CN102929903B - Rapid video retrieval method based on layered structuralized description of video information - Google Patents
Rapid video retrieval method based on layered structuralized description of video information Download PDFInfo
- Publication number
- CN102929903B CN102929903B CN201210230539.7A CN201210230539A CN102929903B CN 102929903 B CN102929903 B CN 102929903B CN 201210230539 A CN201210230539 A CN 201210230539A CN 102929903 B CN102929903 B CN 102929903B
- Authority
- CN
- China
- Prior art keywords
- video
- information
- description
- metadata
- node
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 36
- 238000012544 monitoring process Methods 0.000 claims abstract description 68
- 238000000605 extraction Methods 0.000 claims description 25
- 230000011218 segmentation Effects 0.000 claims description 22
- 238000004458 analytical method Methods 0.000 claims description 18
- 239000000284 extract Substances 0.000 claims description 11
- 230000008569 process Effects 0.000 claims description 6
- 239000000203 mixture Substances 0.000 claims description 5
- 230000001360 synchronised effect Effects 0.000 claims description 5
- 230000008859 change Effects 0.000 claims description 4
- 238000009472 formulation Methods 0.000 claims description 4
- 238000010276 construction Methods 0.000 claims description 3
- 230000008901 benefit Effects 0.000 abstract description 3
- 230000006870 function Effects 0.000 description 20
- 238000001514 detection method Methods 0.000 description 10
- 230000006399 behavior Effects 0.000 description 5
- 230000005540 biological transmission Effects 0.000 description 5
- 238000011835 investigation Methods 0.000 description 4
- 238000005516 engineering process Methods 0.000 description 3
- 238000007689 inspection Methods 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 2
- 230000018109 developmental process Effects 0.000 description 2
- 230000006872 improvement Effects 0.000 description 2
- 238000013549 information retrieval technique Methods 0.000 description 2
- 238000009434 installation Methods 0.000 description 2
- 238000013517 stratification Methods 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 108020005210 Integrons Proteins 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 239000012491 analyte Substances 0.000 description 1
- 230000002547 anomalous effect Effects 0.000 description 1
- 229910002056 binary alloy Inorganic materials 0.000 description 1
- 239000003086 colorant Substances 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 230000000295 complement effect Effects 0.000 description 1
- 238000011840 criminal investigation Methods 0.000 description 1
- 238000007405 data analysis Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 239000004744 fabric Substances 0.000 description 1
- 230000005021 gait Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000006855 networking Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 239000000126 substance Substances 0.000 description 1
- 239000013589 supplement Substances 0.000 description 1
- 230000009897 systematic effect Effects 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Abstract
The invention provides a rapid video retrieval method based on layered structuralized description of video information, which is characterized in that based on graded video content of three databases and three engines, a comprehensive application framework and a research and judging flow are subjected to layered structuralized description, wherein the comprehensive application framework is that in a video monitoring network system, a plurality of nodes are deployed by layers according to an actual condition of each sub-network; and each node takes a video layered structuralized description server as a core and comprises complete three databases and three engines. The rapid video retrieval method based on the layered structuralized description of the video information has the advantages that the cooperative work of the three databases and the three engines with many grades and many nodes is utilized to carry out full-automatic layered structuralized description on monitoring video content; and logical reasoning and judgment are carried out based on a topological relation between the nodes, so as to realize intelligent research and judgment of the video monitoring information and realize automatic alarming and rapid retrieval.
Description
Technical field
The present invention relates to a kind of fast video search method based on the description of video information hierarchical structured, belong to safety anti-
Model field of video monitoring.
Background technology
With economic and society development, and the continuous improvement to safety precaution demand for the people, video monitoring system exists
Obtain rapid development in the decades in past and popularized, by originally single security protection key industry such as government bodies, road
Traffic, financial institution etc., are slowly applied to increasing communal facility such as office building, market, residential area etc..
Social security monitoring creates substantial amounts of video information, but the video of the overwhelming majority is believed in these video informations
Breath all with solve a case unrelated, useful information is mingled in so that search cycle is spun out in a large amount of garbages, and delay is solved a case opportunity.Single
Suspect or ruck action, early warning suspect behavior as early as possible cannot be distinguished with manpower in the video monitoring information of magnanimity, thus,
Need special video information analysis and inquiry system, extract police concern information from these videos and carry out video segmentation
Structural description, is public security department's lock onto target, early warning in daily monitoring, or searched targets, solid during investigation
Determine evidence, discover a clue to provide video investigation service.
Realize such video information analysis and inquiry system typically requires and meets some primary conditions:First, system will
Lay multiple video camera composition monitoring networks;Secondly, system will automatically analyze to the content of monitor video, extracts useful letter
Cease and store, if necessary being capable of automatic alarm;3rd, system can provide and these useful informations be carried out with quick search and gos deep into
The function of analysis.
