WO2006081615A1

WO2006081615A1 - Method and apparatus for detecting significant shielded volumes

Info

Publication number: WO2006081615A1
Application number: PCT/AU2006/000128
Authority: WO
Inventors: Warrick Paul Chisholm; Peter Alaric Hayes; Simon Bedford; David Kenny; John Harold Flexman; Taras Nikolaevitch Rudakov; Vassili Timofeevitch Mikhaltsevitch
Original assignee: Qrsciences Pty Ltd
Priority date: 2005-02-01
Filing date: 2006-02-01
Publication date: 2006-08-10

Abstract

A detection system, including an electromagnetic scanner for detecting explosives and narcotics in objects using an electromagnetic detection technique, includes a significant shielded volume (SSV) detector (7, 9, 11) having multiple detection axes for detecting SSVs in a detection volume containing the objects in advance of the electromagnetic scanner. The SSV detector (7, 9, 11) is capable of determining (13, 15, 17) whether an SSV exists within the detection volume that would otherwise shield the substance within the object from being detected using the electromagnetic detection technique. The electromagnetic detection technique is nuclear quadrupole resonance (NQR) and is provided by an NQR scanner, or is X-ray and is provided by an X-ray scanner, or a combination of both. Probes for a clustered induction detector, arrayed induction detector and a microwave detector are also disclosed.

Description

"Method and Apparatus for Detecting Significant Shielded Volumes"

Field of the Invention

This invention relates to the detection of objects that electromagnetically screen volumes from electromagnetic radiation. The invention has particular, but not exclusive, application to the detection of such objects that have a screening attribute particularly attuned to preventing a probe of an electromagnetic scanner using, for example, Nuclear Quadrupole Resonance (NQR), Nuclear Magnetic

Resonance (NIVlR), Electron Spin Resonance (ESR), X-ray and/or Microwave or Terahertz substance detection, imaging and identification techniques, from irradiating a significant volume.

The invention has utility with detecting such screening objects within mail, airport luggage and other packages, when scanning them for contraband or prohibited substances.

The invention can be embodied in either a stand alone system, i.e. not used in conjunction with a scanner, or in an integrated system with a scanner.

Throughout the specification, unless the context requires otherwise, the word "comprise" or variations such as "comprises" or "comprising", will be understood to imply the inclusion of a stated integer or group of integers but not the exclusion of any other integer or group of integers,

In addition, the term "significant volume" is defined to mean a volume that is of sufficient si∑e to hold a prescribed amount of a substance sought to be detected by an electromagnetic scanner. This amount is determined by both legal and practical limits, and so may vary depending upon the jurisdiction within which the invention is intended to operate. Accordingly, the "significant volume" corresponds to the threshold volume specified for operation of the invention having regard to the legal and practical limits of the substance sought to be detected. Background Art

The following discussion of the background art is intended to facilitate an understanding of the present invention only. It should be appreciated that the discussion is not an acknowledgement or admission that any of the material referred to is or was part of the common general knowledge as at the priority date of the application.

Throughout the following discussion reference is primarily made to the detection of metal objects that shield significant volumes within luggage in airports. However this is only discussed by way of example. The invention also can be used in the detection of objects that shield significant volumes within mail, packages, or any other environment. Furthermore, the invention can be used to detect other significant shielded volumes (SSVs) that may not be contained within a metal or other conductive object or enclosure, but can still screen electromagnetic radiation.

In addition, whilst the following discussion makes specific reference to NQR and NQR scanners, this again is by way of example only. Whilst the invention may have great utility when used in conjunction with NQR scanners given the relative importance of such scanners in the context of the present political and cultural environment and the problems associated with these kinds of scanners detecting substances in SSVs₅ the invention is in no way limited in application to NQR scanners, and may have equal utility in other applications. Accordingly, certain embodiments of the invention described hereinafter describe arrangements that are not limited to use with NQR scanners.

NQR scanners have the potential to be part of the next generation of explosive detectors in airports. However, despite this technique being well known for over 30 years, at present only a few NQR scanners have been deployed in airports around the world. A contributing reason for this delay is that the NQR scan process may have an issue with the detection of contraband and explosive material within a metal enclosure. It is possible that a metal enclosure may form a sufficieπt screen to irradiation by electromagnetic energy in the form of radio frequency waves so as to defeat detection by NQR,

It is commonly known that radio frequency waves (RF) are unable to penetrate thick metal structures. Indeed, most RF shielded rooms are constructed of metal surfaces, which screen electromagnetic fields from every direction, often referred to as a Faraday cage. Rather than ignoring this possible limitation in NQR scanning, one method to solve this problem is to scan for metal enclosed objects and alert the operator that these objects need to be searched by other secondary methods such as an X-ray scanner or by hand.

To detect an SSV in luggage, typically a metal detector is required. Normal metal detectors that are used for the detection of metal objects in soil are not suitable for the detection of metal objects in luggage because they have a limited penetration range and their magnetic field is extremely non-uniform. Metal detectors that have been invented for the detection of gold nuggets, buried coins etc. typically have a drop off in magnetic field intensity that is proportional to 1/r³ and have a limited range of approximately 200mm. This means that small objects that are close to the coil give the same response as those that are large and further away. When looking for a metal shielded object in luggage, the metal detector needs to give a similar response wherever the shielded object lies within the detection volume. As the detection volume has typical dimensions of ~70cm long x ~70cm wide and ~50cm high, then this presents a formidable problem to solve.

Food metal detectors, which are aimed at simply detecting any metal contained within food, are also not concerned with field homogeneity because they are looking for 'any¹ metal, which may harm the consumer of the product. Consequently metal detectors that are used within the food industry are not suitable.

The same can be said for other types of metal detectors, such as those used in the recycling industry. Typically in the recycling industry, objects are transported along a conveyor belt. Heavy items such as metal cans etc tend to fall to the bottom of the recycling materials on the conveyor belt, whereas light items, such as plastic bottles, tend rise to the top. As a result, metal detectors in the recycling industry are usually required to detect over a short-range, just above the conveyor belt surface, which makes them unsuitable for scanning luggage.

Similar deficiencies may be encountered in other scanning environments using electromagnetic irradiation techniques other than NQR, such as NMR₁ ESR₇ microwave, terahertz or other types of scanners where SSVs are concerned. Hence there is a need for a specific design aimed at detecting metal objects in luggage for NQR scanning of explosives, as well as for other types of scanning in similar or different environments.

Specifically with regard to the use of NQR scanners and X-ray scanners, it is expected that NQR scanners will be placed alongside, or be integrated with, X-ray scanners prior to their deployment. The reason for this is that X-ray scanners detect both metal objects, such as guns and knives, and explosives, whereas the NQR scanner is purely an explosives detector. Hence in the process of deriving information from both systems, rather than treating the information separately, it would be useful to merge the data to give the X-ray operator as much information as possible.

Disclosure of the Invention

It is an object of the present invention to provide for the improved detection of objects that electromagnetically screen volumes from electromagnetic radiation.

It is a preferred object of the invention to improve the detection of objects by an X- ray scanner.

According to one aspect of the present invention, there is provided a detection system comprising:

an electromagnetic scanner for detecting explosives and narcotics in objects using an electromagnetic detection technique; and a significant shielded volume (SSV) detector having multiple detection axes for detecting SSVs in a detection volume containing the objects;

wherein said SSV detector is disposed adjacent to said electromagnetic scanner and said SSV detector is capable of determining whether an SSV exists within said detection volume that would otherwise shield the substance within the object from being detected using said electromagnetic detection technique.

Preferably, said electromagnetic detection technique is NQR and is provided by an NQR scanner, or is X-ray and is provided by an X-ray scanner, or a combination of both.

Preferably, said SSV detector is a clustered induction detector (CID), an array induction detector, or a microwave transmission and reflection detector (MTARD), or any combination of same.

Preferably, the SSV detector is disposed in advance of the electromagnetic detector.

In accordance with another aspect of the present invention, there is provided a significant shielded volume (SSV) detector for detecting SSVs within a detection volume comprising:

a probe to irradiate the detection volume with electromagnetic radiation and receive signals therefrom;

transmitting means and receiving means to drive said probe;

processing means to operate said transmitting means in conjunction with said receiving means and process said signals received by said receiving means to identify an SSV within said detection volume; and

alarm means to alert an operator of the presence of an SSV detected thereby. Preferably, the SSV detector is adapted to comprise a clustered induction detector (CID), an arrayed inductive detector (AID) or a microwave transmission and reflection detector (MTARD)₁ or any combination of same.

Preferably, the processing means includes a full metal circumscription routine to determine whether an object disposed within said detection volume is fully enclosed in metal-

Preferably, in the case where the SSV detector is an AID, the full metal circumscription routine ascertains whether the two dimensional size of the object is above a prescribed threshold value and the maximum of an image of the object at a prescribed frequency is above a prescribed threshold and causes the processing means to trigger the alarm means if so.

Preferably, in the case where the SSV detector is a CID, the full metal circumscription routine ascertains whether the resonance frequency shift . is beyond a prescribed limit and causes the processing means to trigger the alarm means if so.

Preferably, in the case where the SSV detector is an MTARD₁ the full metal circumscription routine ascertains whether the size of the reflected microwave signal or the transmitted microwave signal lies above a prescribed threshold and causes the processing means to trigger the alarm means if so.

Preferably, the processing means includes a reinforcing loop routine to determine whether an object disposed within said detection volume is enclosed by a metal reinforcing loop or contains significant amounts of metal.

Preferably, in the case where the SSV detector is an AID, the reinforcing loop routine ascertains whether the integrated amplitude peak of a generated image of the object at a prescribed frequency is above a prescribed threshold and causes the processing means to signal said transmitting means or said probe to increase the magnetic field strength applied to the object sufficiently to counteract the reduction in the magnetic field strength otherwise arising. Alternatively, the reinforcing loop routine in response to ascertaining whether the integrated amplitude peak of a generated image of the object at a prescribed frequency is above a prescribed threshold may cause the processing means to signal said transmitting means or said probe to lower the QR detection threshold.

Alternatively still, the reinforcing loop routine in response to ascertaining whether the integrated amplitude peak of a generated image of the object at a prescribed frequency is above a prescribed threshold may cause the processing means to signal said transmitting means to increase the pulse sequence length sufficiently in order to gather more signal to counteract the effect of the metal reinforcing loop or significant amount of metal within the object.

Preferably, in the case where the SSV detector is a CID₁ the reinforcing loop routine ascertains whether the resonance frequency shift is beyond a prescribed limit and causes the processing means to signal said transmitting means or said probe to increase the magnetic field strength applied to the object sufficiently to counteract the reduction in the magnetic field strength otherwise arising.

Alternatively, the reinforcing loop routine in response to ascertaining whether the resonance frequency shift is beyond a prescribed limit may cause the processing means to signal said transmitting means or said probe to lower the QR detection threshold.

Alternatively still, the reinforcing loop routine in response to ascertaining whether the resonance frequency shift is beyond a prescribed limit may cause the processing means to signal said transmitting means to increase the pulse sequence length sufficiently in order to gather more signal to counteract the effect of the metal reinforcing loop or significant amount of metal within the object.

Preferably, in the case where the SSV detector is an MTARD₁ the reinforcing loop routine ascertains whether the loss of the reflected microwave signal or the transmitted microwave signal lies above a prescribed threshold and causes the processing means to signal said transmitting means or said probe to increase the magnetic field strength applied to the object sufficiently to counteract the reduction in the magnetic field strength otherwise arising.

Alternatively, the reinforcing loop routine in response to ascertaining whether the loss of the reflected microwave signal or the transmitted microwave signal lies above a prescribed threshold may cause the processing means to signal said transmitting means or said probe to lower the QR detection threshold.

Alternatively still, the reinforcing loop routine in response to ascertaining whether the loss of the reflected microwave signal or the transmitted microwave signal lies above a prescribed threshold may cause the processing means to signal said transmitting means to increase the pulse sequence length sufficiently in order to gather more signal to counteract the effect of the metal reinforcing loop or significant amount of metal within the object .

Preferably, the processing means includes a detuning estimation routine to estimate the degree of detuning arising from the amount of metal detected within an object disposed within said detection volume and a capacitor switching means, whereby said detuning estimation routine causes the processing means to signal said capacitor switching means to switch extra capacitance into the probe to counteract the loss in inductance arising from the metal within the object.

Preferably, the processing means includes an RF detection routine to detect the presence of any emitted RF from an object disposed within said detection volume whilst identifying the presence of an SSV and cause said processing means to trigger the alarm to signal the presence of emitted RF from the object.

In accordance with a further aspect of the present invention there is provided a probe for a CID apparatus for detecting a significant shield volume (SSV) within an article passing through a detection volume in a prescribed direction, comprising:

a first coil to produce a substantially vertical magnetic field within the detection volume; a second coil to produce a substantially horizontal magnetic field within the detection volume, parallel to the prescribed direction; and

a third coil to produce a substantially horizontal magnetic field within the detection volume perpendicular to the prescribed direction.

Preferably, the first and third coils are single turn saddle coils and the second coil is a single turn coil.

Preferably, two small continuous single turn loops are added as shim coils to improve the uniformity of the magnetic field of the third coil for producing a magnetic field horizontally and perpendicular to the prescribed direction.

Alternatively, the single turn saddle coil for the first coil for producing the magnetic field in the vertical direction may be replaced by a single turn coil, comprising two loops connected in parallel.

Alternatively still, the second coil for producing a magnetic field in the horizontal direction may be replaced by a single turn coil comprising two loops connected in parallel.

Preferably, another small single turn coil is attached to a hinge and is arranged to brush over the article passing through the detection volume. The field from this coil is preferably substantially in the vertical direction but has a short range.

In accordance with another aspect of the invention, there is provided a probe for an AID apparatus for detecting an SSV within an article passing through a detection volume in a prescribed direction comprising: a series of coils wound upon ferrite rods for both transmitting and receiving signals to and from said detection volume for detecting an SSV within the article.

In accordance with a further aspect of the invention, there is provided a probe for an MTARD apparatus for detecting an SSV within an article passing through a detection volume in a prescribed direction comprising: a series of microwave transmitters and receivers.

in accordance with another aspect of the invention still, there is provided an SSV detector for detecting an SSV within an article passing through a detection volume in a prescribed direction that includes some or all of the following probes:

(i) a first coil arranged to produce a substantially vertical magnetic field, a second coil arranged to produce a substantially horizontal magnetic field, parallel with the prescribed direction, and a third coil arranged to produce a substantially horizontal magnetic field perpendicular to the prescribed direction for detecting metal enclosed objects;

wherein the first and third coils are single turn saddle coils and the second coil is a single turn coil;

(ii) a series of coils wound upon ferrite rods for both transmit and receive; and

(iii) a series of microwave transmitters and receivers.

In accordance with a further aspect of the present invention, there is provided a method for detecting a significant shield volume (SSV) within an article passing through a detection volume in a prescribed direction, comprising:

sensing a preferred position for the object within the detection volume relative to a QR sensor;

positioning the object in the preferred position ready for QR scanning; and

, scanning the object using QR.

Preferably, the method includes sensing the position within the detection volume where the QR field is able to penetrate most of the item; and positioning the item based Dn this information. In this manner, bags or luggage items that have conductive metal structures would be positioned so that the least shielding caused by the metal structures occurs.

Brief Description of the Drawings

Figure 1 is a schematic diagram showing the magnetic field impinging upon a suitcase containing a metal reinforcing loop and the counteracting field subsequently produced.

Figure 2 is a flowchart showing the most basic method of scanning for the presence of an SSV.

Figure 3 is a flowchart showing the basic method for detecting a shielded volume according to the first embodiment.

Figure 4 is a schematic diagram showing the arrangement for the CID within a shield as described in the first embodiment.

Figure 5 is a schematic perspective view showing the first arrangement of coils within the CID described in the first embodiment.

Figure 6 is a schematic perspective view showing the second arrangement of coils within the CID described in the second embodiment.

Figure 7 is a side-on view of the magnetic field generated by the B coil in the first embodiment.

Figure 8 is a schematic perspective view showing the third arrangement of coils within the CiD described in the third embodiment.

Figure 9 is a side-on view of the magnetic field generated by a saddle coil in the first embodiment. Figure 1OA is a schematic showing the circuit diagram for the ClD in accordance with the first embodiment.

Figure 1OB is a schematic showing the circuit diagram for the ClD in accordance with the tenth embodiment.

Figure 11 is a set of graphs showing typical signals observed on the CID as described in the first and tenth embodiments; wherein:

Figure 11A plots the Frequency Shift against the Distance; and Figure 11 B plots the Q Shift against the Distance.

Figure 12 is a flowchart showing one method used to process signal obtained from the coils in the ClD according to the first embodiment.

Figure 13 is a flowchart showing a second method used to process signal obtained from the coils in the CID according to the seventh embodiment.

Figure 14 is a flowchart showing a third method used to process signal obtained from the coils in the CID according to the eighth embodiment.

Figure 15 is a flowchart showing a fourth method used to process signal obtained from the coils in the CiD according to the ninth embodiment.

Figure 16 is a schematic diagram showing the fifth coil arrangement of the CiD as described in the tenth embodiment.

Figure 17 shows the rotational movement of the D coil.

