CN101043629B - Motion picture encoding method and devices thereof - Google Patents
Motion picture encoding method and devices thereof Download PDFInfo
- Publication number
- CN101043629B CN101043629B CN 200710103419 CN200710103419A CN101043629B CN 101043629 B CN101043629 B CN 101043629B CN 200710103419 CN200710103419 CN 200710103419 CN 200710103419 A CN200710103419 A CN 200710103419A CN 101043629 B CN101043629 B CN 101043629B
- Authority
- CN
- China
- Prior art keywords
- field
- motion vector
- vector
- pixel
- mvcy
- 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.)
- Expired - Lifetime
Links
Images
Abstract
The present invention provides a motion picture coding and decoding device. When a prediction is made between fields with different parity, the predicative efficiency of a chrominance vector is improved by adaptively switching the generation of a chrominance motion vector depending on a encoding/decoding field parity (top/bottom) and a reference field parity (top/bottom), and the coding efficiency is improved accordingly.
Description
The application be that September 5, application number in 2003 are 03156610.3 the applying date, denomination of invention divides an application for the Chinese invention patent application of " motion picture encoding apparatus and moving picture decoding equipment ".
Technical field
The present invention relates to have the motion picture encoding apparatus and the moving picture decoding equipment of predictive mode between the field.
Background technology
Usually, the motion image data volume is very big.Therefore, motion image data is being transferred to receiving equipment from transmitting apparatus, or when being stored in it in memory device, is carrying out the high efficiency coding moving image.At this, " high efficiency coding " is one the particular data string converted to another serial data, and compress the cataloged procedure of its data volume.
Two type of motion view data are arranged: a kind of mainly is to be made of frame, and another kind mainly is to be made of the field.Below the main prior art of describing the compression field picture.
As the high-efficiency encoding method of motion image data, frame/field prediction coding is known.
Fig. 1 is the structured flowchart of frame/field prediction encoding device.
This coding method has utilized a plurality of sections of motion image data very high this phenomenon of relevance each other on time orientation.Below the rough operation of describing as shown in Figure 1.The difference image that subtracter 39 generates between input original image and the predicted picture, orthogonal transform unit 31,40 pairs of difference images of quantifying unit 32 and coefficient entropy coding unit are encoded.Inverse quantization unit 33 and inverse orthogonal transformation unit 34 are reproduced from the difference image of quantifying unit 32 outputs.The predicted picture that uses when then, decoded picture generation unit 35 utilizes reproduction difference image that decoded picture generation unit 35 reproduced and coding is decoded to coded image.Decoded picture memory cell 36 storing reproduction images.Subsequently, motion vector computation unit 37 calculates the motion vector between reproduced images and the follow-up input picture, predicted picture generation unit 38 use motion vector generation forecast images.The motion vector that is generated is encoded by vector entropy coding unit 41, and exports through MUX 42 with the code coefficient data of coefficient entropy coding unit 40 codings.In other words, because in motion image data, exist very high similitude usually between frame/field data of particular moment and frame/field data subsequently, the predictive coding method has been utilized this characteristic between frame/field.For example, in the data transmission system that adopts predictive coding method between frame/field, transmitting apparatus produces the motion vector data of the skew between expression previous frame/field and target frame/field, and utilize its motion vector data, and give receiving equipment motion vector data and differential data transmission by predicted picture in the target frame/field of previous frame/field picture generation and the differential data between the true picture in target frame/field.Receiving equipment is according to the image in the motion vector data that receives and the differential data reproduction target frame/field.
The summary of frame/field prediction coding so far, has been described with reference to figure 1.Next descriptor frame predictive coding and field prediction are encoded.