Prior art lays monitoring network by Ethernet or wireless network.In the prior art, for realizing automatization
Video presentation and analysis system, so as to extract from these videos police concern information and realize video monitoring quick-searching and from
Dynamic report to the police, can be processed it is also possible to number in photographic head terminal-pair storage the and real-time video monitoring information obtaining
Focused on according to being sent to Surveillance center.The technology used in video analysis includes target detection, target following, scene are divided
Analysis etc..Analyze the result obtaining and can be used for automatic alarm it is also possible to store for later inquiry.
For example, patent announcement CN101783881A discloses a kind of intelligent network with video structural description function
Network video camera, it is for Image Visual Feature such as the object color in video camera, texture, shape, motion, positioning and contour features
Its attribute and content are made inferences and judges, and pass through extensible markup language(XML), binary system extensible markup language
(Binary XML)Write description language.This invention mainly passes through a kind of video structural description device, for a video camera
In object carry out feature, the reasoning of attribute, judgement.
Disclose a kind of HD recording with video structural description function to set in patent announcement CN101790064A
Standby.A feature extraction unit is passed through in this invention, and multi-path monitoring video image information is carried out with structural description, produces with regard to it
The description information data of attribute and content, realizes intelligent monitoring and warning function.
Disclose a kind of network video server in patent announcement CN101827266A, it includes video acquisition, analysis
With coding output module, the attribute of available video features and content etc. are judged and reasoning, and realize the volume of small data quantity
Code, in order to network transmission and video quick-searching.This invention mainly passes through a video analysis describing module, in present networks
Multi-channel video signal carry out particular analysis, realize the structural description to monitor video.
But, in the above prior art, even achieve part automatically analyzing or auto-alarm function, also it is only limitted to
The data that individually each camera terminal of analysis collects forms structural description information for inquiry;Or in same LAN
The data of multiple camera terminals carries out comprehensive analysis, does not consider in multiple video monitoring sub-networks, massive video information state
Under video content quick search and cross-searching.In the face of police ultra-large networked video monitoring system or transprovincially cross-region
During the demand of cascade network video monitoring system, the powerless performance requirement meeting such ultra-large system of prior art.
In addition, according to the video monitoring function module of prior art or the webserver, main in the face of simple monitors environment
Or simple video monitors network, the analysis to video information and retrieval are all only limitted to single video source/several video source
In object make inferences, judge, analyze and store, all do not referred between multiple video monitoring sub-networks pass through sub-network
Information exchange between node, logical reasoning is carried out by sub-network node topology relation realize across a network intelligent video monitoring
Method, does not refer to the video content hierarchical structured description across multiple sub-networks and the distributed quick search side of video content yet
Method.When in the face of complicated multiple sub-network integrated system, above-mentioned prior art both otherwise affects the respective video of unit one belongs to
Monitoring alarm function, and meet centralized video monitoring intellectuality discriminant alarm function simultaneously, priority assignation can be very difficult, and
System effectiveness is not high, also can be limited in this requirement of a certain node can neatly extending or disconnecting.So it is above-mentioned special
Profit still can not meet the demand of police large-scale cross-region level video monitoring intelligent analysis system.
Content of the invention
In order to solve above-mentioned technical problem, meet police large-scale cross-region level video interconnected monitoring system good application
Demand, meets in massive video information quick-searching target, trans-regional is entered according to target characteristic different geographical is even trans-city
Row intelligently carries out network tracking, the demand of simultaneously automatic alarm linkage of intelligently deploying troops on garrison duty in key monitoring location to suspect/car,
The present invention provides a kind of fast video search method based on the description of video information hierarchical structured.
The a kind of of the present invention is based on classification three based on the fast video search method of video information hierarchical structured description
The video content hierarchical structured of individual data base and three engines describes integrated application framework and studies and judges flow process, described integrated application
Framework is:In video monitoring network system, according to each sub-network practical situation, multiple nodes are disposed in classification, each node with
Video segmentation structural description server(I.e. Metadata server)For core, and include complete three data base and three
Individual engine, described three data bases are video elementary resource database, video segmentation structural description data base(I.e. Metadata
Data base)And monitoring black list database;Described three engines are video segmentation structural description extraction engine(I.e. Metadata
Extraction engine), report to the police study and judge engine and Video content retrieval engine;Described flow process of studying and judging is:By multi-level, multinode three
Individual data base and three engine collaborative works, carry out full-automatic hierarchical structured description to monitor video content, and are based on node
Between topological relation carry out logical reasoning judgement, realize intellectuality and study and judge video monitoring information, finally realize automatic alarm and quick
Retrieval.
Each node of described video monitoring network system is all arranged Metadata server and three data bases and
Three engines are so as to possess the functions such as independent video monitoring, automatic alarm, user search.Described three data bases, both permissible
For centralized data base or distributed data base.
All comprise special video segmentation structural description server in each node of described video monitoring network system
(I.e. Metadata server), for node between interconnection, synchronous logic analysis result, realize simplifying and complete police concern letter
The subscription of breath Metadata data, push, storage, query function.
Fast video search method of the present invention is when being inquired about, mutual by means of the information between different terminals
Association, makes full use of the information such as video camera topological relation in video elementary resource database, using searching of logic-based reasoning
Suo Fangfa realizes quick search.