Figure 18 shows the first block diagram of the ClD combined with the NQR detector according to the twelfth embodiment.

Figure 19 shows the sixth arrangement of the ClD as described in the thirteenth^' embodiment. Figure 20 shows an alternative coil arrangement for the ClD as described in the nineteenth embodiment.

Figure 21 shows the AID design in accordance with the twenty-first embodiment.

Figure 22 shows the magnetic field generated between a transmit and receive pair of the AID design.

Figure 23 shows a 25 kHz image generated by the AID for a bag that has a reinforcing metal loop.

Figure 24 shows a 3.3 kHz image generated by the AID for a bag that has a reinforcing metal loop.

Figure 25 shows the difference between figures.24 and 25.

Figure 26 shows the signal processing flowchart for the vertical image for the AID according to the twenty-first embodiment.

Figure 27 shows the signal processing flowchart for the horizontal image for the AID.

Figure 28 shows the signal processing flowchart for the AID according to the twenty-second embodiment.

Figure 29 shows the signal processing flowchart for the AID according to the twenty-fourth embodiment.

Figure 30 shows the first alternative arrangement for the AID according to the twenty-fifth embodiment where the coils of the transmit rods are individually connected to the power amplifier.

Figure 31 shows the field that is being received on ferrite rod number 1 from all transmit rods, according to the AID arrangement of the twenty-sixth embodiment. Figύre 32 shows the setup for the microwave detection of shielded objects via the transmission signal according to the twenty-seventh embodiment.

Figure 33 shows a flowchart for signal processing the transmission microwave detection signals.

Figure 34 shows the setup for detection of reflected microwave signals according to the twenty-eighth embodiment.

Figure 35 shows a flowchart for signal processing the reflection microwave detection signals according to the twenty-eighth embodiment.

Figure 36 shows the setup for the combined microwave transmission and reflection detection system according to the thirty-first embodiment.

Figure 37 shows the combined microwave transmission and reflection detection system according to the thirty-second embodiment, where the transmitters are individually transmitting signals.

Figure 38 shows, the combined CID and AID according to the thirty-seventh embodiment.

Figure 39 shows the second combined CID and AID according to the thirty-ninth embodiment.

Figure 40 shows the total combined system of the ClD; AID and the MTARD according to the forty-third embodiment.

Figure 41 is a flowchart showing the process performed by the system as described in the forty-fourth embodiment.

Figure 42 is a flowchart showing the process performed by the full metal circumscription routine. Figure 43 is a flowchart showing the process performed by the reinforcing loop routine.

Figure 44 is a flowchart showing a variation of the process performed by the reinforcing loop routine.

Figure 45 is a flowchart showing another variation of the process performed by the reinforcing loop routine.

Figure 46 is a flowchart showing the process performed by the detuning estimation routine.

Figure 47 is a flowchart showing the process of normal measurement performed by the detuning estimation routine.

Figure 48 is a flowchart showing the process performed by the RF detection routine.

Best Mode(s) for Carrying Out the Invention

The best mode for carrying out the invention will now be described by way of various specific embodiments with reference to the accompanying drawings. The embodiments are directed towards different detection systems involving various detection methods and detector apparatus for detecting the presence of a significant shielded volume (SSV) that may be present within an article, prior to scanning the article using NQR detection techniques.

According to the best mode of this invention, in order to be suitable for the detection of SSVs within luggage, the particular system chosen must satisfy some or all of the following criteria:

the field generated by the detector apparatus must give an indication of the SSVs size, so that small clutter, such as from metal, can be separated from more valid RF screens that have sufficient volume to contain a threshold quantity of a substance being searched for. - the field generated by a detector must be reasonably spatially uniform - this is required so that small objects^' that lie close to the coil of a probe are not more readily detected than large objects that are further away and which give weaker signals.

- the system must be configured as a multi-axis system so that every conceivable orientation of a screening object can be detected.

- the system must have a reasonably low reject rate, but maintain a high detection rate to specific targets, and be able to select targets according to some defined rules which may be based. on current security information.

- the system devised desirably would give some indication of where in the detection volume the object lies - this is to give the operator an idea of how to find the object

- although not essential, it would be desirable if the system could also image the object within an item of luggage, for example.

- the system also needs to be fast and minimize any delay to the NQR scanning process.

- it also must not interfere with the surrounding environment, as it may be located in, for example, an airport where electronics emissions may harm the operation; and conversely it must be able to operate in the same electrically/mechanically noisy environment

- it must not harm the volume, such as a passenger's luggage, in any way.

- it must be selectively sensitive to unintended, albeit real, potential shields without triggering false alarms on any conducting structures that don't act as effective shields - an example would be overcoming the problem of 'metal bag reinforcing loops', which are continuous metal bands that exist in most trolley type suitcases today. - It must be relatively thin such that the metal detector system does not add substantial overall length to the NQR scanner - the reason for this is that space within airport terminals is limited and cannot accommodate bulky explosive detectors.

- It must be cost effective.

- It must also be reliable and robust to work in different environments, such as certain airport conditions that may contain some electrical noise and mechanical vibration. It should not be easily influenced by external sources whether accidentally or maliciously, which may require significant shielding of the system itself.

With respect to metal bag reinforcing loops, as shown in Figure 1 , most pieces of luggage are formed with a bag tube 1 provided with a metal bag reinforcing loop 2. When a magnetic field 3 impinges upon the bag loop 2 from above, eddy currents are generated in the loop 2, which generate a magnetic field 4 that opposes the primary magnetic field 3 and thus destroys the primary magnetic field in the vicinity of the bag loop. This interaction with the probing magnetic field 3 can cause a false detection.

In order to achieve the above criteria, the best mode of the present invention provides for an SSV detection system that detects at least tri-axially using a particular probe arrangement that generates magnetic fields in at least three different directions, which are all orthogonal to each other and which detects differences in signals received in response, thereto. Accordingly, the SSV detection system includes transmitting means in the form of a transmitter for causing the probe arrangement to irradiate the object with electromagnetic energy, receiving means in the form of a receiver for monitoring changes in electrical properties of the coils of the probe, processing means comprising a computer and a processing unit to ascertain when a significant change in the electrical property has occurred, and an alarm means to alert an operator of the system when such a significant change has occurred. If the SSV target to be detected is a very thin object and it lies parallel to the conveyor belt motion, then it will be only easy to detect in one of these three orthogonal fields. In the other two directions very little signal will be observed and it will be barely detectable. A system that is built with the ability to only detect in two of these directions may miss some thin objects and thus will be flawed in design.

The general methodology adopted by the best mode is shown by the flowchart in Figure 2 of the drawings. This flowchart shows the process of scanning for an SSV in its most basic form. An article to be scanned by an NQR scanner, in this case a bag, is moved 7 into an SSV tri-axial detector system containing one or more detection coils of a probe and is irradiated with electromagnetic energy 9. The movement of the bag under test could be performed by using a conveyor belt; however it may also be achieved by moving the bag by hand into the search area.

Once irradiated the responses from the receivers are monitored and measured 11 to determine if there is a significant change in an electrical property of the coils compared to when there is no bag present. In this case an electrical property of the receivers might be a voltage or current, however this measured voltage or current might be transformed into another electrical property of the receivers such as quality factor (Q), inductance, or resonant frequency. The measured information is processed to ascertain 13 whether the magnitude in a change in the electrical property or properties being measured is sufficiently significant to suggest the presence of an SSV. If so, then an alarm is triggered or the operator of the machine is alerted 15, otherwise the article passes through the system as being clear of SSVs 17.

The embodiments described hereinafter, present three different designs of significant shielded volume (SSV) detectors, which have been found to satisfy the criteria set out above, and a number of different variations or modifications that arise therefrom. These detectors can be used concurrently as described in one of the embodiments, or used separately as described in other of the embodiments, depending on the circumstances. The three detectors essentially operate in three different frequency ranges and all can perform the scan of the luggage quickly enabling a high throughput:

1) one SSV detector operates in the low MHz region;

2) another in the low kHz region; and

3) the other operates in the high MHz/low GHz region.

The use of these three frequency ranges produces more information about the bag and enables the operator to make a better-informed judgment about the SSV that lies within. In particular, the use of the low kHz and the high MHz/GHz frequency ranges allows penetration of the magnetic field through the reinforcing bag loops and so enables metal objects that lie within these reinforcing loops to be detected.

The three different designs of detectors are:

1) a clustered inductive detector (ClD) for operating in the low MHz region;

2) an arrayed inductive detector (AID) for operating in the low kHz region; and

3) a microwave transmission and reflection detector (MTARD) for operating in the high MHz/low GHz region.

Clustered Inductive Detector (CID):

The first embodiment of the best mode will be described with respect to an SSV detector using the CID design. The SSV detector can be configured for use either by itself or it can be intimately integrated into an NQR detector system for the location of contraband and explosive items.

As shown in Figure 4, the SSV detector comprises a multi-axis resonant coil cluster 39 enclosed inside a metallic conductive shield 40 whose ends are open so as to allow baggage to be transported through the cluster via a conveyor 44, and whose length is around 600mm. An optical sensor 43 is provided to detect the beginning and end of a scanned bag conveyed to and through the coil cluster 39 by the conveyor 44 and the location of the bag in time.

The coil cluster 39 employs a system of coils that define a compact scanning volume within which a bag is temporarily disposed on transport by the conveyor

44 through the cluster. The coils are particularly designed so that the SSV detector is able to detect a shielded volume in a manner where the response of the shielded volume is not strongly dependent on its orientation within the compact scanning volume. This is achieved by using coil designs that have relatively uniform magnetic fields.

The SSV detector also comprises a transmitter/receiver unit 47 and a processing unit 48. The resonance parameters of each coil in the cluster are measured through , transmitting a signal and receiving a signal through the transmitter/receiver unit 47 and then analysing the received signal with the processing unit 48. The results are recorded as a function of the distance of the bag conveyed through the coil cluster 39, relative to the coil cluster.

The signal output by the optical sensor 43 is passed through a driver unit 46 and into the processing unit 48. This signal can either be . used to trigger a data collection cycle or detect the beginning or end of a transported bag in time, which then can be related to data collected from the coil cluster in time.

As shown in Figure 5, in the present embodiment, the multi-axis coil cluster 39 includes three high Q copper coils A 21, B 22, C 23, which are tuned to 1.6MHz, 1.7MHz and ISMHz, respectively, by adding high Q ceramic chip capacitors 31, 32, 33. The three main coils (A 21, B 22, C 23) are orientated to detect objects along three orthogonal directions so that even if the object has a thin profile it will be detected.

The Q's of the resonant coil systems A, B and C are approximately 500, with the inductances of these coils within an external shield ranging from 1uH to 3uH. These numbers are only indicative; any values could be chosen depending upon the application, size of the coils required etc. Although in the present embodiment, the coils are single turn; in another embodiment (not shown) they are multi-turn to increase inductance. The multi-turn configuration, however, is viewed as less desirable as the complexity in building them would increase and the ultimate unloaded Q is lowered.

In the present embodiment, the resonant circuit is constructed so as to be parallel resonant with the applied capacitance. Consequently_* the resonant circuit closely matches the properties of an NQR system with which it is associated, and thus utilises similar components and analysis software systems to that provided with the NQR system.

In an alternative embodiment, the resonant circuit is constructed so as to be series resonant with the applied capacitance; however, in the alternative embodiment, the resonant circuit is not able to closely, match the properties of an associated NQR system, and so would not use similar components. Nonetheless, such an embodiment may have utility as an independent SSV detector, where it is not important for the resonant circuit of the SSV detector to be closely matched with the properties of an NQR detector.

In the present embodiment, the coils A, B and C are arranged symmetrically and orthogonally to each other. In this manner, the coils minimally interfere with each other by reducing the currents that are induced on each coil. The result of this arrangement is that all coils are basically decoupled from each other. By virtue of this decoupling, the resonant frequencies of the coils could be chosen to be the same frequency, if desired, although in the present embodiment they are chosen to be marginally different from each other, as described above.

As a result of this configuration of the coils, the magnetic field directions from coils A, B, and C in the detection volume mainly point vertically, along the conveyor belt 44 and across the conveyor belt, respectively. This system allows essentially three orthogonal measurements of the baggage. The coil layout includes shim coils (not shown) to change the field homogeneity and additional search coiis (not shown) to this basic cluster. The coils are adjustable by bolting straight electrical grade copper bar segments together (not shown). This is useful to adjust the coil dimensions to find an acceptable layout, and allow a design with high conduction in the coil structure. The bars have a series of holes drilled along their length so that the attaching bar can be moved along its length. Several bolts are used to ensure electrical contact and preserve the electrical Q of coil. In other embodiments, the bar is replaced by an alternatively shaped conductor such as pipe, rod or wire, which may be considered desirable depending on the required engineering.

In the present embodiment, coil A 21 is a two loop saddle-like coil that detects planar objects that lie flat within the detection volume. This single turn design enables a high Q to be achieved that is extremely sensitive to metal targets. It is shaped to create a magnetic field that is relatively uniform throughout the detection volume. In designing the coil, care is taken so that the saddle-like form does not create a field that couples in the vertical direction along the width of the luggage tube. This is also useful to avoid coupling to the metal reinforcing loops of a bag, which generally will lie in the same plane as and close to the conveyor belt, when a bag is conveyed through the coil cluster 39 of the SSV detector.

The saddle-like coil shape is most suitable for this purpose; however, in other embodiments other coil shapes are adopted, which also are adequate. One of these is specifically described in the second embodiment.

In the present embodiment, the B coil is a narrow single turn coil 22. A side view of the magnetic field 50 generated by this coil 51 is shown in Figure 7. This coil primarily senses objects that present their greatest surface area in the direction of the conveyor belt motion 34, although the signature of bags with metal reinforcing loops also makes the signal received from this coil extremely useful for detecting objects in these reinforced loop type bags.

Coil C is also a single turn, saddle-type coil 23 constructed by connecting two loops. The magnetic field 63 generated by this coil 23 is shown in Figure 9. This coil detects objects that face across the conveyor belt. The coil structure consists of a continuous metal structure 23 and two disconnected electrically continuous rectangular loops 25, known as opposing loops- These opposing loops can be regarded as field shimming coils.. The shape of the coils and the opposing loops are designed to create a field that is relatively uniform in the direction of the probe formed by the coil cluster and associated circuitry.

The opposing loops reduce the field from coil C in the area of the sides of the luggage tube, as well as helping to shape the fields from coil A and B so they are more parallel to the sides of the luggage tube.

In an alternative embodiment, coil C is constructed from simple rectangular loops, where the loops are wired in parallel. However, this is less desirable because there is some loss in field uniformity.

As described, Figure 4 shows the connection from the coil cluster probe to the electronics, which monitors the probe and receives bag position information. The components of the system used to transmit and receive signals from the coils form a modulator/demodulator circuit as shown in Figure 1OA, whereby the same lines that are used to transmit to each of the channels of the A, B and C coils, are used to receive signals from the A, B and C coils.

A modulator circuit is used to generate a signal on any of the coiis by using a Direct Digital Synthesiser (DDS) 52 to generate sinusoidal waves at the required frequencies on a single transmit line. Accordingly, an N-channel transmit demultiplexer 53 is used to split the single transmit channel into N channels that are sequentially selected, so that a sinusoidal pulse of about 500μs can be applied to each resonant coil in turn. Element 54 is an isolating component to ensure the demultiplexer 53 doesn't significantly load the resonant circuit and cause a deterioration in Q.

A demodulator circuit is used on the receive side to receive signals from the coils. The receive signals are initially amplified by amplifiers 60 and fed into an N- channel receive multiplexer 59 that is used to multiplex the N receive channels into a single receive line connected to a receiver module 61. The multiplexed signals are mixed down to 30 kHz in the receiver module 61 and further amplified before being sent into a single channel ADC card 51, where the signal is sampled at 360 kHz. The sampled data is then sent to the computer 50 for signal processing and is graphically displayed via the display 49.

The method of operation of the SSV detector will now be described with reference to Figures 3, 4 and 10.

In the present embodiment, the CID form of the SSV shield detector operates in the low MHz frequency range. This range is reasonably close to the QR frequencies of interest in an NQR detector system for detecting particular types of explosive. The reaction from the SSV detector at these frequencies mirrors the shielding ability of a shielding material during an NQR excitation. The shielding effect is dependent on the conductivity, permeability and/or geometry of the shielding material. The effectiveness of the shield therefore changes with applied probe frequency, because the conductivity is dependent upon the frequency.

The choice of frequency region can be refined to include other benefits. An example of a benefit might be to operate at frequencies where low RF interference occurs from external sources. In this regime the resonant coils can potentially be employed using limited shielding 40 from external RF sources or in alternative embodiments no shielding at all.

As provided for in the present embodiment, a small offset from the QR frequencies of interest is desirable so that the process can be carried out at the same time as any sensitive QR scan process. For people skilled in the art this frequency range has an advantage in that large coils needed for the volume to be scanned can be easily constructed to have high-Q factors. In general this property of high-Q allows very small changes on the electrical properties of the coil antenna to be identified quickly.