Fig. 2 and 3 has shown at ISO/IEC MPEG-2/MPEG-4 (following be called respectively " MPEG-2 " and " MPEG-4 ") and ITU-TH.264/ISO/IEC MPEG-4 Part10 FCD (advanced video coding (AVC) (" JointFinal Committee Draft (JFCD) of JointVideo Specification (ITU-T REC; 14496-10 AVC H.264|ISO/IEC "), JVT-D157, or ISO/IEC JTC1/S029/WG11 MPEG02/N492, in July, 2002, Klagenfurt, AT) the common field picture coded format of using in (hereinafter referred to as " AVC FCD "), wherein ITU-T and ISO/IEC are in August, 2002 common standardization.Particularly, each frame is made of two fields: top (top field) He Dichang (bottom field).What Fig. 2 showed is luminance pixel and chroma pixel position separately and the affiliated field of each pixel.As shown in Figure 2, odd number brightness line belongs to the field, top as the first brightness line 50a, the 3rd brightness line 50b, the 5th brightness line 50c, the 7th brightness line 50d etc.; Even number brightness line belongs to field, the end as the second brightness line 51a, the 4th brightness line 51b, the 6th brightness line 51c, the 8th brightness line 51d etc.Equally, the odd number chroma line belongs to the field, top as the first chroma line 52a, the 3rd chroma line 52b etc., and the even number chroma line belongs to field, the end as the second chroma line 53a, the 4th chroma line etc.
Different images are constantly represented with field, the end in each field, top.Next time/the spatial configuration of field, top and field, the end is described with reference to figure 3.
In Fig. 3 and later each figure, technology of the present invention relates to the vertical component of motion vector.Therefore, in this manual, reveal competence pixel component not, for simplicity, all horizontal components of motion vector are assumed to be 0.But,, accurately shown brightness and the relation of the position between the colourity in each in order to show prior art problems and effect of the present invention.
In Fig. 3, the longitudinal axis and transverse axis are represented the location of pixels and the time of each interior vertical component respectively.Among Fig. 3 because in the field of the horizontal component of each image not the location change, so do not show and do not describe its horizontal pixel component yet.
As shown in Figure 3, the location of pixels of chromatic component depart from luminance component the field in location of pixels 1/4th vertical pixels.This is because when frame is made of field, top and two of fields, the end, has formed location of pixels relation as shown in Figure 2.If based on the NTSC form, the time interval between then adjacent top and field, the end (64a:65a, 65a:64b, etc.) is approximately 1/60 second.The time interval between between two continuous fields, top (64a:64b, etc.) or two continuous fields, the end (65a:65b, etc.) is approximately 1/30 second.
Next the frame predictive coding pattern and the field prediction thereof of the field picture that adopts among MPEG-2 and the AVC FCD are described.
Fig. 4 has shown the method for utilizing two continuous field (adjacent field, top and field, the end) structure frames in frame prediction mode.
As shown in Figure 4, frame is to be made of two fields continuous in time (field, top and field, the end).
Fig. 5 has shown frame prediction mode.
In Fig. 5, supposed as shown in Figure 4 by two continuous fields (top and field, the end) reconstruct each frame, as 84a, 84b, 84c etc.In this frame prediction mode, to encoding of will encoding by top and the frame that constitutes of field, the end.As the reference image, two continuous (field, top and a field, the end) being used as reference by storage constitutes a reference frame, is used for predicting the target frame that will encode.Then, according to flow process shown in Figure 1 this two two field picture is encoded.In the motion vector method for expressing of this frame predictive coding pattern, zero vector, i.e. (0,0) expression is positioned at the pixel of the same space position.Particularly, the location of pixels 81 of motion vector (0,0) expression frame 1 84a that belongs to the luminance pixel 82 of frame 2 84b.
Next the field prediction coding mode is described.
Fig. 6 has shown the Forecasting Methodology of predictive mode between the field.
In the predictive mode on the scene, the coding target is field at the bottom of import as original image one top (94a, 94b, etc.) or (95a, 95b, etc.).Cun Chu field, top or field, the end can be used as reference picture in the past.In this case, it has been generally acknowledged that the parity of original graph image field and the identical original graph image field and the reference diagram image field of promptly meaning of parity of reference field all are field, top or field, the end.For example, in the prediction 90 between the field with same parity shown in Figure 6, original graph image field 94b and reference field 94a are the fields, top.Equally, it has been generally acknowledged that the parity of the parity of original graph image field and reference field different mean in original image and the reference field one for top another is field, the end.For example, in the prediction 91 between not like parity field shown in Figure 6, the original graph image field is field, end 95a and the reference diagram image field is a top 94a.Then, according to flow process shown in Figure 1 these original images and reference field are encoded.
In the frame pattern and field mode of prior art, come calculating kinematical vector according to the location of pixels in each frame/field.The motion vector calculating method and the pixel corresponding method of giving the timing prior art when motion vector described here.