Fast video search method of the present invention, when carrying out hierarchical structured to video content and describing, need to be followed and regard
The standardized format of frequency content hierarchical structured description.
The comprising the concrete steps that of fast video search method of the present invention:Set up video elementary resource database, will be each
The construction of all kinds of warning in ground and monitoring system, renewal, change conditions and relevant rudimentary information gather warehouse-in in time;From difference
The monitor video of equipment carries out video extraction work by Metadata extraction engine, extracts video basic feature and police
Concern information;These stratification content description informations are saved in storage medium with normalized form, and unified foundation indexes;
Described video elementary resource is stored in video elementary resource database, and described layered representation content is stored in
In Metadata data base;Meanwhile, report to the police and study and judge engine by the video content layered representation information of above-mentioned generation and emphasis people
Data in member, emphasis vehicle, suspicious object image library is compared, or solves a case what experience was formulated with based on public security systems at different levels
Various suspect's behavior interpretation principles and alarm threshold are compared, and when finding suspicious object, automatically export warning message;With
During the retrieval of family, user passes through client input time, place, suspect's feature, suspect vehicle feature, suspicious object feature etc. and examines
Rope condition, Video content retrieval engine extraction feature first from querying condition, then in conjunction with video elementary resource information, pass through
The mode of distributed access carries out parallel search through Metadata server to the Metadata data base of child nodes at different levels, obtains
Obtain retrieval result and associated video is returned to user.
The present invention a kind of based on video information hierarchical structured description fast video searching system it is characterised in that:
Describe integrated application framework and study and judge flow process based on the video content hierarchical structured being classified three storehouse three engine, described integrated application
Framework is:In video monitoring network system, according to each sub-network practical situation, multiple nodes are disposed in classification, each node with
Video segmentation structural description server(I.e. Metadata server)For core, and include complete three data base and three
Individual engine, described three data bases are video elementary resource database, video segmentation structural description data base(I.e. Metadata
Data base)And monitoring black list database;Described three engines are video segmentation structural description extraction engine(I.e. Metadata
Extraction engine), report to the police study and judge engine and Video content retrieval engine;Described flow process of studying and judging is:By multi-level, multinode three
Individual data base and three engine collaborative works, carry out full-automatic hierarchical structured description to monitor video content, and are based on node
Between topological relation carry out logical reasoning judgement, realize intellectuality and study and judge video monitoring information, finally realize automatic alarm and quick
Retrieval;Described video monitoring network system each node all comprise special for node between interconnection, synchronous logic analysis
Result, video segmentation structural description server(I.e. Metadata server), to realize simplifying and complete police concern
The subscription of information Metadata data, push, storage, query function.
The present invention a kind of based on video information hierarchical structured description fast video searching system when being inquired about,
Interrelated by means of the information between different terminals, make full use of the video camera topological relation in video elementary resource database
Etc. information, the searching method using logic-based reasoning realizes quick search.
It is an advantage of the invention that combining public security one line demand, to the video content layering towards safety precaution and criminal investigation and case detection
Structural description content and form are designed and have been defined, from video basic feature and two levels of police concern feature to regarding
Frequency is described, and using described two levels, video signal is carried out with automatization's police concern information retrieval and study and judge, profit
Achieve the work(such as the extraction of police concern information, transmission, storage, retrieval with itself three data bases and three engine techniques
Can, thus realize transprovincially, the police on a large scale video automatically retrieval of cross-region and automatic alarm system, and permissible in theory
Expand endlessly, constantly access new terminal and new storage system, and due to accessing new terminal and new storage system
Continue to expand logical reasoning ability and automatic alarm scope, by multi-level, multinode three storehouse three engine collaborative work, to prison
Control video content carries out full-automatic hierarchical structured description, and carries out logical reasoning judgement based on topological relation between node, realizes
Video monitoring information and automatic alarm and quick-searching are studied and judged in intellectuality.
Brief description
Fig. 1 shows multi-level three data base of the present invention and the schematic network structure of three engines;
Fig. 2 shows the schematic network structure of three data bases in single node for the present invention and three engines.
Specific embodiment
Below in conjunction with the accompanying drawings to a kind of fast video inspection based on the description of video information hierarchical structured of the present invention
Suo Fangfa is described in detail;Three multi-level data bases of the present invention and three Engine-Networks, its principle such as Fig. 1 institute
Show.
Described comprehensive based on the video content hierarchical structured of three multi-level data bases and three engines in the present invention
Close in application architecture, described three multi-level data bases and three engines refer to, in the network comprising multiple nodes, deployment
There are three data bases with respect to its child node higher level and three engine nodes.By three in described higher level node
Individual data base and three engines, can dispatch the resource in its all three data base of child node and three engines.For example, every
Individual local police station can dispose three data bases of a three-level and three engine nodes, can integrate these child node resources in branch office and be formed
One two grades of three data base and three engine nodes, can form three data bases of an one-level by integron node resource in office of city
With three engine nodes.The child node of every one-level, had both included independent Metadata server, data base and engine for only
Vertical realize search function, the resource also by each child node of Metadata server calls and provide service to higher level.Each
The child node of rank has all control system that can work independently, and provides unified interface to the superior and the subordinate.