In the best mode of the invention using the present embodiment, the coil cluster 39, the external shield 40, and the electronic chains of the transmitter/receiver unit 47 are constructed in such a way so as to operate with the high-Q resonant probes. For instance, the electronics required for the transmit and receive modes are lightly electrically coupled to the resonant tank circuit so as not to load the coils 39, and the external shield 40 is a highly conductive material that allows sufficient space to maintain high inductance and a low reluctance return flux path.

The SSV detector essentially measures the response of a small group of predominantly orthogonal coils 39 as a function of bag travel distance so that recorded features related to shielding objects distributed throughout the baggage can be matched to their location in the bag. An analysis is performed on the modification of the measured electrical properties from this group of coils to discern significant shielded volumes (SSV) relative to their location in the bag. The analysis of recorded data discriminates the SSV from other less significant shielded areas.

Figure 3 describes the process, where the bag travels into the coil cluster 39 and the responses are collected from the multi-axis system. The inductive and resistive characteristics change for each resonant coil as bags of varying magnetic and electrical character pass through it. Potentially the luggage within a bag is composed of objects that can be divided into two types: clutter and SSVs. Here "clutter" describes shielding items that don't appear to be SSVs to the applied NQR field. As the bag moves through the coil cluster 39, the change in electrical characteristics for each coil as a function of position or "responses" of bag travel is recorded 63. The responses from clutter are modelled based on measured parameters 65, being produced from computational models for the clutter based on the measured character of the bag. These clutter responses, drawn from the best models of clutter, are compared and removed by subtraction from the recorded responses 67. On subsequent processing 69 if a significant response still exists an alarm is generated 71 , otherwise not 73 and the bag is clear to pass on to NQR or other scanning.

To scan a bag for the presence of an SSV therein, a bag is brought into a scan area via the conveyor belt 44. In the process of proceeding into the scan area the bag breaks an optical fence sensed by the optical sensor 43, which triggers the measurement process.

One aspect of the measurement process involves measurement of the length of the bag. The bag length is determined by measuring the difference in time between when the optical fence is broken and not broken and knowing the average velocity of the bag. Algorithms are applied to account for possible issues such as dangling straps causing the beam to broken multiple times on the same bag.

This length information is useful for the later signal processing of bags with metal reinforcing loops. The correlation between the signals from the optical sensor and the Q and frequency shifts allows magnetic features of the bag and its contents to be located. This aids in processing, as mentioned previously, for metallic baggage structures generally have fixed locations near the edges of the bag. The overlay of the magnetically identified suspect locations with the real dimensions of the bag enables the bag to be more quickly searched by other means which are able to reference these dimensions.

By knowing the position of the bag where the bag is not influencing the coiis significantly, the baseline can be identified. This data region is then used as a ^•baseline so that the processing is able to effectively recalibrate to an empty coil cluster for each bag. The recalibration largely eliminates the effects of drifts in absolute resonant values of the unloaded resonant system caused through, for instance, temperature..

Apart from this one length dimension, it is also desirable to measure the other dimensions of the bag so that the maximum width and height is known for the bag under measurement. These dimensions are used to provide limits to the processing algorithms so as to more accurately define a clutter bag model, in particular the outer dimensions of any contained structural conductive loops. These dimensions are measured using an optical fence where light beams are broken to indicate the desired dimension. Alternatively, and more desirably, a camera is used and the resulting image is processed to find its physical outline dimensions.

After tripping the optical sensor 43, a signal is then generated on a single transmit line to the coils A, B, C by the DDS 52 in the form of sinusoidal waves at the required frequencies. The N-channel multiplexer 53 splits this single transmit channel into N channels that are sequentially selected, so that the sinusoidal pulse of about 500μs is applied to each resonant coil in turn.

A cyclical coil scan process is then applied, where each coil is in turn excited over a stepped narrow range of frequencies and the receive signal is received by the receiver unit 47 and recorded by the process unit 48. Typically there are a low number of frequency steps (NF) for each coil, which produce NF intensity values. A typical value for NF is 10 steps.

The process is then repeated for the frequency range of the next coil and so on.

This whole process is then repeated to monitor the bags travel through and beyond the coil cluster 39. The frequency sweeps of NF steps are designed to cover a range just before the resonant frequency of the coil, through the resonant frequency and a short frequency range just after it.

In the method used in the present embodiment, the frequency is swept for coils A₁ B₁ and C through a range that is 10-30 kHz wide, the range being chosen to suit the responses of the coils. The choice in range depends on the shift in frequency expected from each coil as the baggage passes through. This range is ideally optimised to the population and the expected variation from each coil so as to provide efficient scanning.

For an ideal resonant system the received signal intensity after its transient behaviour follows the text-book Lorentzian shape as the frequency is stepped with fixed increments through the resonance frequency. The peak of the Lorentzian corresponds to the resonant frequency and the width allows the Q to be calculated. The variation in amplitude of the received signal is thus recorded and related to the Q after processing according to the Lorentzian shape.

The pulses that contain successively increasing frequency sinusoids are an efficient method of excitation, in that three coils are able to be scanned every 2cm of baggage movement at a conveyor belt speed of 0.5m/sec. This method allows a reproducible amount of energy delivered to the resonant system that finally generates a signal well above possible noise sources from within the luggage or from external RF noise.

Alternative methods of excitation may be provided. Indeed different methods are described in the fourth and fifth embodiments. However, each method enables the electrical parameters of the coil cluster to be recorded and analysed to calculate the resonant frequency and Q of. the varyingly loaded resonant system as a function of bag position to create responses for each coil.

Once the resonant frequency and Q is known, the presence of an SSV having an area that is able to intercept the field perpendicular to the direction of a particular coil is identified by virtue of it causing a shift in the resonant frequency and generally a noticeable change in Q of the loaded resonant coil.

If coils A, B and C detect a frequency shift and/or effective Q shift above a prescribed threshold, then an alarm is generated for the coil or coils, otherwise the bag is passed as clear.

The most useful response is the frequency shift. This response is strongly correlated to the dimensions of the SSV object The Q response is also useful in that objects that cause its deterioration could act as an SSV. Characteristics of the Q of the shielded item allow some aspect of its conductive character to be determined eg. a thin conductive sheet will have a lower Q than a thicker sheet for instance.

In the present embodiment, the two responses, resonant frequency and Q₁ are correlated to produced a third measure, which is useful to further refine identification. The function that combines them is their multiplication at each measurement distance to generate a new response with distance. The thresholds are a constant with position in one mode, and are shaped as a function of position in another mode to account for the difference in coupling geometry between the NQR coil and the shield probe coils to the bag.

Before comparison to this threshold the responses are filtered or modified mathematically to account for the variations in response caused by the object being either displaced vertically or across the conveyor belt relative to the measuring coil structure.

Even though most coils are designed for uniform response, there still exists a variation in sensitivity, which generally means a small object close to the coils will generate a greater response over smaller travel compared to an object in the centre which would tend to generate a smaller response over larger travel. This variation is mathematically corrected so that the response curves are less dependent on target position.

In the present embodiment, this is achieved by processing a piecewise weighted summation around each point in the response. Alternatively, in another embodiment, this is achieved by using a filter function with the coefficients chosen to best reduce this variation.

A further method employed by the present embodiment to improve the quality of the data for later processing is the application of decoπvolution methods to each response. This partially removes the shaping introduced by the finite position resolution of coil. Applying a controlled deconvolution method aids in extracting distributed small clutter throughout the bag to reveal larger shielded volumes. A reliable deconvolution technique adopted in the present embodiment is the Van Cittert method.

For bags with a metal reinforcing loop, the response is particularly strong, given that the loop can generate large diameter circulating eddy-currents in response to the applied RF field. With placement of the plane of the loop in the plane of the conveyor belt, coil A is the most affected, resulting in a large response that is peaked with the bag at the centre of the coil. This is the typical orientation for trolley bags with loops because they are conveniently loaded in this direction, and the preferred direction for a transmission X-ray as this is the thinnest cross- section.

The loop itself can often be considered a cluttering object as it generally does not act as an effective shielded volume with an exciting QR field whose direction lies in the plane of the loop. Given the relative difference in magnitude of response from the baggage loop and small shielded packages, it is difficult to use the A coil alone to find a shielded package.

Coil B, because of its field, generates a characteristic "M" type shape response as it.alternately couples, decouples and couples again with the loop as it travels with the conveyor belt. This "M" shape originates from the orientation of the magnetic field around the B coil 22, as shown in Figure 7. Within the narrow volume circumscribed by the B field, the field lines 74 are parallel with the conveyor belt 44. However just outside the volume of the B coil 22, the field lines 74 bend and in fact form a circular shape around the copper strip. Hence when a bag first encounters this field, a frequency shift is observed because the field lines are traversing down and a current is induced on the metal loop 2 of the bag. This induced current causes a strong resonant parameter shift on coil B.

When the bag is symmetrically placed with respect to this coil, there is a minimum of current induced on the metal loop 2, as equal numbers of field lines 74 will move into the loop on one side of the coil 22, as they will be leaving the loop on the other side of the coil.

When finally the bag loop 2 exits the coil again, there is a current induced on the bag loop as the field lines cut through the bag loop from underneath.

Because the bag loop 2 would be expected to produce a symmetrical response, this can be exploited to determine if a metal object is present. If a large metal object is present, this expected shape is altered and thus the metal object can be detected on this basis.

A useful method employed in the present embodiment to determine whether the expected shape of the magnetic field has been altered involves selecting two points either side of the "M" shape, and measuring the magnetic field at these points. The least change of these two is then chosen as the baseline.

If the response observed lies outside a prescribed threshold, then an alarm is generated, otherwise the bag is cleared from this stage.

If this resultant minimum is sufficiently large when the bag is symmetrically placed with respect to the B coil, and the responses from the other coils match that of a looped bag, in particular the A coil, the software of the processing unit 48 is designed to calculate the expected response for a looped bag at the recorded bag positions for each of the coils.

The shape of the expected response curves is based on archived results of bag loops measured over a wide range of potential parameters, such as loop dimensions, loop height, electrical conduction of loop etc. These archived results are stored in memory of the processing unit 48, and accessed for comparison purposes using appropriate modelling software.

These archived results may be obtained in a number of ways. In one embodiment they are constructed by solving Maxwell's equations for the specific geometry of the coil and metal bag structures as a function of bag travel, loop dimensions and height.

In the present embodiment, however, these results are estimated mathematically from empirical measurements to first determine the amount of magnetic flux penetrating the bag loop, then the amount of flux disturbed by the metal bag structure, and finally considering this as proportional to the expected response. To determine the amount of magnetic flux penetrating the bag loop, each coil is treated as a segmented line of conductors of current to give the total field at a point. The amount of disturbed flux is found through integration over the area of the bag loop.

A function based on this flux approach is developed for each coil. The equations for the function are refined through comparison of collected data from loops formed on a test jig of varying dimensions and height and actual baggage structures measured with the shield detector. The measured -data contains features due to deflections from the outer shield 40, the physical dimensions of conductors involved in constructing the coils 39, additional baggage frame structure such as handles etc. The equations are corrected to produce response shapes and overall magnitudes that are realistic for typical looped bags.

A well known technical computing language for analysing scientific signals is Matlab™ produced by Mathworks in MA, USA. Matlab™ computer software code for the response from coil B for circulating eddy currents in just a horizontal loop is shown as an example below:

b_dist = dist - bag_length/2; a_dist = dist + bag_length/2; scale_freq = loop_width*scale_mag;

Calculate flux through loop from bottom of coil

Ioop_freq1 = abs(log(1./(loop_height^Λ2 + b_dist.*b_dist)) - log(1./(loop_height^Λ2 + a_dist.*a_dist)));

Calculate flux through loop from top of coil

Iooρ_freq2 = abs(log(1./((coil_heighWooρ_height)^Λ2 + b_dist^*b_dist)) - log(1./((coil_height-loop_height)^Λ2 + a_dist.*a_dist))); Find approximate loop frequency shift based on scaled total flux through loop that has been corrected for influence of external shield

loopjreq = scale_freq*(loop_freq1 - Ioop_freq2). ^*exp((dist.^*dist)/ (2*(coilj-ange)^Λ2));

where:

'dist' is an array of bag locations relative to the centre of the coil. 'coil_range' is the range of the coils fringe field, modified by shield, 'loopjπeight' is the height of the loop relative to the bottom of the coil. 'coil_height' is the ditsnce between the top and bottom of coil B. 'scale_freq' is a value used to scale the flux result to a frequency shift.

It can be seen from the above code that the bag loop signature "M" shape depends upon bag loop length, width and height. As the bag loop length is approximately known from the optical sensor measurements, with typically the loops traversing the entire length of the bag, then the other two parameters can be varied to determine the expected bag loop signature. Once the bag loop signature is determined, this signal is subtracted from the received signal, and if the remaining signal lies above a threshold, then an alarm is generated.

Figure 11 shows graphically the types of signals obtained from four coils (A, B, C and D.) for a loop bag. As can be seen, the signal obtained from coil B shows the characteristic "M" shape produced by a looped bag. Coil D is a supplementary coil that will be explained in another embodiment described below.

The processing performed by the processing unit 48 uses information from the coils to predict the responses from clutter, and in particular, the structure of the metallic reinforcing loop 2 of the bag.

The change in response for loop dimensions and height control largely the magnitude and shapes of the responses for each coil in very different ways. As noted earlier, the response from each resonant coil is not entirely uniform with displacement along the sampling field direction.

These features allow a system of equations to be constructed that approximately describes the response as a function of the parameters: loop height and bag width, so that a mathematical fitting process can find a satisfactory loop bag model. This fitting process is bounded by the physical limit imposed by bag length, and employs observed real-world correlations in the dimensions of bag lengths and bag widths to help define the unknown bag width.

The processing then constructs a list of appropriate models from the measured data. The responses are then calculated for these models and adjusted based on loop height and width within ranges to fit to collected responses.

The fitting process is forced towards minimums of selected response regions with position so that the derived model does not over-emphasise clutter. This avoids an overestimate of the responses from just the bag structure alone so that any shielded volume will be more likely to trigger an alarm.

In general the information from the responses of coils A, B and C is enough to find acceptable estimates of bag loop width, height and, to some extent, other structures such as metallic handles.

In the present embodiment, the bag width and height, as well as length, are measured. In addition, potential positioning errors are accounted for, where the centre of the loop is found to be displaced or the loop is of slightly different size than that calculated from the optical sensor information. Accordingly, the centre of the models are displaced and scaled in a small range of values to attain a good model.

A laptop computer or other relatively continuous metal target may trigger processing related to a looped bag based on the threshold comparison above, but its response shape will lead to a detection. For example, coil B won't create an M shaped response, producing an increasing frequency response and generally decreasing Q response as the object is about centred relative to B. This difference in response shape from the model responses of the coils for the looped bag is used to discern a significant shielded volume. These differences are examined and if determined to be sufficient then an alarm is generated.

Once the best model for clutter has been determined, the associated responses are derived and a compare/subtraction process is performed. Each coil response is compared to its own set of characteristic thresholds to decide if the difference is sufficient to trigger an alarm.

The threshold is uniform with position in the one mode, or it is shaped so as to pick out selected areas of the bag in the second mode. These individual coil alarms are mathematically combined to give an overall assessment on the likelihood that the bag contains an SSV.

The specific process flow adopted by the processing unit 48 using the various techniques described above is shown in figure 12.

Firstly, responses are created by finding the resonance parameters comprising frequencies, amplitudes and Q's, and processing those 101 as a function of distance. The responses are then offset 103 relative to recorded parameters without luggage. They are then filtered for noise (if any) and adjusted 105 depending upon the sensed proximity of the bag to the coils.

A comparison is then performed to ascertain 107 whether the responses cross the thresholds. If so, an alarm value and a shield position value are calculated and inserted 109 into an alarm array.

^■ The process then proceeds to the next stage where the minimum magnitude of responses from selected areas of the bag associated with selected coils is found 111. A further comparison is performed 113 to ascertain whether the offset responses cross their thresholds, and if so, an alarm value and a shield position value are calculated and inserted 115 into the alarm array.

The responses are then checked 117 to see if they are consistent with a looped bag of the measured length. If not, the total alarm value from the alarm array is determined 119 and compared 121 to see if it is above a threshold, whereby if it is, it is concluded that a shield has been found and the position from the weighted alarm positions is calculated 123, whereas if not, the bag is cleared 125.

If the checked responses 117 are consistent with a looped bag of the measured length, a list of looped bag models is constructed 127, based on the measured bag length. A list of model responses is then constructed 129, and the model responses that best fit selected bag response characteristics are selected 131.

The best model responses are then subtracted 133 from the bag responses to obtain difference responses, and the difference responses are then compared 135 to determine whether they cross their thresholds. If so, an alarm value and a shield position value are calculated and inserted 137 into the alarm array.

The total alarm value is then determined 119 from the alarm array, and compared 121, to determine whether it is above a threshold. As before, if it is, a shield is concluded to have been found, and its position is calculated 123 from the weighted alarm positions. If it is not above the threshold, then the bag is cleared 125.

In general the model responses which may be effectively zero are subtracted from the responses and the response weighted distance is calculated. For simple objects this gives the position of the object to within a few centimetres. As the clutter and hence model becomes more complex, the error between the calculated position becomes greater to the most significant shielded target. In a very high percentage of cases, as shown in real-world testing, the SSV is still found to be within 10 cm, allowing the bag to be usefully segmented to reduce search time and/or enhance detection which may involve other techniques.