Defined the coordinate system of the frame/field picture of extensive use in MPEG-2 coding, MPEG-1 coding, the AVC FCD coding etc. among Fig. 7.White circle among Fig. 7 is the pixel definition position in target frame/field.In the coordinate system of this frame/field picture, initial point (0,0) is appointed as in the upper left corner, and sequentially to level and vertical pixel definition position designated value 1,2,3 etc.Particularly, be positioned at n horizontal level and m upright position pixel coordinate for (n, m).Equally also stipulated the coordinate of the position between the pixel.Particularly, be positioned in the horizontal direction apart from pixel 1.5 pixels in the upper left corner with the position 180 of black circle mark among Fig. 7, on 2 locations of pixels, the coordinate of position 180 can be expressed as (1.5,2) in vertical direction.In field picture, has only a half-pix of two field picture in vertical direction.But, even in this case, according to the location of pixels in each, with the coordinate of the same manner definition pixel shown in Figure 7.
Next use coordinate system shown in Figure 7 to describe the definition of the motion vector between the field.
Fig. 8 has shown the Traditional calculating methods of the motion vector between the respective pixel between the field.Motion vector need to determine the position of coded fields and the position of reference field.Between these 2, determine motion vector.So, but calculation code field coordinate 201 (X
s, Y
s) and reference field coordinate 202 (X
d, Y
d) between motion vector.In the conventional method of the motion vector between the pixel of calculating between two fields,, all use same procedure described below to come calculating kinematical vector no matter coded fields or reference field are field, top or field, the end.Particularly, coded fields coordinate 201 (X
s, Y
s) and reference field coordinate 202 (X
d, Y
d) input to motion vector computation unit 200, and provide (X
d-X
s, Y
d-Y
s) as the motion vector 203 between these 2.
Fig. 9 has shown the conventional method of the pixel that motion vector between calculated field is pointed.In this case, suppose to come calculating kinematical vector with method shown in Figure 8.Reference frame/field Coordinate Calculation needs the position and the motion vector of coded frame/field.Under situation shown in Figure 9, be assumed to be coded fields coordinate 212 (X
s, Y
s) (X, Y), (X is Y) with coded fields coordinate 212 (X then can to use motion vector 212 to have provided motion vector 211
s, Y
s) calculate the reference field coordinate.In the Traditional calculating methods of the motion vector between the field between the respective pixel, calculating the reference field position with following identical method, and no matter coded fields or reference field are top or field, the end.Particularly, (X is Y) with coded fields coordinate 212 (X motion vector 211
s, Y
s) input to pixel corresponding unit 210, provide coordinate (X
s+ X, Y
s+ Y) as a reference coordinate 213.
What concern between this vector and location of pixels determines to be equally applicable to luminance component and chromatic component.In being both the MPEG-1/MPEG-2/AVC FCD of ordinary movement method for encoding images, only the vector of luminance component to be encoded, the vector of chromatic component calculates by dwindling luminance component.Particularly, in AVC FCD, because the Vertical number of pixels of chromatic component and Horizontal number of pixels are respectively luminance component half,, regulation obtains to be used for to calculate the motion vector of the predict pixel of chromatic component so should accurately narrowing down to half to the motion vector of luminance component.
Figure 10 has shown the conventional method of calculating chroma motion vector with brightness movement vector.
Particularly, if brightness movement vector 221 and chroma motion vector 222 are respectively that (MV_x is MV_y) with (MVC_x, MVC_y), then chroma motion vector generation unit 220 can calculate chroma motion vector 222 according to following formula.
(MVC_x，MVC_y)＝(MV_x/2，MV_y/2) (1)
No matter motion vector is used to have same parity or has prediction between the not like parity field, can use this Traditional calculating methods.
In AVC FCD,, can use the precision of 1/4 pixel as the precision of the motion vector of luminance component.Therefore,,, can use vector, promptly be accurate to the decimal fraction precision with 1/8 pixel precision as the precision of the motion vector of chromatic component as the result of formula (1).
Figure 11 has shown the computational methods of the interpolating pixel of the chromatic component that defines among the AVC FCD.