Each node of described video monitoring network system is all arranged Metadata server and three data bases and
Three engines are so as to possess the functions such as independent video monitoring, automatic alarm, user search;Described three data bases, both permissible
For centralized data base or distributed data base.
In order to meet performance requirement and the flexible expansion demand of extensive cross-region network type video monitoring system, the present invention
Comprise special video segmentation structural description server in each node(Metadata server 1), for node between interconnection,
Synchronous logic analysis result, realizes simplifying and the complete subscription of police concern information Metadata data, pushing, store, looking into
Ask function.
Preferably, in the method that the present invention provides, by adopting distributed data base, it is possible to use the net developing rapidly
Network disposal ability, the Metadata data of transmission more a small amount of is for retrieval, and positions correlation by way of logical reasoning
Monitor video so that retrieve a large amount of monitor video resource informations time shortened to a few houres by several days.It is additionally, since distributed
The intrinsic advantage of data base, it is not that design a certain in prior art such as is necessary for using a certain specific video monitoring
System, and can advantageously carry out integrated with different video monitoring system, or integrated with existing video monitoring system it is only necessary to
They supplement complement, modular Metadata server and three data bases and three engines.And due to dividing
Grid in cloth data base is substantially dynamic, the resource in grid can be dynamic occur it is allowed to user as needed
Resource is added in grid, or deletes from grid.After deleting in a certain node is from network type video monitoring system, this section
Point possesses Metadata server and three data bases and three engines, so still possessing independent video monitoring, automatically reporting
The functions such as police, user search, simply do not possess and study and judge functionally and logically inference function with associating of neighbor node, and delete this
Networked system after node still possesses whole associations and studies and judges functionally and logically inference function.In addition, distributed data base provides
Enhanced extensibility, is simplified and complete logic associated video information by transmission, makes system be not only restricted to be physically adjacent to
Or network delay, it is to avoid the defect of centralized data base, and in this way, can flexibly realize letter between node
Just interconnect, easily separate, easily delineation of power and quickly cross-region information transfer and video information inquiry.
Special Metadata server in single node of the present invention and three data bases and three, three engines
Data base and three engine schematic diagrams, its principle is as shown in Figure 2.
Fig. 2 also shows the police concern information retrieval integrated application based on fast video search method of the present invention
Framework.Described integrated application framework is that each node is all with video segmentation structural description server(Metadata server
1)For core, and include complete three data base and three engines, described three data bases are video elementary resource data
Storehouse 4, video segmentation structural description data base(Metadata data base 2)And monitoring black list database 3;Described three engines
It is video segmentation structural description extraction engine(Metadata extraction engine 5), engine 6 is studied and judged in warning and Video content retrieval draws
Hold up 7.
Fast video search method of the present invention is when being inquired about, mutual by means of the information between different terminals
Association, makes full use of the information such as video camera topological relation in described video elementary resource database 4, using logic-based reasoning
Searching method realize quick search.
Described video information hierarchical structured description refers to by video intelligent analytical technology, by the video counts of amorphousness
According to by the abstract of its content different levels and understanding, generating structure and the annotation of textual form, i.e. video segmentation structuring
Description data (i.e. metadata hereinafter described or Metadata data).These transmission needed for Metadata data and storage set
Standby resource is far smaller than the video data after compressed encoding, but contains main police concern information in video.To this
A little Metadata data carry out compiling rope it is possible to realize advanced video retrieval data to excavate application.The side providing in the present invention
In method, in order to realize the data interaction between each video monitoring sub-network and the application of across sub-network data complex reasoning, need first
Set up the standardized format of video content hierarchical structured description.
The searching method of described logic-based reasoning, refers to describe data by means of video content hierarchical structured, passes through
Time To Event, place, suspect's vehicles and translational speed are carried out with integrated logic reasoning, generates logical reasoning and pass judgment on knot
Really, i.e. logical data.These logical datas are carried out compile with rope, and the various suspicion with experience formulation of being solved a case based on public security systems at different levels
Criminal's behavior interpretation principle and alarm threshold are compared it is possible to realize intellectuality to study and judge video monitoring information, and finally realize
Automatic alarm and quick-searching.
Described video segmentation structural description server, i.e. Metadata server 1, is the core of each child node, uses
In between node interconnection, synchronous logic analysis result, realize simplify and the complete subscription of police concern information Metadata data,
Push, storage, query function.
Described Metadata data base 2, shoots video corresponding video segmentation knot for storing each video camera in child node
Structure describes data(I.e. described Metadata data).
Described monitoring black list database 3, stores the personnel's blacklist that need to deploy troops on garrison duty, face characteristic, apparel characteristic, gait spy
Levy, contour of the vehicle feature, vehicle color feature, vehicle license plate characteristic, Mass disturbance, that personnel hover, vehicle drives in the wrong direction etc. is various police
Concern event description etc..