Other processes may be employed that are ancillary or alternative to that described with respect to Figure 12, and are described in subsequent embodiments.

In the present embodiment a graphical response with position (not shown), before and after model subtraction, is displayed to an operator or overlayed with some other image such as optical or X-ray generated image to help in identifying areas of interest.

The second embodiment is substantially identical to the first embodiment expect that instead of using a saddle-like coil shape for coil A, it includes two rectangular loops connected in parallel.

As shown in Figure 6, a simple rectangular looped coil 21' is used as coil A.

The third embodiment is substantially similar to the preceding embodiments except that instead of using a single rectangular coil for coil B, a coil comprising a pair of electrically connected parallel loops, separated by a distance along the direction 34 of the conveyor belt 44 is provided to create an extended uniform field.

As shown in Figure S₁ this embodiment provides for only two loops coils that are electrically connected in parallel to make the coii B 22^s. In addition, the disconnected opposing loops 25 provided in the first embodiment are removed, which causes some loss in uniformity response within the detection volume.

This arrangement extends the magnetic field created by coil B, in contrast to the B-field created by the arrangement of the single coil in the preceding embodiments. This arrangement improves uniformity of the magnetic field from its centre to the edges of the detection volume; however, it generally lowers the effective resolution along the bag length and makes the corresponding image harder to interpret. Notwithstanding this, the extended uniform field is more useful if the coil structure acts as an NQR coil as well, which will be discussed in a subsequent embodiment.

The fourth embodiment is substantially the same as the first embodiment except that another method of excitation is used for the coils. Moreover, a short rectangular pulse is employed to impulse excite a signal that rings down on the resonant coil. This method can be described as "pinging" the resonant coil.

This ring down, or exponential change in signal amplitude, with time provides the same information as does analysing the change in resonant frequency and Q of the loaded resonant coils. The advantage of this method is the increase in speed and efficiency obtained by using a single exciting pulse that causes each resonant coil to electrically ring. The primary frequency of the ringing is the resonance frequency and the decay rate is related to the Q of the loaded resonant coil antenna.

The result of the excitation is measured, as described previously, and analysed through a Fourier Transform (FT) to convert the ring down signal in the time domain to a Lorentzian shaped response in the frequency domain, which again is described by the resonance frequency and Q values. Analysis is also possible in the time domain for those skilled in the art.

The fifth embodiment is similar to the fourth embodiment, except that it employs another method to simultaneously excite the coil cluster via a rectangular pulse delivered to all coils and then monitor the ring down from all coils.

This is achieved through direct connections to each resonant coil or antenna. Moreover, all coils are excited via an electrical connection by delivering a "ping" and using a single ADC to digitise a combined signal for all coils received through a high impedance summing amplifier. The combined signal is then decomposed through standard Fourier analysis to find the resonant frequencies and Q of each coil. The unloaded resonant frequency, being very different for the described coils A₁ B, and C, minimises any cross talk of information between separated coil channels.

The sixth embodiment is substantially the same as the first embodiment, except that the signals from the coils are measured without the use of a demodulator circuit. In this embodiment, the signals are electrically rectified and filtered and sampled by a slow ADC or directly sampled by a fast ADC without the use of a demodulator.

The seventh embodiment is substantially the same as the first embodiment, except that the main process flow undertaken by the processing unit 48 is supplemented to combine the clutter removed responses from each coil through a mathematical function to form a total and then compare the result against a threshold.

The process flow is shown in Figure 13, whereby after checking 113 whether the responses are , consistent with a looped bag of the measured length, and determining that they are not, a supplemental set of steps is performed before the total alarm value is determined from the alarm array 119. Moreover, like responses of the various coils, such as frequency, are combined 139 from the previous steps to form a total, and a comparison is performed 141 to ascertain whether these responses cross their respective thresholds. If so, an alarm value and a shield position value are calculated 143 and these values are inserted into the alarm array.

After these supplemental steps have been performed, the process flow continues, as in the first embodiment with determining the total alarm value from the alarm array 119, and continuing with the remaining steps.

According to the combining process 139, the responses are conveniently normalised by multiplication with pre-determined scaling constants so that they are approximately equal for a spherical shielded volume. Each response belonging to coil A₁ B and C is taken as an orthogonal component and a potential target magnitude and potential target orientation calculated for each bag location, before the combined responses are compared 141 with a threshold. ' .

For a simple flat sheet SSV, the magnitude and orientation can be calculated accurately from the responses of the resonant coils and related directly to the size and real orientation of the SSV. For relatively symmetric objects such as spherical shapes the calculated orientation will be 45° to each axis. This result implies only that the target is symmetrical to each resonant coil.

The, advantage of combining responses as components is demonstrated in the ideal situation where the shielded volume response is independent of its orientation. This allows a single uniform threshold that would be higher than any threshold set when considering the individual response components from each resonant coil. This difference desirably leads to a lower alarm rate from clutter.

The magnitude of the threshold can be set up to depend as a continuous function of various parameters that may selectively choose a volume shape and/or orientation. For example, a useful shaped threshold is one whose magnitude depends on the calculated shielded volume orientation. This may be useful when trying to discover preferential sheet-like shielded volumes with a major surface that is parallel to bag panels.

The eighth embodiment is substantially identical to the seventh embodiment, except that it omits the steps of finding the minimum magnitude of the responses from selected areas of the bag, and offsetting the responses 111; and comparing these offset responses to see if they cross their thresholds 113 to calculate an alarm value and shield value and insert these values into the alarm array 115.

The process flow is shown in Figure 14 for this particular embodiment. In the context of the supplemental steps 139 to 143 for combining like responses mathematically and comparing these to with their thresholds to insert values into the alarm array, the offsetting is secondary and can be omitted.

The ninth embodiment is substantially the same as the eighth embodiment, except that it provides an alternative process flow to that of the preceding embodiments, whereby several steps are omitted and refined in favour of the steps associated with mathematically combining clutter removed responses from each coil.

As shown in Figure 15, after the initial steps 101 to 105 are performed, steps 107 and 109 associated with initial checking of the responses with thresholds and supplementing the alarm array if appropriate are omitted, along with steps 111 to

115 concerned with offsetting the responses, and the process moves directly to checking the consistency of responses with the bag length 117. If the responses are found not to be consistent with a looped bag of the measured length, then the best model is found through a comparison 145 that matches identified clutter in the responses.

After subtracting the best model responses from the bag responses to arrive at difference responses, again the comparison of difference responses with thresholds at steps 135 and 137 are omitted and the process moves into directly performing the mathematical combination of like responses 139 and relying upon a comparison of these with their thresholds 141 to determine total alarm values 119, as in the preceding embodiment.

The tenth embodiment is substantially similar to the first embodiment, except that it involves a different coil cluster arrangement 39 to that of the preceding embodiments. Moreover, essentially the same coil cluster 39 is used with coils A, B, C, but one or more additional inductive coils designed to find an SSV in a selected portion of the bag, are provided.

As shown in Figure 16 of the drawings, the coil cluster 39 is formed with an additional coil D 24. Coil D, is a special type of coil for detecting objects near the top of bags. It travels up and down sliding over the top of tall bags. It is special in that it is only designed to be short range. The D coil adds extra detection sensitivity on the top of the bag.

As coil D lies in the plane of coil A, it is placed outside the strong parts of the field of the other coils, to reduce its influence due to its movement.

This system of coils yields good detection results of a shielded volume and provides an ability to discern baggage structures and clutter. It also is a single turn loop.

As shown in Figure 17, coil D 24 can move up and down through the use of a hinge 28 and supporting arm 37. As bags pass through the machine, they generally do not come in contact with coil D. However, occasionally a bag which is higher than the average will push coil upwards. A counterweight (not shown) makes upward movement of coil D easier.

In the present embodiment, coil D is encased within a continuous frame (not shown) so that bags do not get caught on it. Further to this, coil D is made to recede into the frame upon which the hinge 28 is mounted for large bags.

Triangular patches 38 composed of metal sheet are provided to enable the upward movement of coil D without causing a response change relative to the fixed metal structure of the coil cluster. These patches balance the coils response so that movement alone does not induce a significant response on D.

Coil D is operated in a similar fashion to coils A, B, and C in that it is constructed to be of high Q and its resonance parameters are measured through an additional channel D in the transmit and receive electronics as shown in figure 1OB.

The parameters of the coil such as resonance frequencies are chosen in a similar method to that of coils A, B and C. The processing of its response function follows a similar method to that described in the first embodiment to generate a set of alarms and shield positions which are combined with the results of the other coils of the cluster to find if a significant shielded volume exists. Due to its more localised field, coil D improves the position resolution of shielded volumes.

The eleventh embodiment is substantially identical to the tenth embodiment, except that Coil D is located in the middle of the other coil structure, so as not to take up extra space. In this embodiment, the coil is shaped so as to provide a circular field to reduce the interaction with the other coil directions, i.e. a magnetic field that approximately matches that created by a single wire.

This design is considered desirable because, by symmetry, it will have low net flux coupling to the surrounding coils from coil D.

The twelfth embodiment is directed towards an integrated SSV detector and NQR scanner apparatus, where the resonant coils used for NQR scanning and those that are used for SSV detection are shared.

The schematic arrangement of the apparatus for this embodiment is shown in Figure 18, where two subsystems -comprising the SSV detection subsystem 75 and the NQR subsystem 77 provide information from the shared use of one or more resonant coils within the cluster 39 through switching 79.

A bag position subsystem 81 is provided to control the position of a bag conveyed to the coil cluster 39 to be scanned, relative to the coil cluster, by receiving information about the bag position and dimensions and controlling the operation of the conveyor belt 44.

The process unit 83 sends and receives information from all subsystems to control the process and perform analysis on received sensor data that can be passed to a display 85 for the operator.

Coil B is removed and replaced by an NQR coil, which is located serially along the conveyor belt, relative to the remaining shielded volume coils. In this arrangement the NQR coil is able to look for similar SSV features to that -of the B coil. The implementation of the scanning process for SSVs on the NQR coil is similar to that described in previous embodiments, with the required modification that electrical switching 79 is applied so ^'that both the NQR process and shield detection can occur on the same coil. In this case shielded volumes are detected in a similar orientation to that delivered by the coil B through the NQR coil being monitored as to its resonant properties, as the bag proceeds to and from its effective detection volume. The data is correlated with bag location as previously described, combined and processed in a similar manner with the other coii directions.

The advantage of this embodiment is the reduced complexity of additional coils in that the NQR coil already provides a high Q resonant coil that can also provide SSV information similar to the B coil. This partial integration also leads to a reduction in length as allowances to external shield sizes are reduced because the return field of coil B does not have to be allowed for.

The thirteenth embodiment is substantially identical to the preceding embodiment except that system of coils for the SSV are integrated into functioning also as a typical NQR probe to reduce overall machine length, as opposed to coil for the NQR probe functioning as the B coil.

In the arrangement of the present embodiment, coil B acts as the NQR coil or part of that coil. The conducting surface area of coil B has to be relatively open to allow the system of coils to operate effectively as described previously for the extended B coil.

The B coil is effectively lengthened to increase its sample volume so that an NQR scan can be performed over a large volume. This is achieved by coil B being constructed as parallel loops that are spaced apart for the magnetic field from the sides to penetrate and form a complete, flux path without coupling strongly to B coil. This arrangement is shown in Figure 19.

The conductive loops of coil B are electrically linked in parallel to maintain high Q in a resonant circuit. The design has the loops constructed of high conductivity bar or preferably pipe, where the spacing between loops is far enough to overcome proximity effects of conductors.

The process involves:

• the bag entering the system and performing an SSV scan on the bag with the coil array;

• storing the part response for the SSVs once the bag reaches the correct position for the NQR scan;

■ stopping the bag;

• electrically switching the system to perform an NQR scan;

• conveying the bag out of the coil cluster and performing the second part of the SSV scan;

• joining both response sets from the first part and second part of the SSV scan; and

• analysing the response sets as in one or more of the previous embodiments.

The fourteenth embodiment is substantially the same as the preceding embodiment except that the SSV scan data is collected as the bag proceeds into the coil cluster and dispensed with as the bag leaves the coil system.

In this arrangement, the derived data does represent a complete scan of the luggage, although such an arrangement would not be as effective as the preceding embodiment.

The fifteenth embodiment is substantially the same as the thirteenth embodiment, except that the process involves conveying the bag through the coil cluster twice. The first time through the coil cluster an SSV scan is performed on the bag, whereas the second time the bag is specifically positioned for the requisite time in the coil cluster and an NQR scan performed, if required.

The sixteenth embodiment is substantially the same as the twelfth embodiment, except that more than one resonant coil is tuned to the same frequency for the integrated SSV and NQR scanner.

In this case the SSV detection is carried out as before, with the advantage that' during NQR scanning, the signal from the same NQR excitation line is derived from more than one coil to obtain a better NQR SNR than from a single determined axis.

An example of this is the case is where a bag loop may partially shield the explosive signal to the B coil. This circumstance could be recognised from the SSV scan and an appropriate coii chosen for the NQR scan that maximises the NQR signal. In this example either A or C could be chosen to yield a better NQR result as their fields lie in the plane of the loop.

The seventeenth embodiment is substantially the same as the preceding embodiment, except that shield scanning is used with each resonant coil approximately tuned to different explosive/contraband NQR lines of frequencies.

In this arrangement, the SSV detection occurs as with the previous embodiments, except the resonant frequencies are chosen to match more than one NQR frequency. The NQR scans are then performed without the significant retuniπg that is usually required for covering the frequency band of all spectral lines with a high Q coil.

Typically, contraband/explosive lines are separated by MHz so that significant tuning is required to shift a single high Q coil from one frequency to another to achieve an effective result. The retuπing of the NQR coils is accomplished by adding or removing capacitance to the resonant circuit that is often a large fraction, or multiples, of capacitance required for one or another NQR spectral line. To shift this capacitance requires switching which must not compromise the Q for high SNR. This often is a costly procedure to implement that at the same time compromises the performance and reliability of the machine.

With the present embodiment, the coils are arranged to scan the luggage for shielded volumes as the bag enters and leaves the coil structure. In between, an NQR scan is performed in a similar manner as that described earlier, the only difference being that the shielded volume detection frequencies are chosen to be close to the scanned explosive/contraband frequencies.

The process involves:

• performing an SSV scan as the bag proceeds into the coil cluster,

• if necessary, fine tuning the resonant coils based on the shield scan information for that resonant coil,

• performing the NQR scans with the coils either:

(1) serially, ie. one NQR pulse sequence is used after another is applied serially to the appropriately tuned coil, or

(2) potentially in parallel, ie. more than one NQR pulse sequence that are synchronised is applied to the coil cluster at the same time, or

(3) a mixture of (1) and (2).

• conveying the bag out of the coil cluster, and

• recording response data.

As previously described, the response data from the two scans is joined and analysed as a single data set.

As in the previous embodiment the process occurs with all of the shield information gathered in one sweep, then an NQR scan is applied to a repositioned bag. A desirable example would be for the resonant coil B to be tuned to approximately 890 kHz and the resonant coil C tuned to 5.19 MHz to perform both an SSV scan at these frequencies and also perform a PETN and RDX plastic explosive scan without significant retuning. In this example, coil A is used for primarily SSV scanning and is tuned to an intermediate frequency.

It should be appreciate that application of the shielded volume detector to luggage provides a large amount of information that can be used to achieve a better NQR scan- As described in the sixteenth embodiment, using a methodology that takes the information from the shielded volume scan to change the NQR scan method so that it achieves a better result, has great utility.

The eighteenth embodiment is directed towards a combined SSV detector and NQR scanner, substantially the same as the preceding embodiment, but where the processing method involves performing an SSV scan and then finding the position of least coupling to the NQR field to the clutter, using the results of the SSV detector.

As explained earlier the coupling of the magnetic fields is very dependent of the actual location and orientation of the luggage and clutter within. There exist regions with the positioning and orientation variables of a luggage item where the NQR scan is expected to produce a better result from the general bag for the detected clutter. These regions are computed from the SSV detector and the bag positioned so that a better NQR response is achieved. An example of this is where a looped bag is scanned to find the point at which the response changes are a minimum for the field direction that will be used for the NQR scan. In the case where the NQR field is essentially in the direction of coil B, as shown, a looped bag has least coupling when the bag is placed symmetrically relative to coil B. This suggests that the best location for NQR scanning is to place the bag symmetrically relative to the NQR coil.

The present embodiment also provides for using the change in the tuning parameters from the shield detection system to predict the requirements for tuning of the NQR probe. Moreover, the resonance frequency changes found with bag location are scaled and applied to the NQR system so that very near correct fine tune settings are then applied. This generally saves the time required to tune without the SSV detector.

In addition, the frequency shifts and/or Q's found with the shield detection system are used to alter the pulse sequence parameters to achieve a required level of detection. This includes extending the pulse sequence in time in order to average the predicted increase in thermal noise to a lower level.

With several of the embodiments described above, it is desirable to locate the shield volume and/or NQR detected material with more precision. This task can be achieved by modifying the coil arrangement in the manner described in the ensuing embodiment.