In Figure 11, Hei Quan and white circle are represented integer pixel and interpolating pixel respectively.At this, the horizontal coordinate of interpolating pixel G 256 is by with α: the ratio internal point of division A 250 of 1-α and each horizontal coordinate between the C 252 obtain, and vertical coordinate can be by with β: the ratio internal point of division A 250 of 1-β and each vertical coordinate between the B 251 obtain.At this, α and β are the values between 0 and 1.Can be as follows by the interpolating pixel G 256 of these location definitions with being positioned at integer pixel A 250, B 251, C 252 and D 253 around the interpolating pixel G 256 and α and β rough calculation.
G＝(1-α)·(1-β)·A+(1-α)·β·B+α·(1-β)·C+α·β·D (2)
Chromatic component interpolating pixel computational methods according to method shown in Figure 11 are an example, also can use other computational methods.
Under the situation of this coding mode, different at the original graph image field with reference field, promptly in the prediction that has between the not like parity field, its zero vector separately of the motion vector of luminance component and the motion vector of chromatic component is uneven in the definition of AVC FCD.Particularly, predict by the chromatic component motion vector of luminance component motion vector computation, then will depart from the pixel of luminance component with reference to the locus if use according to traditional definition.This situation is described below with reference to Figure 12.In Figure 12, suppose that field, top 130, field, the end 131 and field 132, top are continuous in time.At this, utilize top 130 to be encoded in field, the end 131.In this interfield coding, vertical motion vector of same line is set to zero in each.Therefore, if zero vector (0,0) is composed to the second-line luminance pixel 133a that belongs to field, the end 131, this pixel can be predicted by the pixel 135a in the field, top 130.Same, when zero vector (0,0) being given the first-line chroma pixel 133a that belongs to field, the end 131, this pixel can be predicted by the pixel 137a in first line of the chromatic component of field, top 130.Equally, belong to top 132 the in three-way luminance pixel 133b and chroma pixel 134b can be respectively by brightness in the field, the end the in three-way pixel 135b and the pixel 137b in second line of colourity predict.Because preferably chroma motion vector is arranged essentially parallel to brightness movement vector, if brightness movement vector is kept intact, then chroma pixel 134a and 134b should be predicted by position 136a and 136b respectively.
As previously mentioned, explained in not parallel this fact of zero vector that has between not like parity field brightness and colourity in the prediction.Under the situation of AVC FCD, this fact is brought following point for the prediction between the different field of all parities.Figure 13 and 14 has shown these problems.Problem in the AVC FCD situation is described below.In the explanation below, for the purpose of simple, in all cases, the horizontal component of motion vector is set at zero.
Figure 13 has shown when reference field and coded fields are field, the end and top respectively, has used brightness movement vector to calculate the caused problem of chroma motion vector.In AVC FCD, from equation (1) as seen, be half of luminance component because stipulated the vertical and horizontal pixel number of chromatic component, so be used for calculating half that the colorimetric prediction pixel motion vector should narrow down to the luminance component motion vector.This and motion vector are to be used for interframe, to have between the field of same parity, and the prediction that still has between not like parity field is irrelevant.
Explained later the problem that when using brightness movement vector between the different field of parity to calculate chroma motion vector, causes.In Figure 13, the coded fields top field brightness pixel 140 in first line has predictive vector (0,1), so it specifies the reference field luminance pixel position, field, the end 141 in second line to be predicted value.
At this,, can calculate the chroma motion vector that belongs to same and be (0,1/2) according to equation (1).If predict that as the predicted value of the top of the coded fields in first line chroma pixel 142 then location of pixels 143 is used as predicted value with motion vector (0,1/2), its pixel from first line of end reference field chromatic component moves down half-pixel.
At this, brightness movement vector (0,1) and chrominance vector (0,1/2) are uneven.Preferably use end reference field colorimetric prediction location of pixels 145, on this position, applied the chroma motion vector that is parallel to brightness movement vector.
Figure 14 has shown when reference field and coded fields are top and field, the end respectively, has used the caused problem of brightness movement vector calculating chroma motion vector.As Figure 13, shown in Figure 14, the end coded fields luminance pixel 150 in first line has predictive vector (0,1), so it points to reference top field brightness location of pixels 151 in second line as predicted value.
At this, calculate the chroma motion vector that belongs to same according to equation (1) and be (0,1/2).If predict that as the predicted value of end coded fields chroma pixel 152 then location of pixels 153 is used as predicted value with motion vector (0,1/2), its reference field chroma pixel position, top 153 skew half-pixels from first line.