Described video elementary resource database 4, for the unit type of each video camera, installation ground in storage monitoring child node
The Back ground Informations such as point, working condition and video camera topological relation figure.Described video camera topological relation figure, describes each video camera
Close on video camera and suspect/car needed for a video camera moves to another video camera coverage must be through route
And the shortest time.This shortest time is determined by camera installation locations, setting angle, mode of transportation, road conditions etc..
Described Metadata extraction engine 5, its function is that multi-path monitoring video is carried out with real-time data analysis, extracts bid
Accurate video segmentation structural description information, and be saved in described Metadata data base 2.Described Metadata extraction engine
It is the general name of all equipment that can extract hierarchical structured description from video or module:Can be an equipment or
Multiple equipment;Can be leading portion embedded device or central server even PC cluster resource;Can be independent
Equipment is it is also possible to be embedded into video camera, in DVR;Accessible analog video, also can access the digital video through overcompression.Described
Metadata extraction engine both can be extracted to the video flowing shooting in real time, also the video file of storage can be taken out
Take.
Engine 6 is studied and judged in described warning, by being analyzed in real time to Metadata data flow, for described monitoring blacklist
Information in data base 3 is deployed troops on garrison duty, when detect meet black list database and suspect/car or event and occur after, immediately
Send warning.
Described Video content retrieval engine 7, the intelligent video retrieval according to user or tracking request, generate Metadata inspection
Rope computer instruction, enters line retrieval to described Metadata data base 2, returns to user and comprises the target of search condition of coincideing
Video segment.
Described camera network refers to adjacent camera terminal the multichannel video camera of logic association and information mutual communication.
In the method that the present invention provides, for realizing the Video Information Retrieval Techniquess of logic-based reasoning between node, firstly, it is necessary to
By the video camera model in monitoring network, infield, operating status record is to video elementary resource database 4, and, passes through
To live on-the-spot investigation and experiment, record each video camera closes on video camera, and suspect/car moves from a video camera
Move must ultimately form video camera topological relation figure and remembering through route and shortest time needed for another video camera coverage
Record video elementary resource database 4.
Secondly, dispose various video intelligent analytical equipments, form Metadata extraction engine 5.Described Metadata extracts
Engine 5 automatically can extract target routinely from the video content that affiliated node camera/video collecting device 8 shoots
And its feature(As the shape of target, color, texture, profile etc.), and police concern information(As suspect, suspect vehicle, suspicious
Article, anomalous event etc.), through index storage in Metadata data base 2.For being stored in video storaging equipment 9, not
It is analyzed in real-time and extracts the video of Metadata data, provisional injunction Metadata extraction engine 5 can enter when needing retrieval
Row video record extracts, and its result is also stored in Metadata data base 2.
When user carries out intelligence and deploys to ensure effective monitoring and control of illegal activities, the blacklist being inputted by software interface according to user and strategy of deploying troops on garrison duty, and
Monitoring blacklist and strategy of deploying troops on garrison duty that superior node is assigned, form monitoring black list database 3 and dynamically update.Meanwhile, in institute
State warning and study and judge engine 6, by being analyzed in real time to Metadata data flow, in described monitoring black list database 3
Information deployed troops on garrison duty, or with based on public security systems at different levels solve a case experience formulation various suspect's behavior interpretation principles and warning circle
Limit is compared, when the suspect/car meeting blacklist and the suspicious suspicion of trace surmounting police's preset alarm boundary is detected
After doubtful people/car or accident, then send warning immediately.When there is downstream site, this node report to the police study and judge engine also can be by
Monitoring blacklist and policy assignments of deploying troops on garrison duty are deployed troops on garrison duty to corresponding downstream site.
When user carries out intelligent video retrieval, retrieval time is determined in described Video content retrieval engine 7 middle finger by user
Scope, regional extent, target characteristic, rule of conduct, event type etc. limit, and are generated by described Video content retrieval engine 7
Metadata retrieves computer instruction, enters line retrieval to described Metadata data base 2, returns to user and comprises the retrieval bar that coincide
The video segment of the target of part.When there is downstream site, retrieval tasks also can be assigned by this node Video content retrieval engine 7
Downstream site to respective regions.
When user carries out intelligent networking tracking to suspect/car, mutually closed by means of the information between different cameras
Connection, in conjunction with camera network node topology graph of a relation, using the searching method of logic-based reasoning, realizes suspect/car is gone
To complex reasoning, draw the video related information coincideing the most, or show that most probable finds suspect/car temporally again
Point, helps the police automatically to obtain associated video.
By the geographically relevant information and the system node closing between on each node, and police's on-the-spot investigation exploration
Draw substantial amounts of distance, the relation information of required time being gone to by this node and closing on system node, can be gone out by logical reasoning
A certain suspect vehicle or a certain suspect are gone to by certain mode of transportation and close on system node required time, and conclude therefrom that
May need to transfer the priority of neighbor node video information.The remote system that suspect vehicle or a certain suspect can not possibly reach
Node, the priority of its Video Information Retrieval Techniques is reduced to zero, then avoid the need for doing retrieving further or transferring regarding on network again
Frequency information, saves police's retrieval therefrom and transfers time of video, ensure realize large-scale networked video monitoring system and from
Dynamic warning system is unlikely to be restricted by systematic function or network performance.