Accordingly, the nineteenth embodiment is substantially the same as the first embodiment, except that coils A and C are split into two pieces as shown in Figure 20,

Moreover, coil A is split into two components: coil A1 91a and coil A2 91b; and coil C is split into components: coil C1 93a and coil C2 93b. Coil B 22 still provides the location of an SSV in the direction of conveyor belt motion 34, with the split resonant A coils A1, A2 and C coils C1, C2 providing the location vertically and across the conveyor belt direction, respectively. The coils are made resonant by adding capacitors 95a, 95b, 97a and 97b to A1, A2, C1, and C2 respectively.

The best place to split the described resonant coils is along the symmetry plane whose normal is in the direction of the primary field they produce. These paired resonant coils are arranged so as not to be strongly coupled to each other, which may require them to have additional spacing or shielding from each other.

For detection, the responses are measured in a similar way to that described earlier, except with the addition of two extra channels in transmit and receive. Considering the responses from these paired coils, if the sought SSV object is closer to one coil than the other of this pair, the change in response from that coil is greater, The position of the target in the direction of the axis of separation of the coil pairs is calculated from the response of one coil as a ratio of the overall response of both coils.

An example for the shielded volume detector operation with this coil structure is for detecting a target that is close to the top of the sampling volume. This target location creates a greater change in response signal in the top coil of the pair compared to the bottom. The proximity to the top coil is calculated by the magnitude of that response relative to the total response of the combination. By combining the results from the three position measurements the location of the object is found in a Cartesian coordinate system.

In the preceding embodiments, the SSV detector has been described using a measurement process in a resonant system. This process is also able to be performed with a non-resonant system of coils.

Accordingly, the twentieth embodiment is directed towards an SSV detector system that is substantially similar to the first embodiment, except that the impedance change of each coil is measured instead of the resonant parameters. Here the resonating capacitors are removed from the electrical circuit and the impedance measured directly.

Similar information is obtained from this process as with the resonant coil system, where the inductance and resistance changes that are related to the resonant frequency and Q, also change, correspondingly. This measurement process has the desirable effect in that the sampling frequency is more easily varied.

The measurement of the impedance of each coil is accomplished with off-the shelf components that not only perform the measurement quickly in the megahertz frequency range, but at other frequencies to obtain more information about any shielded volume. The system of the present embodiment operates with at least two different frequencies when sampling the luggage as it proceeds through the described coil structure. Initially, it operates at a low frequency where the ratio of the inductive responses from thick and thin conductive targets is much greater than one due to the changes in skin depth. Then the system operates at a higher frequency, where the ratio of inductive responses from thick and thin conductive targets are close to one. Therefore by comparing the responses from the two different frequencies, coarse knowledge about the thickness of conductive target can be determined and this knowledge can be used in processing to discern thin metal contained shielded volumes for thicker conductors that may include shielded bag loops.

The performance of the SSV detector of the CID type described in each of the preceding embodiments is extremely good with low reject rates and a high probability of detection. However, in the vertical direction the coupling to bag loops can hinder the effectiveness of this system. Hence two alternative arrangements are constructed to detect shielded objects in the vertical direction, and in particular inside bags with reinforcing loops.

Arrayed Inductive Detector (AID):

The first of these alternatives is the AID, which operates in the radiofrequency kHz region. The second of these alternatives is the detection of metal objects by microwaves, which will be discussed later.

Accordingly, the twenty-first embodiment is directed towards an SSV detector system using the AID design. The general configuration of the SSV detector using the AID is different to that of the CID described in relation to the first embodiment, and is shown in Figure 21.

As shown in the drawings, the AID SSV detector comprises a probe consisting of 32 ferrite rods each with a coil wrapped around to produce 32 discrete coil assemblies, which are spaced around a rectangular frame 165. The frame 165 is made of some suitably non-metallic substance such as wood or plastic. The ferrite rods are Amidon material type-33 with a length of 195 mm long and a width of 12mm. The Amidon type-33 material is particularly suited to operating in the low kHz region. The advantage of operating in the radiofrequency region is the magnetic field produced by the coil assemblies is able to penetrate the metallic reinforcing loop of a bag to some extent, unlike the situation with the higher frequencies of the CID, resulting in shielded objects being able to be detected within bag loops. The magnetic field is able to partially penetrate metal reinforcing loops because it is measured over a more confined volume relative to the size of the conductive structure, i.e. the field will induce eddy currents in the structure but because the structure is large, smaller eddy currents will exist to counter the input flux from the sensor.

There are ten transmitting coils 157 and ten receiving coils 159 in the vertical direction, and six transmitting coils 160 and six receiving coils 158 in the horizontal direction. All coils are equally spaced apart by approximately 60 mm. There are more coils in the vertical direction because the distance across the scan area is typically longer than the vertical dimension, hence more coils are required. The vertical and horizontal transmit coils are connected in two separate circuits in parallel. That is, all of the vertical transmit coils are linked in parallel in one circuit and all of the horizontal transmit coils are connected in parallel in a separate circuit.

Ideally the AID frame 165 is mounted inside a metal conductive shield with a conveyor belt passing through it (not shown), almost identical to Figure 4, except that obviously the CID probe is replaced by the AID probe described above. In such an arrangement, before a measurement is begun using the AID, typically a bag is moved along the conveyor through an optical fence identical to 43 in Figure

4. The breaking of the optical fence triggers the measurement process described

• below, after a short time delay to allow the bag .sufficient time to be located within the AID.

To generate a transmit burst on the vertical and horizontal transmit coils, a digital sine wave is synthesised in software at two different frequencies 3.3 kHz and ^• 25 kHz. The purpose of these two frequencies shall be discussed later. These two signals are added together digitally and then transmitted through two output channels of a 16 channel simultaneously updating 100 kHz ADC/DAC card 164. From here the signals are sent to two input channels of a high power audio amplifier 163 and from there to the vertical and horizontal transmit circuits, which, as previously stated, are connected in parallel. The transmit bursts for both the vertical and horizontal coils take 5.12 ms to occur. Although the signals can be transmitted to all of the coils simultaneously, in the present embodiment they are transmitted in bursts to the vertical and horizontal coils, in an interleaved manner, 25-40 ms apart. If the transmit bursts were all transmitted at the same time, then the signal seen on some coils of ferrite rod receivers would be overloaded by the magnetic field from the adjacent transmit rods.

The magnetic field created by the coils of the transmit rods is received by the individual coils on the ferrite receiver rods and input into 16 pre-amplifiers 155, which in turn amplify the received signals into the volts range. These signals are input into the 16 ADC channels of the ADC/DAC card 164, which in turn is connected to a computer 156. The ADC card thus simultaneously samples all sixteen channels at once.

The simultaneous sampling of the channels is superior to multiplexing the channels because it is fast and enables the user to 'slice' through the bag simultaneously, much like an X-ray scanner. If the system were multiplexed then the pixels generated would be staggered in time (and hence in distance along the bag) and thus more difficult to interpret and process. Simultaneous sampling is also less problematic since issues such as finding sufficient time to multiplex so many transmit and receive signals to and from the coils on the rods makes the resolution more dependent on the conveyor belt speed. In the present embodiment, the belt speed is 0.5m/s and so a 60cm bag passes through the scanner in only 1.2 seconds.

The previously described step is repeated approximately 30 times as the bag moves through on a conveyor belt, with a 50-80 ms gap between the slices. This process results in slices which are spaced 3-4 em apart_. and in the process an irnage of the metal contained within the luggage is built up for the operator to 'see' any metal objects within the bag.

As the system has vertical rods and horizontal rods, two images are formed of the bag. At each of the thirty slices, 512 data points are recorded, which corresponds to a sampling time of 5.12 ms if the sampling is performed at 100 kHz. To form the images these 512 data points are baseline corrected by subtracting off the thirtieth time slice. Then each of the other 29 slices for each receiver channel are DC detrended, windowed and Fast Fourier Transformed (FFT) into frequency space. In performing the FFT, the absolute value of Fourier co-efficients (i.e. peak heights) at the two frequencies of interest are calculated and input into two arrays, which form the two metal images at 3.3 kHz and 25 kHz. To aid the following description these will be called Amp25khz and Amp3khz. In separate images the two phases of the baseline-corrected signal are stored. These are called

Phase25khz and Phase 3khz. There are also four corresponding amplitude and phase images of the raw received signal (not corrected for the 30^th time slice). These images are called Raw25khz, Raw3khz, RawPhase25khz, RawPhase3khz. The images containing the raw received signal phases are then background corrected after they have been transformed into a phase by subtracting off the phases calculated in the same process for the thirtieth time slice. In another series of two images, instead of taking the absolute value of the Fourier co-efficients, the raw complex co-efficients at the two frequencies of interest are input into two more images. These are called the CplxReal25khz, Cplxlmag25khz, CplxReal3khz and Cplxlmag3khz.

While it may appear the system can only detect objects in two directions, the system can detect objects in the third dimension. This is because the transmitted field from the coils on the vertical and horizontal rods expands out in three dimensions, which results in the detection of objects that lie in the vertical plane parallel to both the vertical and horizontal rods. Consequently, the system is capable of detecting any shielded object in any orientation.

The field homogeneity should ideally be entirely uniform across the gap between the rods. However, in order to achieve this in practice is very difficult, given that the magnetic field decreases as 1/r³ from most coil designs. The AID design does, however, have a reasonably uniform field, which is helped by the fact that the coils on both the transmitter and receiver rods tends to concentrate field in their vicinity and thus counteract the drop off in field intensity from the coils of the transmit rods. The shape of the magnetic field 167 generated between any two coils of transmit receive ferrite rods 158 and 160 is shown in Figure 22.

The system operates at low power levels not harming passenger's luggage and routinely at a 0.5 m/s conveyor belt speed The system is also relatively inexpensive, and emissions from the device are limited by adding an appropriate metal shield around the device, while leaving an opening for the luggage to pass into and out of the device. As the device is extremely thin (only -40-70 mm wide), the shield used is quite small and the overall dimension of the device adds only a small amount of length to the NQR scanner.

As previously discussed, with respect to Figure 1 , bag loops 2 are contained in most trolley type bags. As these are continuous around the bag structure, any magnetic field 3 that impinges upon the bag from above will induce an eddy current signal 4 in the bag loop. This signal counteracts the impinging field and tends to cancel it. This makes it very difficult to detect metal objects contained within metal loop bags. This is further compounded with the latest trend of manufacturers to include three or more metal loops in bags and steel bars along the bottom of the bag for the extendable trolley handle, although generally this metal section does not form a closed loop.

If this type of bag were to pass underneath an ordinary metal detector then every looped bag would cause an alarm, which does not necessarily accurately represent the shielding ability of the bag.

To overcome this problem in the present system, two images of the bag are formed at two different frequencies. One image is formed at a relatively high frequency (25 khz) and the other image is formed at- a relatively low frequency (3.3 khz). Both of these images contain the signal from a bag loop that, due to the bag loop's size, is large in both amplitude and area. Smaller objects that are in the bag also appear in these images, however the targets almost always induce a signal, which has a slightly different phase to the bag loop signal. Hence superimposed upon the bag loop signal in both images is a smaller signal from a smaller target. The small target can either slightly increase the height of the bag loop signal where it occurs in the image, or slightly decrease the height of the bag loop signal. Invariably most targets increase the height of the bag loop signal. In these two images if the bag loop signal is subtracted out of the images then the smaller target within will be revealed.

Accordingly, this technique is used in the present embodiment to cancel the signal of the bag loop to reveal metal targets contained within. To subtract out the bag loop signal a symmetrical three dimensional Gaussian like surface using a simplex search is fitted to lie just under the peak shape of each of the two images.

This fitted peak is then presumed to represent the bag loop signal and is subtracted out of each image revealing any targets superimposed upon the bag loop signal.

Figures 23, 24 and 25 show the subtraction process graphically. Figure 23 is the image generated from a bag with a reinforcing loop at 25 kHz and Figure 24 is the corresponding image at 3.3 kHz. A 3D surface is fitted to each image (not shown) and then subtracted out to reveal the smaller target as shown in Figure 25, which shows the metal object revealed in the top right hand corner of the bag, whereas in either of the original images it was unable to be seen.

Some objects, such as steel targets, rather than adding to the bag loop signal, actually lower the peak locally. To overcome this problem, a pre-screening step is performed to identify if there is a "dip" in the peak- Once identified the dip is inverted into a "bulge" upon the bag loop signal and then the subtraction proceeds as per normal.

Bag loops can be distinguished from other large metal objects by examining several different parameters. These parameters can include: (i) Size: If the number of points in the 25 kHz amplitude image above a threshold is larger than a minimal area threshold then the signal is deemed to be large enough to be possibly a bag loop.

(ii) 25 khz Peak Height: If the maximum of the Amp25khz is above another threshold then the bag is flagged as possibly containing a bag loop.

RawPhase25khz: At the corresponding point at which the 25 khz image reaches a maximum, the nine closest points in the RawPhase25khz image are averaged. This average is then compared to upper and lower bounds. If it lies within these bounds then it is deemed to be possibly representative of a bag loop.

(iv) Meanar: If the average of the image, formed when the Amp25khz image is divided by the Amp3kHz image, lies above another threshold, then the bag possibly contains a bag loop

If the received signal passes these four criteria then it is flagged as being a looped bag. Many other parameters or combinations of parameters derived from the initial signal processing can be used to distinguish a bag loop from a plain metallic object.

Presently in the luggage market, bag loops tend to fall into three different category types, which for the sake of simplicity will be referred to hereinafter as: Type I₇ Il & III. The first two types dominate the marketplace and Type III has very similar, but not the same, characteristics as Type II. Through experimentation it has been found that the three different types arise from the different arrangement of metal loops in bags. For instance Type I bags have typically a solid steel band around the centre of the bag, steel tubes for a trolley handle, and wire loops top and bottom of the bag for extra reinforcement. Type III on the other hand can have only wire loops top and bottom of the bag.

Bag loop types are detected by examining the magnitude of the Meanar parameter. If this is small in value then the bag loop type is a Type I, if it is medium sized then the bag loop type is Type II, and Type III has the largest values.

The computer 156 performs the aforementioned processes, as well as additional processes according to the specific flowchart shown in Figure 26.

Initially, in order to determine if a shield is present in the vertical image, the 25 kHz bag image signals derived from the sensors are processed 171 to determine if any peaks or maxima lie above a baseline threshold.

If all peaks or maxima lie under the baseline threshold, then the bag is passed as clear 173. If any of the peaks or maxima do exceed the baseline threshold, then the same peaks or maxima signals are checked 175 to see if they lie above a second higher bag loop threshold, which is what would be expected for a bag with a reinforcing loop or laptop. If any of the peaks lie below this second threshold, then it is concluded that it is not a bag loop or laptop, but is probably a metal shield and so an alarm is generated 177. If the peak lies above the bag reinforcing loop/laptop threshold, then the signal area 25 kHz peak height and the RawPhase25khz are used 179 to determine if the signal is a laptop or a bag loop using the aforementioned bag loop detection process. If the signal is not detected as a bag loop then the object is deemed to be a large metal object and an alarm is generated 181, accordingly. If the object is found to be a bag loop then the aforementioned subtraction process 183 is used to cancel the bag loop signal and reveal any underlying objects. Another comparison is then performed 185 to determine if the residual signals lie above a third prescribed threshold representative of the high probability of a significant metal object being present. If any residual signals lie above this third threshold, a significant metal object is deemed to be contained within the bag and an alarm is signalled 187, otherwise the bag is passed as clear 189.

As a back up to the foregoing procedure, the operator can often 'see¹ which bags contain reinforcing loops and which contain laptops and therefore the operator can make a judgment call if he/she feels the computer based decision making process has failed. After completing the vertical image processing, the horizontal image is processed. As generally no bag loops appear in the horizontal image derived from the coils of the horizontal rods, no subtraction is required, hence simply if the peak in the 25 kHz horizontal coil image lies above a specified threshold then an alarm is generated_s otherwise the side image is passed as clear. One slight complication with the side image is that when a bag loop or laptop passes through the device flat, it deflects magnetic field into the horizontal rods, causing a 'phantom' image. This phantom image is easily removed by multiplying the 25 kHz against its phase and determining if any peaks lie above the threshold. This is done because the reflections off bag loops and laptops observed in the coils of the horizontal rods produce a signal with negative phase, whereas 'real' signals always have positive phase. Hence, multiplying the signal against a negative phase turns the phantom image into a negative amplitude and thus it can be removed from the image by removing any objects with an amplitude less than zero in this new image.

The flowchart of the specific process followed by the computer 156 is shown in Figure 27. Initially, the maxima of the 25 kHz signals in the horizontal image are checked 191 to see if any lie above a prescribed threshold. If not, the bag is cleared 193. If so, the 25 kHz image is multiplied 195 by its corresponding phase image. The signal in this new image is then compared 197 to see if it lies above a prescribed threshold. If so, then it is concluded that the signals probably represent a metal object and the operator is alerted 199 by an alarm. If not, then it is concluded that the bag is clear 201.