At this, brightness movement vector (0,1) and chrominance vector (0,1/2) are not parallel.Preferably use top reference field colorimetric prediction location of pixels 155, on this position, applied the chroma motion vector that is parallel to brightness movement vector.
As mentioned above, if the parity of reference field is different with the parity of coded fields, then according to traditional Forecasting Methodology, will be with reference to the pixel that is positioned at the luminance component position of spatially departing from the chromatic component position, and, not only be zero vector but also for all vector generation forecast images, in predicted picture, be positioned at the position that the pixel of luminance component position spatially departs from chromatic component.Notice that in the above description, say that vector is parallel or not parallel under the situation of the time orientation of considering brightness movement vector and chroma motion vector, promptly the time orientation from the coded fields to the reference field is comprised in the motion vector.In the following explanation also is like this.
Summary of the invention
One object of the present invention is to provide a kind of dynamic image encoding method and dynamic image encoding device, and it can improve the forecasting efficiency of chromatic component in the coding between the different field picture, and therefore improves code efficiency.
Dynamic image encoding method of the present invention carries out motion compensated prediction and execution encoding process between the field to motion image signal, each frame in the described motion image signal is made of field, top and field, the end, and the pixel count of colourity vertical component is different with the pixel count of brightness vertical component, it is characterized in that, when the value of vector component is that the luminance component motion vector of movement in vertical direction of a pixel of the unit luminance component of establishing the expression field picture is MVy with 4, the value of vector component is that the chromatic component motion vector of movement in vertical direction of a pixel of the unit chromatic component of establishing the expression field picture is when being MVCy with 8, according to reference field and coded fields be respectively the top or the different situations of field, the end generate the chromatic component motion vector, promptly, when reference field and coded fields all are top or when all being field, the end, MVCy and MVy have the relation such as MVCy=MVy; When reference field was field, the end for a top coded fields, MVCy and MVy had the relation such as MVCy=MVy+2; When reference field was the field, top for field, end coded fields, MVCy and MVy had the relation such as MVCy=MVy-2.
Dynamic image encoding device of the present invention carries out motion compensated prediction and execution encoding process between the field to motion image signal, the field constitutes with field, the end each frame in the described motion image signal and the pixel count of colourity vertical component is different with the pixel count of brightness vertical component by pushing up, it is characterized in that this dynamic image encoding device has: judge that reference field and coded fields are respectively the field, top or the unit of field, the end; And a kind of like this unit, it generates the chromatic component motion vector, is that the luminance component motion vector of movement in vertical direction of a pixel of the unit luminance component of establishing the expression field picture is MVy in the value of vector component with 4, the value of vector component is that the chromatic component motion vector of movement in vertical direction of a pixel of the unit chromatic component of establishing the expression field picture is when being MVCy with 8, when reference field and coded fields all are top or when all being field, the end, MVCy and MVy have the relation such as MVCy=MVy; When reference field was field, the end for a top coded fields, MVCy and MVy had the relation such as MVCy=MVy+2; When reference field was the field, top for field, end coded fields, MVCy and MVy had the relation such as MVCy=MVy-2.
Motion picture encoding apparatus of the present invention is used for the motion image signal that is made of a plurality of fields is carried out motion compensation between the field, this equipment comprises: a plurality of chroma motion vector generation units, and it uses brightness movement vector to generate chroma motion vector in motion picture encoding apparatus; Selected cell, it utilizes the reference field and the coded fields parity separately of motion vector, selects with a chroma motion vector generation unit that generates chrominance vector.The chroma motion vector generation unit of being selected by selected cell generates the colorimetric prediction vector according to the motion vector information of monochrome information.
Moving picture decoding equipment of the present invention is used for the motion image signal that is made of a plurality of fields is carried out motion compensation between the field, and it comprises: a plurality of chroma motion vector generation units, and it generates chroma motion vector by brightness movement vector; And selected cell, it utilizes the parity of reference field of motion vector and the parity of coded fields, selects with a chroma motion vector generation unit that generates chrominance vector.The chroma motion vector generation unit of being selected by selected cell generates the colorimetric prediction vector according to the motion vector information of monochrome information.
According to the present invention, owing to used according to the parity of coding/decoding field and reference field and, solved field, top and the configuration of field, the end or the deviation of the different chroma motion vectors that cause of the method for salary distribution owing to luminance pixel and chroma pixel by the chroma motion vector that proper method generates.