Preferably, in order to meet, extensive cross-region network type video monitoring system interconnects and adjacent node is comprehensively patrolled
Collect the requirement of reasoning, in the Metadata data base 2 in each node, the Metadata data of storage all need to be according to identical lattice
Formula, identical indexed mode are stored, and ensure that the video elementary data in adjacent node can be by the retrieval of same logic
Instruction retrieval is out.
Preferably, Metadata data of the present invention, is made up of following two parts data:
1. video basic feature
Video basic feature is the low level information for describing target in video.Including following information:
1)Moving object detection information:Including time, target designation, target location, target sizes, speed;
2) motion target tracking information (track);
3) color of object rectangular histogram:Color histogram describes the distribution of different colours on object.Domain color feature description
Constitute the primary color of target.As a green car, its primary color is exactly green;
4) discrete cosine transform (DCT) coefficient:Target image carries out the coefficient of frequency domain transform gained, and this feature effectively describes
The texture features of target;
5) edge direction characteristic:The vector carrying out obtained by statistics with histogram according to direction to object edge, for effective
Distinguish variform target.
2. police concern information
1)Target classification information;
2)Face, Car license recognition information;
3)Leave analyte detection, article remove detection, detection of running, intrusion detection, drive in the wrong direction detection, hover detection etc. event inspection
Survey class output information, specifying information includes:Current time, Case Number, event type, event level, state-event;
4)People flow rate statistical output information includes:Current time, tangent plane pass through number;
5)Crowd density estimation output information includes:Current time, current crowd density grade;
6)Video quality detection information.
Above-mentioned two parts information collectively forms Metadata data, is immediately extracted by Metadata extraction engine,
Store in Metadata data base.
During video information layered representation of the present invention it is preferable that monitor video source can be as Fig. 2 in institute
The video capture device 8 that shows is it is also possible to the video file that is stored in video storaging equipment 9.
In sum, according to embodiment of the present invention, its idiographic flow is:Set up video elementary resource database
4, the construction of all kinds of for various places warnings and monitoring system, renewal, change conditions and relevant rudimentary information are gathered warehouse-in in time;Come
Pass through Metadata extraction engine 5 from the monitor video of distinct device, extract video basic feature and police concern information;This
A little stratification content description informations are saved in storage medium with normalized form, and unified foundation indexes;Described video base
Plinth resource is stored in video elementary resource database 4, and described layered representation content is stored in Metadata data base 2
In;Meanwhile, report to the police study and judge engine 6 by the video content layered representation information of above-mentioned generation and emphasis personnel, emphasis vehicle, can
Data in doubtful images of items storehouse is compared, or with based on public security systems at different levels solve a case experience formulation various suspect's behaviors sentence
Read principle and alarm threshold is compared, when finding suspicious object, automatically export warning message;In user search, user
By search conditions such as client input time, place, suspect's feature, suspect vehicle feature, suspicious object features, in video
Hold search engine 7 extraction feature first from querying condition, then in conjunction with video elementary resource information, by distributed access
Mode carries out parallel search through Metadata server 1 to the Metadata data base 2 of child nodes at different levels, obtains retrieval result
And associated video is returned to user.
The above, the only specific embodiment of the present invention, but protection scope of the present invention is not limited thereto, not
In the case of departing from spirit and substance of the present invention, increase more engines or more data base on the basis of the present invention, by
This and realize more complete fast video retrieval, these modifications and improvement are also considered as protection scope of the present invention, and any ripe
Know those skilled in the art in scope disclosed by the invention, the change or replacement that can readily occur in, all should cover
Within the scope of the invention as claimed.