Vertical targets, which lie in the same plane as the vertical and horizontal rods, are detected by the pattern they produce in the vertical and side images. Upon entering the detection area they produce a signal on both vertical and side images with negative phase, similar to the bag loops and laptops above. Accordingly, the present embodiment is able to detect objects in all three directions by correlating the shape observed in the vertical and horizontal images in space and time. The use of phase and amplitude information is also used to determine the type of metal detected. For instance, it is well known in the field of metal detection that ferrous metal objects produce a negative phase in metal detectors, aluminium foil produces a slightly positive frequency and other metals such as gold, silver and copper produce larger positive phased signals.

There are alternative methods to process the images from a bag that contains metal reinforcing loops to that described in the preceding embodiment.

The twenty-second embodiment is substantially similar to the twenty-first embodiment, except that it employs another method of revealing objects contained within bags with reinforcing loops by scaling the Amp25khz and Amp3khz images such that they have the same height and then subtracting these images.

As shown in Figure 28, steps 171 to 181 are the same as in the twenty-first embodiment; however, instead of performing step 183, at step 209 the Amp25khz and Amp3khz images are scaled to have the same peak height and then subtracted from each other to determine if any anomalies appear present.

The twenty-third embodiment is substantially similar to the twenty-second embodiment, except that rather than subtracting one image from the other, the 3.3 kHz and 25 kHz images are divided point by point. When the 25 kHz and 3.3 kHz are free of any SSVs the ratio of the two images produce a new image which has relatively constant value. If an SSV is present in the bag, then it shows up in this ratio image as a 3D peak.

Similar to the processing described previously if the height of this peak lies above a threshold then an alarm is signalled, otherwise the bag is passed as clear.

This ratio method is particularly useful for detecting objects that lie in the centre of the bag. In this position the subtraction methods can sometimes completely subtract out the object of interest, whereas this ratio method does not.

The twenty-fourth embodiment is similar to the twenty-first to twenty-third embodiments, except that it uses more sophisticated processing techniques than described in the preceding embodiments for the AID design. Moreover, instead of performing a single processing step 183 or 209, an additional step is performed to produce a more accurate image of SSVs detected within the detection volume.

The additional step involves calculation of a "bias factor" which helps to effectively discriminate false SSVs from real SSVs. The bias factor is determined by multiplying several parameters together to create a SignalStrength value, plotting this SignalStrength value as a point on a graph, and determining how far above a demarcation line the point lies. If the point lies below the line then the bias factor is set to zero and hence any peaks in the residual image left from subtraction or division are not detected.

Any number of parameters can be combined to produce the SignalStrength value. It has also been found through experimentation that a different SignalStrength value is required for each bag loop type.

For type I bag loops, the following equation is used to calculate the SignalStrength value:

SignaIStrength=1 /((PeakHeight25/PeakHeight3)/(PeakPhase25/PeakPhas e3))*Meanarmid/Meanar

Where:

PeakHeight25 is the peak height observed in the Amp25khz image;

PeakHeight3 is the peak height observed in the Amp3khz image;

PeakPhase25 is the peak height observed in the Phase25khz image;

peakPhase3 is the peak height observed in the Phase3khz image;

Meanar is the mean value obtained when the Amp25khz is divided by the Amp3khz image and all pixel values are averaged; and

Meanarmid is the same as Meanar except that only 9 points are averaged. The central point of these nine points occurs at the same co-ordinates as the corresponding peak value in the Amp25khz image, and the remaining eight points are the eight pixels that that surround this co-ordinate value.

For Type Il & Type III similar equations are used. For each bag loop that passes through the AID detector, initially the bag loop type is determined by examining the Meanar parameter and then the SignalStrength value is calculated according to the equation for this bag loop type. This value is then plotted on a graph against the average of the RawPhase25khz value. Similar to Meanarmid above, this average is calculated by averaging the corresponding point at which the peak in the Amp25khz image occurs and the surrounding eight points. The result produces a graph like Figure 29 which shows: the bag loops which contain SSVs

223, bag loops which do not contain SSVs 224, and the demarcation line 225 between the two datasets. Any point which falls below the demarcation line is tagged with a zero bias factor and all other points are deemed to be valid SSVs. The bias factor is calculated by determining the distance the point lies above the line. Once the bias factor is determined this number is multiplied against the peak height of the residual left from either the subtraction or division process in embodiments twenty one to twenty three. This new peak height is then compared to a new threshold and if it lies above this threshold then the bag is deemed to contain an SSV, otherwise the bag is passed as clear.

The twenty-fifth embodiment is directed towards an alternative arrangement of the SSV detector using the AID design. In this arrangement, the transmit side of the SSV detector is modified such that rather than simultaneously transmitting the same signal from coils on all rods, each coil transmits at a particular frequency or frequencies.

As shown in Figure 30, the operation of the setup is similar to the operation of the SSV detector of the twenty-first embodiment, except that the power amplifier 163 is now a 16 channel amplifier 163' along with the ADC/DAC card 164' and the coils of the transmit rods 157 and 160 are individually connected to different output channels of this power amplifier. Apart from this, the operation is the same. in the ideal arrangement the transmit frequencies on coil of a transmit rod comprise a low frequency {near 3.3 kHz) and a high frequency (near 25 kHz). As each coil of a transmit rod has its own identifying frequencies, it is possible to avoid cross contamination on the receive side, whereby each transmitter is linked up to a corresponding receiver, directly below for the vertical orientated rods or across for the horizontally orientated rods.

In the parallel transmit system described in embodiments twenty-one to twenty- three, any one receiver receives a signal from all coils of transmit rods, and thus a loss of signal from any one of the coils of the transmit rods results in the detection of an object. However, in the present embodiment, any one coil of a receive rod receives signals from all rods but these signals are separated in frequency and are much easier to distinguish apart.

The previously described signal processing methods can still be used in the present arrangement to determine if an SSV is present. However, instead of constructing a 25 kHz array and a 3.3 kHz array, a high frequency array and a low frequency array is constructed. Thus, where it was previously mentioned that the 25 kHz image signals be subtracted from the 3.3 kHz image signals, now simply the high frequency images are subtracted from the low frequency images.

An additional advantage this arrangement provides over the previously described parallel transmit system embodiments is that crude tomography can be employed. For example, as well as knowing the degree of attenuation in the vertical direction, the degree of attenuation in the oblique directions can be obtained.

Accordingly, the twenty-sixth embodiment is substantially the same as the preceding embodiment, except that the computer 156 employs a method of tomography.

As shown in Figure 31, it can be seen that the first receiver rod receives signal from the coils of all transmit rods. The lines represent magnetic field traversing from the transmit side to the receive side. If a metal object breaks the transmit signal from transmit rod 10 to receiver rod 1, but not the field between transmit rod 1 to receiver rod 1, then it is possible to deduce that the object lies underneath rods 2 to 9. By similarly collecting all information of this type from all receivers, the shape and size of the object can be deduced by reconstruction.

Such reconstructing techniques are called Algebraic Reconstruction Techniques (ART). These techniques enable reconstruction of objects from remotely detected information. These are most commonly used in Computer Tomography (CT) X- ray.

Accordingly, the present embodiment uses such a system to obtain a more accurate image of the object. The method used in the present embodiment for algebraic reconstruction of the objects that lie between the transmitter and receiver rods involves constructing a matrix of elements between the transmitters and receivers. Then the amounts of attenuation seen on the coil of each rod from each transmitter are iteratively fed in, and the matrix is modified such that it agrees with observed attenuations.

This process requires several hundred or thousands of iterations to arrive at a matrix that approximately agrees with all of the signals observed on the receivers. This final matrix contains a representation of any metal objects within the bag in both terms of height and lateral displacement.

In an alternative embodiment, more ferrite rods are added to the system to improve resolution. Resolution is also improved by decreasing the time In between measurements as the bag passes through the ferrite rod, that is, the 50- 80 ms between slices is reduced.

In a further embodiment, an additional set of transmit/receive ferrite rods are also added to the system and slanted in an oblique direction to the direction of travel along the conveyor. This enables direct measurement of thin targets that lie in the direction parallel to all of the ferrite rods. In other embodiments, the choice of frequencies is altered to suit the particular application. For instance, lower frequencies may result in better performance in the subtraction process as compared to 3.3 kHz and 25 kHz. In addition, a multiplexed system is used instead of a simultaneous sampling system.

Microwave Transmission and Reflection Detector (MTARD):

The second alternative design to use in the SSV detector to the ClD₁ is the MTARD. Microwaves have the fortunate benefit of being mostly transparent, to metal loops, such as the metal reinforcing loops that exist in suitcases. The reason for this is that microwave frequencies have wavelengths in the order of millimetres to half a metre, whereas bag loops have dimensions in the order of a half a metre. This means that the metal bag loop only affects the microwave field near the metal loop surfaces, not areas that are further away from the metal loop. In waveguide technology, it is known that waveguides cannot stop frequencies that occur above the cut off frequency of the waveguide. As the metal loop can be treated simplistically as a waveguide then the microwaves will be mostly transparent to the loop,

The twenty-seventh embodiment is directed towards an SSV detector using a MTARD design.

• As shown in Figure 32, the MTARD SSV detector is similar to the AID design in that it comprises a power amplifier 221, which is connected to the coils of the transmitter rods of a probe arrangement including horizontal and vertical transmitter rods 223 and 225, and coils of corresponding horizontal and vertical receiver rods 227 and 229. The coils of the receiver rods are connected to a 16 channel pre-ampljfier 231 , which is selected by a 32 channel ADC card 233 via 16 mixers 235. The power amplifier 221 is driven by a microwave generator 237 under the control of a computer 239, which also controls the operation of the ADC card 261. The microwave generator 237 is also used to operate the mixers 235, when selected by the conpputer 239.

By arranging the transmitters and receivers in an array similar to the AID design, very low power microwaves are able to be transmitted through the bag loops, for the most part uπattenuated. Solid metal objects, however, will stop this transmission and so it is possible to detect metal surfaces on this basis.

This fortunate set of circumstances make microwaves in the range of 0.5 GHz to 30 GHz almost the ideal wavelength at which to detect metal objects within looped bags, and so frequencies in this range are used in the present embodiment- Their wavelength is low enough to achieve, sufficient penetration of typical luggage, interact strongly with metal, particularly thin metal surfaces, have high enough wavelength to give acceptable resolution of the- target in an image and not interact strongly with large metal loop structures found in bags. However, the benefits of this wavelength range are counteracted by some drawbacks. At higher GHz frequencies, the waves begin to be absorbed by materials such as very dense plastics and water, which attenuate microwaves similar to a metal object. Another drawback is that microwaves are easily reflected off metal objects possibly leading to false images.

Hence, there is a small narrow band of microwave frequencies at which it is best to detect SSVs in luggage. From experimentation this range has been determined to be 3 to 10 GHz. At the high end of this range there is low interaction with the bag's reinforcing loop, but a relatively large interaction with plastic items. At the low end of this range the converse is true. Hence, the ideal wavelength to detect microwave objects has been determined to be close to 5GHz.

Due to the fact that high frequency waves (>10 GHz) have limited penetration depth this can also be used to advantage. Since the it will be known that the depth of penetration is limited, high frequency microwaves can be used to detect objects that lie close to the surface of a bag, a location in a bag which is the most likely position for an SSV. Hence, by optimising for the appropriate wavelength, the depth of penetration^' can be adjusted.

The present embodiment overcomes the absorption of the microwave by objects such as plastic and water by determining how much signal is reflected off and how much is transmitted through the object. If the object shows properties that indicate most of the signal is reflected, then it is probably solid metal. However, if the object shows only partial reflection, then the object is probably plastic or some other non-metallic object.

The primary piece of data is the received intensity by each sensor, a significant and often practically useful amount of information can be derived from this alone in the described embodiment. To derive just the intensity, the microwave signal is transformed into an easily monitored DC signal through electronic is rectified and filtered. However, it is preferred that the phase of the transmitted signal be determined as well, and so the present embodiment uses a method involving quadrature analysis through the mixers 235 mixing the received signal in two channels with some of the direct source signal provided by the microwave generator 237 and a 90° phase shifted version of the source signal, and then retrieving the low frequency output from both channels. These two channels are known as I and Q. The phase and intensity information that is passed through I and Q channels is then readily digitized as it is now close to DC.

Phase changes that occur from transmission delays in varying dielectric media, varying path lengths and reflection or multiple reflections can be identified. The method used in the present embodiment for removing reflections from the image is to examine the phase of the returned signal. If the signal phase is greatly different to what would be expected, then the image formed is deemed to be only a reflection and is removed from the image.

Similar to the CID in Figure 4, the microwave design is housed within a shield and a bag is moved along a conveyor belt and allowed to pass through an optical fence. By breaking the optical fence, and allowing a short time delay, the measurement process begins according to the following procedure.

As shown in Figure 32, the computer 239 initiates the microwave generator 237 to generate a single microwave sine wave pulse at 0.5-30 Ghz, which is sent to the microwave power amplifier 221. Like the AID design, here the signal is sent to two lines - one for the vertical transmitters 225 and one for the horizontal transmitters 223. Each transmitter group is connected in parallel and thus transmits simultaneously together. In a variation to the present embodiment, a single traπsmitter is provided instead of a series of transmitters as shown in Figure 32. In the case of a single transmitter the ADC card requires less channels, four channels being sufficient. As microwaves can 'fan¹ out rather than bend like kHz and low MHz magnetic fields, both the vertical and horizontal microwave transmitters can be operated simultaneously. Alternatively, in another variation of the present embodiment, both transmitter groups are operated in an interleaved manner in time, like the AID design. In this variation, lenses, such as wax or plastic domes, are used to focus or defocus the microwave beams, allowing collection of the return microwaves from a much wider area.

Returning to the present embodiment, after being sent to the transmitters each signal is received upon corresponding receivers 227 and 229. From there, these signals are input into the 16 preamplifiers 231, which are then mixed down by the mixers 235 in quadrature to a lower frequency and sampled by the low frequency ADC card 233. The signal is then stored in the memory of the computer 239.

The microwave system is housed inside a shield that absorbs most of the microwaves directed at its walls, This is achieved by coating the walls with a microwave absorbent material, rather than leaving the surface plain metal, as this is highly reflective. This reduces spurious reflections from the walls that may interfere with a signal transmitted from the baggage. The absorption of the transmitted or reflected microwave energy serves to control the leakage of microwave energy to the outside environment.

The pulsing of the bag is repeated for 30 times at 3-4 cm apart and the data generated is stored within the memory of the computer 239. After the sequence of 30 slices has been completed, the data from each slice for each channel is background corrected (using the thirtieth time slice), filtered and recorded. Here the signal amplitude and phase are read for the frequencies of interest and input into two arrays.

The processing performed by the computer 239 follows the flowchart shown in

Figure 33. By analysing 241 the amount of signal representing the microwave image to reveal any large loss of transmitted signal, it is assumed that such signal has probably been attenuated by any metal in the bag 243. If there is no loss, the bag is cleared 245. Signals which appear to be weak are removed and those that appear to show strong attenuation are left, resulting in an image that should primarily contain only metal objects.

From the metal image thus obtained, a further comparison is performed 247 to see if the area of the object lies above a threshold value. If the signal shows a metal object that has the dimensions above a threshold level, then the operator is alerted and/or an alarm is signalled 249 for a metallic SSV. Otherwise the bag is passed as clear 251.

The twenty-eighth embodiment is substantially the same as the preceding embodiment, except that the receivers are arranged adjacent to their corresponding transmitters rather than oppositely.

As shown in Figure 34, in this design the horizontal and vertical receivers 255 and 257 are placed alongside the horizontal and vertical transmitters 259 and 261 on one horizontal side and one vertical side of the frame only. Thus the receivers only measure the reflected signal from a bag that is conveyed into the detection volume of the detector.

Metal objects are strong reflectors of microwaves and hence produce large signals, whereas plastics and the like are relatively weak reflectors and produce weaker signals. The microwaves are projected parallel to the normal of surface that is required to be identified. The received intensity, ie where a 180 degrees reflection has occurred, is approximately proportional to the transverse size of the object.

The reflected microwave intensity is recorded as a function of bag position so that highly reflective areas of the bag are identified along the length of the bag. Again this recorded array forms a response. From this response the extent of the object can be determined along the length of the bag, by determining where the response increased from background and clutter signals through processing. The computer 239 processes the response according to the flowchart shown in Figure 35. The response is initially analysed 263 to determine whether the microwave image shows any large reflection signals. If not, the bag is cleared 265. If so it is concluded that the bag probably contains a metal object 267 and the effective reflective size is found from the measured intensity and length of the identified target. This reflective size is then compared 269 to a threshold to see if the target is sufficient to be an SSV. Those signals that remain after this process are deemed to be SSVs and thus the operator is alerted and/or an alarm is signalled 271. If the area of the object is deemed to not lie above the threshold, then the bag is cleared 273.

The twenty-ninth embodiment is substantially the same as the preceding embodiment, except that microwave transmitters and receivers are placed on the top and the bottom of the frame.

In this arrangement (not shown), the transmitters on the top of the frame transmit simultaneously and the transmitters on the bottom of the bag transmit simultaneously at another time well after the top transmitters have finished transmitting.

This arrangement enables reflections on top of the bag to be detected in the top set of receivers and receivers on the bottom of the frame to receive reflections from the bottom of the bag. This is important because at some shorter microwave lengths the penetration depth becomes lower because of attenuation- of the microwaves by plastics, clothing etc.