In addition, by the present invention,, can solve in conventional method the problem of reference pixel offset between the luminance component and chromatic component even, also can obtain to be parallel to the chroma motion vector of brightness movement vector for the different field of parity.
Description of drawings
Fig. 1 has shown the configuration of inter prediction encoding equipment;
Fig. 2 has shown the position separately of brightness and chroma pixel and the field under each pixel;
Fig. 3 has shown brightness and chroma pixel vertical time and locus separately in the field picture;
Fig. 4 has shown the relation of frame encoding mode midfield and frame;
Fig. 5 has shown the Forecasting Methodology of inter prediction encoding pattern;
Fig. 6 has shown the Forecasting Methodology of predictive coding pattern between the field;
Fig. 7 has shown the coordinate of field picture;
A Traditional calculating methods of motion vector between the respective pixel between Fig. 8 has shown;
Fig. 9 has shown the Traditional calculating methods of motion vector indication pixel;
Figure 10 has shown the Traditional calculating methods of using brightness movement vector calculating chroma motion vector;
Figure 11 has shown the computational methods of the interpolating pixel of chromatic component;
Figure 12 has shown the principle of traditional Direct Model, in order to explain the null vector between the different field of parity;
Figure 13 has shown when reference field and coded fields are field, the end and top respectively, has used brightness movement vector to calculate the caused problem of chroma motion vector;
Figure 14 has shown when reference field and coded fields are top and field, the end respectively, has used the caused problem of brightness movement vector calculating chroma motion vector;
Figure 15 has shown the method for using brightness movement vector generation chroma motion vector among the present invention;
Figure 16 has shown the operation of a preferred embodiment of the first chroma motion vector generation unit of the present invention;
Figure 17 has shown the operation of a preferred embodiment of the second chroma motion vector generation unit of the present invention;
Figure 18 has shown the operation of a preferred embodiment of the 3rd chroma motion vector generation unit of the present invention;
Figure 19 has shown the operation of a preferred embodiment of selected cell of the present invention;
Figure 20 has shown when reference field and coded fields are field, the end and top respectively, has used brightness movement vector to calculate example of the present invention of chroma motion vector;
Figure 21 has shown when reference field and coded fields are top and field, the end respectively, uses example of the present invention of brightness movement vector calculating chroma motion vector;
Figure 22 has shown the operation of another preferred embodiment of the first chroma motion vector generation unit of the present invention;
Figure 23 has shown the operation of another preferred embodiment of the second chroma motion vector generation unit of the present invention; And
Figure 24 has shown the operation of another preferred embodiment of the 3rd chroma motion vector generation unit of the present invention.
Embodiment
Coding principle among the present invention is at first described.
Motion picture encoding apparatus of the present invention is used for the motion image signal that is made of a plurality of fields is carried out motion compensation between the field, and this equipment comprises: a plurality of chroma motion vector generation units are used to use brightness movement vector to generate chroma motion vector; Selected cell utilizes the reference field and the coded fields parity separately of motion vector, selects with a chroma motion vector generation unit that generates chrominance vector.The chroma motion vector generation unit of being selected by selected cell generates the colorimetric prediction vector according to the motion vector information of monochrome information.
If the chroma motion vector from the coded fields to the reference field is parallel to the brightness movement vector from the coded fields to the reference field, then the spatial deviation of brightness movement vector equates with the spatial deviation of chroma motion vector, promptly, relation between the locus of brightness movement vector and chroma motion vector is maintained, and the misalignment between disappears.
Herein importantly, in conventional method, even be parallel to chroma motion vector according to the mathematical expression brightness movement vector, but when these vectors were mapped to concerning between the luminance pixel that constitutes each and between the chroma pixel, each vector was also not parallel.
A plurality of chroma motion vector generation units comprise following three types.
When reference field and coded fields had identical parity, selected cell was selected the first chroma motion vector generation unit.When reference field and coded fields were top and field, the end respectively, selected cell was selected the second chroma motion vector generation unit.When reference field and coded fields were field, the end and top respectively, selected cell was selected the 3rd chroma motion vector generation unit.