Claims (4)
1. a kind of based on video information hierarchical structured description fast video search method it is characterised in that:Based on classification three
The video content hierarchical structured of individual data base and three engines describes integrated application framework and studies and judges flow process, described integrated application
Framework is:In video monitoring network system, according to each sub-network practical situation, multiple nodes are disposed in classification, each node with
Video segmentation structural description server is core, and includes complete three data base and three engines, and described video divides
It is Metadata server that Rotating fields describe server, and described three data bases are video elementary resource database, video divides
Rotating fields descriptive data base and monitoring black list database, described video segmentation structural description data base is Metadata number
According to storehouse;Described three engines are video segmentation structural description extraction engine, report to the police and study and judge engine and Video content retrieval engine,
Described video segmentation structural description extraction engine is Metadata extraction engine;Described flow process of studying and judging is:By multi-level, many
Three data bases of node and three engine collaborative works, carry out full-automatic hierarchical structured description to monitor video content, and
Between based on node, topological relation carries out logical reasoning judgement, realizes intellectuality and studies and judges video monitoring information, and final realization is reported automatically
Police and quick-searching;Metadata server and three data are all arranged on each node of described video monitoring network system
, so as to possess independent video monitoring, automatic alarm, user search function, described three data bases are collection for storehouse and three engines
One of Chinese style data base or distributed data base;The comprising the concrete steps that of described fast video search method:Set up video elementary money
Source database, the construction of all kinds of for various places warnings and monitoring system, renewal, change conditions and relevant rudimentary information are gathered in time
Warehouse-in;From distinct device monitor video by Metadata extraction engine to carry out video extraction work, extract video
Basic feature and police concern information;These hierarchical structureds describe video information and are saved in storage medium with normalized form
In, and unified foundation index;Described video elementary resource is stored in video elementary resource database, described hierarchical structured
Description video information is stored in Metadata data base;Meanwhile, report to the police and study and judge engine by the hierarchical structured of above-mentioned generation
Description video information compare with the data in emphasis personnel, emphasis vehicle, suspicious object image library, or with based on public affairs at different levels
Peace system solve a case experience formulation various suspect's behavior interpretation principles and alarm threshold compare, when find suspicious object when,
Automatically output warning message;In user search, user passes through client input time, place, suspect's feature, suspect vehicle
Feature, suspicious object characteristic key condition, Video content retrieval engine extraction feature first from querying condition, then in conjunction with regarding
Frequency basic resource information, by way of distributed access through Metadata server the Metadata number to child nodes at different levels
Carry out parallel search according to storehouse, obtain retrieval result and associated video is returned to user.
2. a kind of fast video search method based on the description of video information hierarchical structured according to claim 1, its
It is characterised by:All comprise special video segmentation structural description service in each node of described video monitoring network system
Device, i.e. Metadata server, for node between interconnection, synchronous logic analysis result, realize simplify and complete police concern
The subscription of information Metadata data, push, storage, query function.
3. a kind of fast video search method based on the description of video information hierarchical structured according to claim 1, its
It is characterised by:Described fast video search method is when being inquired about, interrelated by means of the information between different terminals, fills
Divide using the video camera topological relation information in video elementary resource database, the searching method using logic-based reasoning is realized
Quick search.
4. a kind of fast video search method based on the description of video information hierarchical structured according to claim 1, its
It is characterised by:Described fast video search method, when carrying out hierarchical structured to video content and describing, need to follow video content
The standardized format of hierarchical structured description.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210230539.7A CN102929903B (en) | 2012-07-04 | 2012-07-04 | Rapid video retrieval method based on layered structuralized description of video information |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201210230539.7A CN102929903B (en) | 2012-07-04 | 2012-07-04 | Rapid video retrieval method based on layered structuralized description of video information |
Publications (2)
Publication Number | Publication Date |
---|---|
CN102929903A CN102929903A (en) | 2013-02-13 |
CN102929903B true CN102929903B (en) | 2017-02-08 |
Family
ID=47644701
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201210230539.7A Active CN102929903B (en) | 2012-07-04 | 2012-07-04 | Rapid video retrieval method based on layered structuralized description of video information |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN102929903B (en) |
Families Citing this family (23)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN104052780A (en) * | 2013-03-15 | 2014-09-17 | 南京理工大学常熟研究院有限公司 | Social public security video information retrieval and service system based on convergence of three networks |
CN103347167B (en) * | 2013-06-20 | 2018-04-17 | 上海交通大学 | A kind of monitor video content based on segmentation describes method |