The thirtieth embodiment is directed towards another microwave SSV detector, but rather than being a series of transmitters and receivers, as in the arrangement of the twenty-seventh embodiment as shown in Figure 34, there is a single transmitter and receiver or transceiver.

In this embodiment, the spread of microwaves is controlled from the source so that the object is uniformly illuminated, as in a fan like beam that stretches right across the conveyor belt. The advantage of this system is that it is simple and cheap; however it is limited to targets that have a significant component of their normal in the beam direction.

In a variation of the present embodiment, the system is expanded to two more axes by locating two or more transceivers around the axes of the detection volume. In this case the transceivers are operated at the same time, with mutual interference in the received signal reduced by operating at different frequencies, polarizations or operating at different times and interlacing the collected data from each axes as a function of position.

The responses desirably are combined as described earlier mathematically to give an overall response and orientation of reflecting targets, which is compared to predetermined thresholds to determine if a conductive SSV exists.

The thirty-first embodiment is substantially similar to the twenty-sixth and twenty- seventh embodiments, effectively being a combination of both.

The preferred combination of both transmission and reflection microwave arrays used in the present embodiment is shown in Figure 36. As shown, there are 16 transmitters in the vertical and horizontal directions (six horizontal 275 and ten vertical 277) and 32 receivers for both reflected and transmitted signals (six horizontal reflected 279, ten vertical reflected 281, six horizontal transmitted 283 and ten horizontal transmitted 285).

In this arrangement, the microwave generator 237 creates microwave signals, irradiating the bag with a burst of microwaves from each transmitter. The transmitted and reflected signals are captured on both sets of receivers and sent to the preamplifiers 231 where they are amplified, then mixed down to base band by the mixers 235 and then sampled by the ADC card 233. After sampling the signals are sent to the memory of the computer 239 for signal processing.

This process is repeated for 30 slices spaced 4 cm apart. At each slice the signal is stored in an array for later signal processing. Once the 30 time slices have been performed the data that has been saved into the time data array is broken into four separate arrays - vertical transmission and reflection arrays and horizontal transmission and reflections arrays. Initially the vertical transmission and reflection arrays are multiplied against each other, point by point, to produce a new array. Within this new array any object that appears on the image shows up as a strong absorber/reflector of microwaves Fn the vertical direction.

This scenario works best for objects having a reflective surface normal to the direction of the microwave beam. If the shielded target is on an angle relative to the path of the microwaves, then the microwaves are reflected predominantly into a different receiver. In this case the modelling of the measured reflected signal is used to determine if the reflected signal could be correlated with the transmission signal. In particular if the transmission signal begins and ends at the same time as the reflected signal, then it is assumed that both signals originated from the same object. The intensities measured at both receivers are also used to determine which object was responsible for the transmission loss.

For a flat metal object whose surface normal lies in the direction of travel of the conveyor, or there is a strong orientation towards this direction, the detection is more difficult as the transmission tends to be high, except for dips in intensity due to the glancing reflection off the surface interfering with the direct beam, and little reflected energy is seen by the sensors. The response from the transmission receivers, however, has a characteristic form, to help in their identification along with the lack of reflective signal.

For this target orientation, in one variation to the present embodiment, additional microwave receivers are included to measure a chosen specular reflection angle.

Alternatively, as is described in a subsequent embodiment, additional information from another SξV detector is obtained, whereby the TSARD detector is combined with a CID detector, to overcome the difficulty.

The thirty-second embodiment is substantially identical to the preceding embodiment, except that the transmitters are no longer connected in parallel, but are connected separately to a multiplexer. As shown in Figure 37, a multiplexer 287 is used on the transmit side, and the receivers are simultaneously sampled. This arrangement permits the user to turn on one transmitter and receive any signal on all receivers simultaneously.

By using this method it is not only easier to determine the amount of microwave signal transmitted through a bag for a particular transmitter but to* also determine where the reflections originated from within bag, as the object must lie along a line from the excited transmitter and the opposing receiver which measures the transmission signal. By cycling through all sixteen transmitters and measuring the signals on all 32 receivers, there will be a clearer picture built up of the objects within the bag. An Algebraic Reconstruction Technique (ART) is then used to reconstruct the object that caused the signals to be observed. As discussed before, the ART approach is extensively used for Computerised Tomography (CT) scanners to reconstruct objects detected via the use of X-rays, and so will not be further described here.

One technique that enhances the microwave detection of shielded objects is the use of multiple frequencies. Plastics and water are highly absorbent at microwave frequencies above 1 GHz and have reduced absorption below 1 GHz. On the other hand, metal objects show strong absorption regardless of the frequency. Therefore by measuring absorption responses at two or microwave frequencies, greater discrimination between plastics and metals is able to be achieved, while gaining the higher resolution and greater metal reinforcing loop penetration obtained at the higher frequency. If an object shows strong absorption at all frequencies then it is more likely to be a metal object. _t

The thirty-third, thirty-fourth, thirty-fifth and thirty-sixth embodiments are substantially similar to the twenty-seventh, twenty-eighth, thirty-first and thirty- second embodiments, respectively, being variations of each,

In these embodiments, similar to the AID design, instead of transmitting at a single microwave frequency, multiple microwave frequencies are used. The method that is used to achieve this involves adding suitably selected microwave generators of different frequencies. Hence as in any one of the arrangements of the twenty-sixth, twenty-seventh, thirty-first and thirty-second embodiments, as shown in Figures 34, 35, 36 and 37, respectively, the microwave generator 237 in the corresponding embodiments thirty-three to thirty- six produces two or more microwave signals and these signals are sent to the transmitters and are received on the receivers. The multiple frequency receivers are then mixed with the appropriate transmitter signal frequency to arrive at I and Q signals. These signals are recorded and processed to intensity and phase of the received signal at each frequency, to determine if it has characteristics of metal or has the characteristics of plastic, water or the like. If it has been determined that the signal appears to be a metal object and the size of the object is significant, then the operator is alerted or an alarm is signalled.

It should be noted that there are many different ways of achieving the same result in microwave detection of metal objects. Techniques such as microwave tomography, which uses the aforementioned ART technique, and time domain reflectometry could also be performed. In certain embodiments, the transmitter is designed to operate at higher frequencies such as the terahertz range to improve resolution of the device. Furthermore, as with the AID design, better resolution is achieved by moving to a finer spacing between the transmitters and receivers and taking more time slices.

Combined System:

Due to the compact nature of the AID and MTARD designs, it is possible to combine two or more designs, or parts thereof, within one shield structure. It may be desirable to reduce potential interference to separate each detector by a metal plate, however since each detector operate in very different frequency ranges with relatively narrow bandwidths there is little need.

This combinatorial design has advantages over any of the previous designs used alone. The processing methods from the combined system would each use their corresponding individual processing method, but combine the results at the ^■7R _

highest level, through a simple logical operation, eg. find the result from each detector and "OR" them together to produce a final detection result. . An alternative method is to weight the result produced by each detector relative to the other, based on its ability relative to the other detectors, then calculate the result of the total unit. A further method again is to use information from both detectors in combined lower level processing, where the measured attributes of the target bag are calculated to arrive at a self consistent model for all detection schemes. An example is to use the information derived by each detector to define clutter that occurs in the image/response of each detector, then use that result to remove clutter from the image/response of each detector. Another example is to use the image of each detector to enhance the probability of detecting a shield by comparing information at each location within the bag arriving at a correlated identification of the SSV.

Some desirable combinations of the three principal techniques previously described will be discussed below, but these do not limit the combinations available to be used from the described components.

The thirty-seventh embodiment is directed towards a combined AID and CID SSV detector, where components of the CID detector are added to the AID detector to enhance the operation of the latter.

In general terms the AID design can be improved by combining two planar SSVs so that their respective normal axes are coaxially disposed, in the present instance, along the conveyor direction. This is achieved in the present embodiment, by combining the AID design with a ClD design using just the B coil component. In this manner, the results of the B coil of the CID design combined with the vertical and horizontal images derived from the AID design result in a superior system to that of a single SSV AID detector.

Figure 38 shows the two-design system of the present embodiment with the CID design 300 and the AID design 310 surrounded by a shield 330 and separated by a metal plate 340. In this case the coil cluster of the CID design only consists of coil B. As in the AID design, the decision making process on the vertical and horizontal coils of the AID design follows that described in any one of the twenty-first to twenty-fourth embodiments described with reference to Figures 26 - 29. For coil B₁ the detection is performed according to the first, seventh, eighth and ninth embodiments as described with reference to Figures 12-15, respectively.

If either system detects an SSV₁ an alarm is signalled or the operator is alerted, otherwise the bag is passed as clear.

The thirty-eighth embodiment is substantially identical to the preceding embodiment, except that all of the coils A, C, and D of the coil cluster in the CID system are included, and all results from all coils are used to determine if a metal object was present.

To combine results from this arrangement, a Boolean AND or OR is used. As an example, under the Boolean AND condition, if either the A or D coils detect an SSV and the coils of the vertical rods of the AID design detect an SSV, then in the combined result a positive detection is signalled. However, if only one or none of the systems detect an SSV the bag is signalled as clear.

Similarly, under the Boolean OR condition, if both or one of the systems detect an SSV₁ then in the combined result an SSV will be determined to have been detected. Otherwise the bag is passed as clear.

The present embodiment also provides for processing at the lower level whereby information derived from the AID design is used to calculate the approximate dimensions, location and yaw of a detected looped bag, and this information is then passed on to the CID processor unit to be used in its calculations to improve the modelling of clutter.

The present embodiment also provides for the overlaying of responses as a function of bag travel from both detectors, so that unless the SSV occupies the same volume, the correlated response does not cross a set threshold. The thirty-ninth embodiment is directed towards a combined CID and AID SSV detector, where components of the AID detector are added to the CiD detector to enhance the operation of the latter.

The CID design can be enhanced in detecting a planar SSV whose normal is vertical by combining it with the AID design. This is particularly useful for bags containing a significant amount of small clutter or conductive bag loops, where the combination of broad uniform detection and localised but non-uniform detection provides very different pieces of information from each detection system.

In the present embodiment, a reduced version of a combined ClD and AID design combines the vertical ferrite rod sensors of the AID design with the B and C coils of the CID design to produce a system, which is superior to either system alone.

in Figure 39, the coils of the ferrite vertical rods 350 of the AID design are combined with the B and C coils 360 of the CID design, but are disposed in two separate chambers, separated by a metal plate 340 and surrounded by a shield 330, as in the preceding embodiment.

In a variation of the present embodiment, both the coils of the vertical rods 350 and the B and C coils 360 are used in the same chamber and the frequency shift design is employed to avoid interference.

The decision making process for the combined system of the present embodiment follows that described in the first, seventh, eighth and ninth embodiments with respect to Figures 12-15 and the twenty-first to twenty-fourth embodiments with respect to Figures 26-29, as in the thirty-sixth embodiment.

If any of these methods detect an SSV, then an alarm is produced or the operator is alerted.

The fortieth embodiment is substantially identical to the preceding embodiment, except that coils A and D of the CID design and. the horizontal coils of the AID design are left in the system and all results from all coils are used to determine if an SSV is present.

As before, the processing occurs at the high level where the results from each system are combined using Boolean logic or a weighted system is used. In addition, the data is combined at the response level from each system to more correctly describe the bag for a combined result.

It is possible to pass information between the AID and CID designs, to enhance the detection process. Accordingly, the processing unit of the computer is adapted to transfer this information and process it to enhance the detection results that can be achieved by either detector design alone.

This is accommodated two ways. Firstly, because the field profiles are slightly different, the detection of an object in one image allows its precise size and location to be determined. For instance, an object, which lies close to the coils of the AID design, produces a large signal, but this decreases as its relative location to the coils is further away from these coils. The CID, on the other hand, detects an object with a similar signal whether it is close or further away to the coils. Therefore the present embodiment combines these results to gauge the depth of the object and its size. This information provides a better determination of whether the object was an SSV.

Secondly, the results from these two systems are combined so that the CID design is provided with information on height, width and length of the reinforcing loop to correctly remove it from the signal observed on the B coil. All of these parameters are approximately derived from the AID design and consequently on passing them through to the CID processing software, better modelling of the reinforcing loop of the bag is achieved.

An improved detection system will also be obtained by combining either of the described magnetic methods with a transmission and/or reflective microwave technique. The intention of this combination is to search areas of the bag that might be partially obscured to the magnetic methods and to provide more informatioπ to processing algorithms, such as location and size of any baggage loops or metallic clutter.

For instance, the reflection of microwaves from a looped bag and its contents would enhance the detection of shielded volumes located at specific orientations and locations principally near the surfaces of the bag, compared to magnetic methods alone. It has been found that there are areas in some bags and orientations of shielded volumes that have lower sensitivity when compared to other orientations for the magnetic methods. These areas of the bag where the magnetic methods have lower sensitivity determine the physical location of microwave transmitters and receivers for the best reception of reflected and/or transmitted energy from the bag.

Accordingly, the forty-first embodiment is directed towards a combined ClD and MTARD SSV detector system, where the components of a simple reflective microwave system are combined with the CID design.

In this embodiment, only a single microwave transceiver, as described earlier in the thirtieth embodiment is provided to enhance the detection in certain locations of the bag with special shield formats.

In this arrangement, the microwave detector would principally look for shields that are planar and lie parallel to the conveyor belt, ie have a vertical normal, to enhance detection of same compared to the CID alone. Hence the microwave system is situated to view the specular reflected beam from these targets. This enhances detection, especially in the case of high rates of clutter, including the continuous conductive bag loop.

The processing occurs, as described earlier, at a number of different levels, from combining logically the high level result of each independent system to combining information at the response level for better clutter and shield discrimination. In particular, the present embodiment provides for a processing algorithm that weights the detection result of each method according to measured parameters from both techniques. For instance, if a large conductive bag loop was perpendicular to the beam direction of the microwave then the microwave response is more highly weighted than the detected response from coil A.

By combining the microwave design with the CID design, better sensitivity to objects contained within looped bags is achieved, as the microwaves easily penetrate the bag loops, whereas the low MHz frequencies have more difficulty in achieving this.

The forty-second embodiment is directed towards a combined AID and MTARD SSV detector system, where the combined images of the AID design and the microwave design are used for detecting SSVs.

(n this embodiment, images from both the AID and MTARD designs are correlated so that reflections in both images are removed to obtain a more accurate detection can be achieved by either design alone.

The information obtained from the microwave method is processed with information obtained from the AID method.

The two principal advantages of using either of the two systems described in the thirty-second and thirty-third embodiments, are: better probing of specific areas of the bag and/or shielded volume orientation: and more constraints on the model of the baggage contents, which includes any metallic structure that supports the bag.

It may be desirable to combine all^' three techniques into one shield detector system to achieve a machine that has the highest probability of detecting shielded objects.

Accordingly, the forty-third embodiment is directed towards a combined SSV detector including the CID, AID and MTARD designs.

With reference to Figure 40, the three designs are shown to be serially positioned so that the CID design 420; the AID design 410 and MTARD design 400 are combined within the one shield 430, with each design separated by metal sheets 440 to stop each design interfering with the others.

In the seria! design it is advantageous to rearrange the order according to specific engineering issues such as EM mitigation to and from the outside. For length restrictions, an overlapping combination is constructed with care taken to reduce the mutual interference between designs.

In the. preferred arrangement of the present embodiment the combination specifically includes the CID's B coil or the NQR coil, and the C coil, the AID'S vertical rod array system and a coarse microwave system, particularly designed to look for shielded targets in the plane of the conveyor belt. The microwave system utilises reflection and/or transmission to differentiate targets in this plane.

This system searches for SSVs in all orientations, as it is able to produce sufficient location information for any found targets, and provide better selectivity with conductive looped bags, since the microwave component has improved contrast between real SSV targets and bag loops in the direction of the plane of the conveyor belt, compared to the magnetic methods.

The image produced from this combination would aid in the further assessment through possibly hand searches or combining with other technologies such as X- Ray.

As described in the preceding embodiments, ordinarily the SSV detector is placed in front of a QR scanning system. This placement allows information derived during the SSV scanning to be fed into the QR scanning process to aid in the detection of explosives. The information that can be fed into the QR scanner includes:

(i) Whether the bag scanned was fully enclosed in metal,

(ii) Whether the bag contained a reinforcing metal loop, (iii) Whether the bag contained significant amounts of metal,

(iv) An estimate of the detuning that will occur when the bag is within the QR scanner.

(v) Whether the bag's contents were emitting RF noise.

In the forty-fourth embodiment, which is substantially similar to any one of the preceding embodiments, information derived from the SSV detector is passed to the QR scanner to aid in the detection of explosives within luggage.

As shown in Figure 41, the bag to be scanned is transported into a SSV detector, where it is scanned for SSVs as it travels S90. If required, during this scan the SSV detector also listens passively for RF emissions. After processing the results of the SSV scan S100 some of the information is used to either adjust the ■ contents of the bag scanned or it is used to adjust the QR scanning process

S120. When adjusting the bag scanned, primarily any active electronics that are detected during the passive scan are switched off. When adjusting the QR scanner or the QR scanning process, any of the alternatives discussed below are implemented, In the final step S130, the QR scan is performed, unless the bag is fully enclosed in metal in which case the bag is not scanned.

A small percentage of the luggage that flows through a typical airport is fully enclosed in metal. These metal cases almost completely block the QR scanner's magnetic field from entering the suitcase and therefore there is little value in scanning these cases inside the QR scanner. In fact to do so is inefficient use of scanning time, as the time could better spent to scan another bag where the magnetic field penetration is normal. The AID, CID and MTARD detectors can help in identifying these bags and prevent them from being scanned. Accordingly, in the present embodiment, the processing means includes a full metal circumscription routine to determine whether baggage disposed within the detection volume is fully enclosed in metal. In the case of the AID detector, such kinds of metallic baggage appear as large 2D objects with very large 25 kHz maximum peak heights similar to the image shown in Figure 24. The full metal circumscription routine ascertains whether the two dimensional size of the object is above a prescribed threshold value and the maximum of an image of the object at a prescribed frequency is above a prescribed threshold and causes the processing means to trigger the alarm means if so.

Hence, as shown in Figure 42, if the two dimensional size of the metallic object is' above a certain value and the maximum of the 25 kHz image is above a certain height then the bag is flagged as being fully enclosed in metal and it is not scanned within the QR scanner and the operator is alerted that that bag should be hand searched.

In the case of the CiD detector, the full metal circumscription routine ascertains whether the resonance frequency shift is beyond a prescribed limit and causes the processing means to trigger the alarm means if so. If the resonance frequency shift is beyond a certain limit, then the bag is flagged that it is not to be scanned by QR and it should be hand searched. .

In the case of the MTARD detector, the full metal circumscription routine ascertains whether the size of the reflected microwave signal or the transmitted microwave signal lies above a prescribed threshold and causes the processing means to trigger the alarm means if so, to indicate that it should not be scanned by QR and needs to be hand searched.

While other bags are not fully enclosed in metal, they do contain significant amounts of metal. These metallic objects can either be the metal reinforcing loop that exists around the centre of many bags or it could be other objects. The presence of large amounts of metal within a bag results in the strength of the applied magnetic field to be diminished. This magnetic field loss is caused by eddy currents being induced on metal surfaces and these eddy currents generate opposing magnetic fields which cancel the applied magnetic field. The loss of transmit magnetic field in the QR scan process means that the QR signal received back from quadrupolar nuclei is also weaker and hence the QR scanner is less likely to detect the explosive. To counteract this loss the transmit B field is increased to counteract the loss as shown in Figure 43.

Accordingly, the processing means also includes a reinforcing loop routine to determine whether an object disposed within the detection volume is enclosed by a metal reinforcing loop or contains significant amounts of metal.

In the case where the SSV detector is an AID, the reinforcing loop routine ascertains whether the integrated amplitude peak of a generated image of the object at a prescribed frequency is above a prescribed threshold and causes the processing means to signal the transmitting means or the probe to increase the magnetic field strength applied to the object sufficiently to counteract the reduction in the magnetic field strength otherwise arising

Hence, the method for this technique is as follows:

(a) The AID detector produces two dimensional peak(s) indicating the presence of metal, similar to what is shown in Figure 24.

(b) If the integrated peak height under the peak(s) lies above a certain threshold then the B field that is applied to excite the QR nuclei is increased to counteract the loss of B field.

In a variation to the present embodiment, the reinforcing loop routine in response to ascertaining whether the integrated amplitude peak of a generated image of the object at a prescribed frequency is above a prescribed threshold, causes the processing means to signal the transmitting means or the probe to lower the QR detection threshold instead of increasing the B field. This adjustment has the same net effect as shown in Figure 44. In another variation to the present embodiment, the reinforcing loop routine in response to ascertaining whether the integrated amplitude peak of a generated image of the object at a prescribed frequency is above a prescribed threshold, causes the processing means to signal the transmitting means to increase the pulse sequence length sufficiently in order to gather more signal to counteract the effect of the metal reinforcing loop or significant amount of metal within the object. In this manner, the QR signal is detected with the same intensity as if the metal was not present in the object at all as shown in Figure 45.

Similarly in the CID detector if the resonance frequency has moved beyond a certain value then the B field is increased. Thus the reinforcing loop routine ascertains whether the resonance frequency shift is beyond a prescribed limit and causes the processing means to signal said transmitting means or said probe to increase the magnetic field strength applied to the object sufficiently to counteract the reduction in the magnetic field strength otherwise arising.

In a variation to this embodiment involving the CID detector, the reinforcing loop routine in response to ascertaining whether the resonance frequency shift is beyond a prescribed limit, causes the processing means to signal the transmitting means or the probe to lower the QR detection threshold.

In another variation to this embodiment involving the CID detector, the reinforcing loop routine in response to ascertaining whether the resonance frequency shift is beyond a prescribed limit, causes the processing means to signal the transmitting means to increase the pulse sequence length sufficiently in order to gather more signal to counteract the effect of the metal reinforcing loop or significant amount of metal within the object.

In the case where the SSV detector is an MTARD, the reinforcing loop routine ascertains whether the loss of the reflected microwave signal or the transmitted microwave signal lies above a prescribed threshold and causes the processing means to signal said transmitting means or said probe to increase the magnetic field strength applied to the object sufficiently to counteract the reduction in the magnetic field strength otherwise arising. In an alternative to the present embodiment involving the MTARD, the reinforcing loop routine in response to ascertaining whether the loss of the reflected microwave signal or the transmitted microwave signal lies above a prescribed threshold, causes the processing means to signal the transmitting means or the probe to lower the QR detection threshold.

In a further alternative still to this embodiment involving the MTARD, the reinforcing loop routine in response to ascertaining whether the loss of the reflected microwave signal or the transmitted microwave signal lies above a prescribed threshold, causes the processing means to signal the transmitting means to increase the pulse sequence length sufficiently in order to gather more signal to counteract the effect of the metal reinforcing loop or significant amount of metal within the object .

The introduction of metallic objects into a QR coil causes the resonance frequency of the coil to shift. This frequency shift occurs because the metal object lowers the inductance of the coil and the resonance frequency is dependent upon the inductance (Equation 5). To counteract this loss in inductance, extra capacitance is switched into the circuit which restores the circuit to the correct resonance frequency.

ω2=l/(L.C) ..., (5)

where: ω = angular resonance frequency;

L = inductance;

C = capacitance.

Accordingly, the processing means includes a detuning estimation routine to estimate the degree of detuning arising from the amount of metal detected within ^• an object disposed within said detection volume and a capacitor switching means, whereby said detuning estimation routine causes the processing means to signal said capacitor switching means to switch extra capacitance into the probe to counteract the loss in inductance arising from the metal within the object.

The retuning process involves the following steps:

(a) Bag is moved into the QR coil.

(b) The capacitors are swept from the lowest value through to the highest value,

(c) At each capacitive step a small fixed frequency voltage signal is transmitted through the coil.

(d) At each capacitive step the signal the coil generates is received upon a small receiving loop outside the QR coil located halfway between the coil and the shield.

(e) At each capacitive step the root mean square (RMS) value of this received signal is recorded.

(f) Then a plot is produced of capacitive value versus RMS value.

(g) The maximum of this plot is where the resonance point lies and capacitance point can be found by obtaining the corresponding x axis value.

(h) The capacitance value obtained is then used to tune the coil to the correct resonance frequency.

Relatively speaking, this process can take a long time, i.e. in the order of seconds, to perform. Ideally it should be as small as possible to maximise throughput of bags through the machine.

The CID has within its structure a single turn coil that is orientated in the same direction as the standard single turn coil used in QR scanning of luggage, as is shown in Figure 5 as the B coil 22. The frequency shift measured on this coil indicates the expected resonance frequency shift to be observed within the QR coil.

Thus in the present embodiment involving a CID, the B coil is used to tune the QR coil, rather than performing a tuning sequence once the bag is within the QR coil, as was the case with the previous method used in other embodiments. The new method requires a calibration stage,.as shown in Figure 46, before it can be used during QR scanning, however, this calibration stage needs to be only performed occasionally.

The calibration method is as follows:

(a) A first bag passes through the ClD detector's B coil and a plot of frequency shift versus distance is calculated (similar to the "B" plot in Figure 11).

(b) The area under this curve is integrated and used to produce an "integrated area signal".

(c) The same bag would then pass into the QR coil and a normal tuning sequence would be performed from which a capacitance value is derived.

(d) Then steps (a)-(c) are repeated over and over for different bags with varying metal content and the "integrated area signal" and capacitance values are stored.

(e) These are then used to generate another plot of integrated area versus capacitance value.

(f) A curve of best fit and corresponding equation are then calculated.

The area underneath the plot needs to be integrated because of the difference in length between the B coil in the CID detector and the length of the QR coil. As the QR coil is much longer, the accumulated frequency shift over a distance needs to be calculated rather than use a single instantaneous value. The normal measurement procedure as shown in Figure 47 begins by:

(a) A bag passes through the ClD's B coil and the integrated area, signal is calculated.

(b) This value is input into the calibration equation determined in step (f) above, to calculate the required capacitance to tune the coil.

(c) While the bag is moving into the QR coil the capacitance calculated in step (b) is set as the tuning capacitance.

(d) The QR scan begins.

This process negates the need for a tuning sequence and effectively allows more time for the QR pulse sequence to operate, generating more SNR or, rf the pulse sequence was left unchanged, increases the throughput rate of the QR scanner which decreases passenger frustration.

A further feature of the present embodiment involves the processing means including an RF detection routine to detect the presence of any emitted RF from an object disposed within said detection volume whilst identifying the presence of an SSV and cause said processing means to trigger the alarm to signal the presence of emitted RF from the object.

While the bag is moving through the CID or AID detectors, the transmitting and reception phase occurs for 5ms out of 40ms for every slice. This means that there is effectively 35ms in which both systems are idle. Accordingly, some of this time is used to perform a passive scan of the bag as it passes through each detector. A passive scan does not reveal any metallic objects contained within the system but 'senses any RF noise emanating from the bag. The sensing of this noise is used within the QR detection process to flag the operator that there was noise coming from the bag and this may have caused a false alarm or alternatively the operator opens the bag turns off the electronics and rescans the bag. Normally the electronic items once switched off do not cause a false alarm. With the involvement of an AID detector, the process is as follows;

(a) During this passive listening stage, no transmission of power occurs and the 16 receivers sample the noise induced on each coil

(b) 2D images of the bag are generated as discussed previously at 25kHz and 3.3kHz.

(c) These images are scanned to determine if any peaks above a threshold are present.

(d) If there are peaks present, then the operator is signalled that there is electronic noise within this bag.

(e) The operator then either hand searches the bag or turns the offending item off and rescans the bag.

Similarly for the CID, the passive scan occurs and the 3 resonant coils simply listen for noise emanating from the bag. Obviously in step (b) the ClD would only generate 1D graphs as shown in Figure 11. Figure 48 shows the general methodology for detecting and isolating RF noise emissions.

The forty-fifth embodiment is substantially similar to the forty-fourth embodiment, but involves the set up of a pre-screening station some distance prior to and separated from the QR scanner. This pre-screening station consists of all or of some of the coils of the AID and ClD designs and is primarily designed to listen to RF noise coming from bags.

Accordingly, a pre-screening station is built housing either the AID or CID designs or both.

If the CID design is used, the arrangement would look identical to the CID design in Figure 4, except that the TxJRx Unit 47 would become a receiving unit only and the processing unit would perform different tasks. In this embodiment the coils receive noise as the bag passes through the SSV detector, and Fourier Transforms this noise into frequency space. If this noise lies above a predetermined threshold, which indicates that significant noise exists, then an alarm is sounded.

Similarly, if the AID device in Figure 21 is incorporated, the structure is similar to that shown in Figure 4, except that the power amplifier 163, and transmit rods 160,157 are removed from the system so that it functions as a passive listening device. As previously explained the bag is moved through the AID device and a series of thirty slices are taken. At each slice the receiver collects noise emanating from the bag for 5ms, Fourier Transforms this data into frequency space and if any peaks lie above a predetermined threshold then an alarm is signalled. The AID design because of its imaging capability can locate the object in two dimensions whether vertically or horizontally. Such images will save the operator time in trying to locate the offending item.

'If a bag scanned in the pre-screening station has been found to contain electronics that are switched on, then the operator of the system instructs the bag owner to switch off the electronics before the bag is passed through to the QR scanner.

It should be appreciated, that many different embodiments have been described herein, and that the list of embodiments is not to be considered as exhaustive, and hence the scope of the invention is not limited thereby. In practice, other embodiments of the invention could be envisaged and adapted to work equally as well by any one skilled in the art. For example, the basic coil configurations described for the coil cluster are meant to represent some useful arrangements and are not intended to be limiting of the invention. As will be apparent to those skilled in the art, a number of arrangements are possible from the described elements of each embodiment, which include the increase or decrease in the number of measurement axes.

Claims

The Claims Defining the Invention are as Follows

1. A detection system comprising:

an electromagnetic scanner for detecting explosives and narcotics in objects using an electromagnetic detection technique; and

a significant shielded volume (SSV) detector having multiple detection axes for detecting SSVs in a detection volume containing the objects in advance of said electromagnetic scanner;

wherein said SSV detector is capable of determining whether an SSV exists within said detection volume that would otherwise shield the substance within the object from being detected using said electromagnetic detection technique.

2. A detection system as claimed in claim 1, wherein said electromagnetic detection technique is NQR and is provided by an NQR scanner, or X-ray and is provided by an X-ray scanner, or a combination of both.

3. A detection system as claimed in claim 1 or 2, wherein is a clustered induction detector (CID), an array induction detector, or a microwave transmission and reflection detector (MTARD), or any combination of same.

4. A detection system as claimed in claim 1 or 2, wherein is disposed in advance of the electromagnetic detector.

5. A significant shielded volume (SSV) detector for detecting SSVs within a detection volume comprising:

transmitting means and receiving means to drive said probe; processing means to operate said transmitting means in conjunction with said receiving means and process said signals received by said receiving means to identity an SSV within said detection volume; and

alarm means to alert an operator of the presence of an SSV detected thereby.

6. An SSV detector as claimed in claim 5, further comprising a clustered induction detector (CID)₁ an arrayed inductive detector (AID) or a microwave transmission and reflection detector (MTARD), or any combination of same.

7. A probe for a CID apparatus for detecting a significant shield volume (SSV) within an article passing through a detection volume in a prescribed direction, comprising".

a first coil to produce a substantially vertical magnetic field within the detection volume;

a second coil to produce a substantially horizontal magnetic field within the detection volume, parallel to the prescribed direction; and

8. A probe as claimed in claim 7, wherein the first and third coils are single turn saddle coils and the second coil is a single turn coiL

9. A probe as claimed in claim 7 or 8, wherein two small continuous single turn loops are added as shim coils to improve the uniformity of the magnetic field of the third coil for producing a magnetic field horizontally and perpendicular to the prescribed direction.

10. A probe as claimed in claim 7 or 8, wherein the single turn saddle coil for the first coil for producing the magnetic field in the vertical direction may be replaced by a single turn coil, comprising two loops connected in parallel.

11. A probe as claimed in claim 7 or 8, wherein the second coil for producing a magnetic field in the horizontal direction is replaced by a single turn coil comprising two loops connected in parallel.

12. A probe as claimed in any one of the preceding claims, wherein, another small single turn coil is attached to a hinge and is arranged to brush over the article passing through the detection volume.

13. A probe as claimed in claim 12, wherein the field from the small single turn coil is substantially in the vertical direction but has a short range.

14. A probe for an AID apparatus for detecting an SSV within an article passing through a detection volume in a prescribed direction comprising: a series of coils wound upon ferrite rods for both transmitting and receiving signals to and from said detection volume for detecting an SSV within the article.

15. A probe for an MTARD apparatus for detecting an SSV within an article passing through a detection volume in a prescribed direction comprising: a series of microwave transmitters and receivers.

16. An SSV detector for detecting an SSV within an article passing through a detection volume in a prescribed direction that includes some or all of the following probes:

wherein the first and third coils are single turn saddle colls and the second coil is a single turn coil; (ii) a series of coils wound upon ferrite rods for both transmit and receive; and

(iii) a series of microwave transmitters and receivers.

17. A method for detecting a significant shield volume (SSV) within an article passing through a detection volume in a prescribed direction, comprising:

positioning the object in the preferred position ready for QR scanning; and

scanning the object using QR.

18. A method for detecting as claimed in claim 17, including sensing the position within the detection volume where the QR field is able to penetrate most of the item; and positioning the item based on this information.

19. A detection system substantially as herein described with reference to the accompanying drawings as appropriate.

20. A significant shielded volume (SSV) detector for detecting SSVs within a detection volume substantially as herein described with reference to the accompanying drawings as appropriate.

21. A probe for a CID apparatus substantially as herein described with reference to tbe accompanying drawings as appropriate.

22. A probe for an AID apparatus substantially as herein described with reference to the accompanying drawings as appropriate.

23. A probe for an MTARD apparatus substantially as herein described with reference to the accompanying drawings as appropriate.

24. A method for detecting a significant shield volume (SSV) within an article substantially as herein described with reference to the accompanying drawings as appropriate.