CN104281598A (en) * | 2013-07-05 | 2015-01-14 | 上海众恒信息产业股份有限公司 | Population database system based on cloud computing |
CN105138525B (en) * | 2014-05-30 | 2019-10-01 | 株式会社日立制作所 | Traffic video processing unit and method and retrieval device and method |
TWI533239B (en) * | 2014-12-12 | 2016-05-11 | 由田新技股份有限公司 | Distant facial monitoring system, method, computer readable medium, and computer program products |
CN105959621A (en) * | 2016-04-28 | 2016-09-21 | 泰华智慧产业集团股份有限公司 | Quasi-real time control distribution system and method based on multi-source video structural data |
CN106339428B (en) * | 2016-08-16 | 2019-08-23 | 东方网力科技股份有限公司 | Suspect's personal identification method and device based on video big data |
CN106330928B (en) * | 2016-08-29 | 2019-12-31 | 上海航盛实业有限公司 | Management system for monitoring information |
CN106339973A (en) * | 2016-08-31 | 2017-01-18 | 安徽四创电子股份有限公司 | Guard security system based on data platform and and guard security method thereof |
CN106709424B (en) * | 2016-11-19 | 2022-11-11 | 广东中科人人智能科技有限公司 | Optimized monitoring video storage system |
CN108153799B (en) * | 2016-12-06 | 2022-03-25 | 阿里云计算有限公司 | Database access control method and device and database system |
CN107277470A (en) * | 2017-08-10 | 2017-10-20 | 四川天翼网络服务有限公司 | A kind of network-linked management method and digitlization police service linkage management method |
CN107832475A (en) * | 2017-12-01 | 2018-03-23 | 浪潮软件集团有限公司 | Deployment control method, deployment control engine and deployment control system |
CN108364438A (en) * | 2017-12-28 | 2018-08-03 | 武汉烽火众智数字技术有限责任公司 | Method for early warning based on front end perception data and device |
CN108399261A (en) * | 2018-03-13 | 2018-08-14 | 成都蓉兴伟业科技有限公司 | A kind of case and video analysis method based on integration method |
CN110519204B (en) * | 2018-05-21 | 2020-07-24 | 北京仁光科技有限公司 | Apparatus, method and system for interacting computers within a subnet |
CN109523797A (en) * | 2018-12-12 | 2019-03-26 | 武汉中科天宇科技有限公司 | A kind of wisdom traffic police service linkage system and corresponding method |
CN110933291B (en) * | 2019-06-06 | 2020-07-28 | 北京仁光科技有限公司 | Cross-network interaction system and cross-network interaction method |
CN110825893A (en) * | 2019-09-18 | 2020-02-21 | 北京旷视科技有限公司 | Target searching method, device, system and storage medium |
CN110602461A (en) * | 2019-09-26 | 2019-12-20 | 广脉科技股份有限公司 | Future community regional perception information processing feedback system |
CN113448622B (en) * | 2021-01-20 | 2022-05-06 | 北京仁光科技有限公司 | Multi-signal-source interface interaction system |
CN113569096B (en) * | 2021-07-23 | 2024-03-29 | 北京百度网讯科技有限公司 | Structured information extraction method, device, equipment and storage medium |
CN115080094B (en) * | 2022-08-16 | 2022-11-29 | 张家港金典软件有限公司 | Security system updating method and system based on hierarchical interaction |
Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6463433B1 (en) * | 1998-07-24 | 2002-10-08 | Jarg Corporation | Distributed computer database system and method for performing object search |
CN1691591A (en) * | 2004-04-16 | 2005-11-02 | Ut斯达康公司 | Method and apparatus for a loosely coupled, scalable distributed multimedia streaming system |
CN101917553A (en) * | 2009-11-27 | 2010-12-15 | 新奥特(北京)视频技术有限公司 | System for collectively processing multimedia data |
-
2012
- 2012-07-04 CN CN201210230539.7A patent/CN102929903B/en active Active
Patent Citations (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
US6463433B1 (en) * | 1998-07-24 | 2002-10-08 | Jarg Corporation | Distributed computer database system and method for performing object search |
CN1691591A (en) * | 2004-04-16 | 2005-11-02 | Ut斯达康公司 | Method and apparatus for a loosely coupled, scalable distributed multimedia streaming system |
CN101917553A (en) * | 2009-11-27 | 2010-12-15 | 新奥特(北京)视频技术有限公司 | System for collectively processing multimedia data |
Also Published As
Publication number | Publication date |
---|---|
CN102929903A (en) | 2013-02-13 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN102929903B (en) | Rapid video retrieval method based on layered structuralized description of video information | |
Laufs et al. | Security and the smart city: A systematic review | |
CN107196910B (en) | Threat early warning monitoring system, method and deployment framework based on big data analysis | |
Vieira et al. | On-line discovery of flock patterns in spatio-temporal data | |
CN101873414B (en) | Event video detection system based on hierarchical structure | |
TWI450207B (en) | Method, system, computer program product and computer-readable recording medium for object tracking | |
CN106649801A (en) | Time-space relationship analysis system based on multi-source internet-of-things position awareness | |
CN104616205A (en) | Distributed log analysis based operation state monitoring method of power system | |
WO2014121340A1 (en) | A surveillance system | |
Al Boni et al. | Area-specific crime prediction models | |
CN105335387A (en) | Retrieval method for video cloud storage system | |
CN113347170B (en) | Intelligent analysis platform design method based on big data framework | |
CN104317918A (en) | Composite big-data GIS (geographic information system) based abnormal behavior analysis and alarm system | |
CN106649809A (en) | Internet of Things analysis method | |
CN112766119A (en) | Method for accurately identifying strangers and constructing community security based on multi-dimensional face analysis | |
CN106646110A (en) | Low-voltage distribution network fault positioning system based on GIS and Petri technologies | |
CN106777279A (en) | A kind of time-space relationship analysis system | |
CN115664703A (en) | Attack tracing method based on multi-dimensional information | |
CN102651813A (en) | Intelligent video monitoring and advertising system | |
Choi et al. | Activity recognition in wide aerial video surveillance using entity relationship models | |
CN114706988A (en) | Monitoring and early warning method and device based on knowledge graph | |
CN105208326A (en) | Video-cloud-based early warning method and system of public security threat of urban region | |
Hampapur | Smart video surveillance for proactive security [in the spotlight] | |
Kwee et al. | Traffic-cascade: Mining and visualizing lifecycles of traffic congestion events using public bus trajectories | |
Beiji et al. | Crime hotspot detection and monitoring using video based event modeling and mapping techniques |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
C14 | Grant of patent or utility model | ||
GR01 | Patent grant |