The method that calculating is parallel to the chroma motion vector of brightness movement vector depends on the coded fields parity and the reference field parity of brightness movement vector.These computational methods are variant in following three kinds of situations: a kind of situation is that the coded fields parity is identical with the reference field parity, a kind of situation is that coded fields and reference field are respectively field, top and field, the end, and also having a kind of situation is that coded fields and reference field are respectively field, the end and field, top.Therefore, in the present invention, from three types chroma motion vector generation unit, select one of the best to calculate the chroma motion vector that is parallel to brightness movement vector according to coded fields and reference field.
Particularly, if the reference field parity is identical with the coded fields parity, the following calculating chroma motion vector of the first chroma motion vector generation unit then, suppose value " 1 " with vector component be a unit luminance pixel representing field picture vertical displacement brightness movement vector and be that the chroma motion vector of the vertical displacement of a unit chroma pixel representing field picture is respectively MVy and MVCy with the value " 1 " of vector component.
MVCy＝MVy/2 (3)
If reference field parity and coded fields parity are respectively field, top and field, the end, the following calculating chroma motion vector of the second chroma motion vector generation unit then, suppose value " 1 " with vector component be a unit luminance pixel representing field picture vertical displacement brightness movement vector and be that the chroma motion vector of the vertical displacement of a unit chroma pixel representing field picture is respectively MVy and MVCy with the value " 1 " of vector component.
MVCy＝MVy/2+0.25 (4)
If reference field parity and coded fields parity are respectively field, the end and field, top, the following calculating chroma motion vector of the 3rd chroma motion vector generation unit then, suppose value " 1 " with vector component be a unit luminance pixel representing field picture vertical displacement brightness movement vector and be that the chroma motion vector of the vertical displacement of a unit chroma pixel representing field picture is respectively MVy and MVCy with the value " 1 " of vector component.
MVCy＝MVy/2-0.25 (5)
Sometimes, the unit of brightness and chrominance vector changes according to its definition.As situation about giving a definition in: during when brightness movement vector component change value " 4 ", brightness movement vector is represented the displacement of a brightness campaign pixel, and when chroma motion vector component change value " 8 ", chroma motion vector is represented the displacement of a colourity motion pixel, if the reference field parity is identical with the coded fields parity, then the following calculating chroma motion vector of the first chroma motion vector generation unit supposes that brightness movement vector and chroma motion vector are respectively MVy and MVCy.
MVCy＝Mvy (6)
In identical definition, if the parity of reference field and coded fields is respectively field, top and field, the end, then the following calculating chroma motion vector of the second chroma motion vector generation unit supposes that brightness movement vector and chroma motion vector are respectively MVy and MVCy.
MVCy＝MVy+2 (7)
In identical definition, if the parity of reference field and coded fields is respectively field, the end and field, top, then the following calculating chroma motion vector of the 3rd chroma motion vector generation unit supposes that brightness movement vector and chroma motion vector are respectively MVy and MVCy.
MVCy＝MVy-2 (8)
Moving picture decoding equipment of the present invention has and the motion picture encoding apparatus identical functions basically, and operation in the same manner.
Below the main preferred embodiment of describing encoding device.This encoding device has above-mentioned configuration.Because the present invention relates to the vertical component of motion vector, for simplicity, the horizontal component of supposing all motion vectors is 0.At this, decoding device has the configuration identical with encoding device.
Below hypothesis has adopted AVC FCD and has described the preferred embodiments of the present invention.
Figure 15 has shown the method for using brightness movement vector calculating chroma motion vector.The preferred embodiment that uses brightness movement vector to generate the equipment of chroma motion vector in field prediction comprises chroma motion vector generation unit and a selected cell of three types.
Operation of the present invention shown in Figure 15 is described below.At first suppose given brightness movement vector 231 be (MV_x, MV_y).This luminance vector is transfused to all first chroma motion vector generation unit 233, the second chroma motion vector generation unit 234 and the 3rd chroma motion vector generation unit 235.Then, their output separately is transfused to selected cell 230.Selected cell 230 bases are about the coded fields parity 237 of input motion vector and the information of reference field parity 238 thereof, select in the output of first, second and the 3rd chroma motion vector generation unit, and its be output as chroma motion vector 232 (MVC_x, MVC_y).
Figure 16 has shown the operation of the first chroma motion vector generation unit.In the preferred embodiment, brightness movement vector 261 (MV_x MV_y) is transfused to the first chroma motion vector generation unit 260, export first candidate's chroma motion vector 262 (MVC1_x, MVC1_y).Chroma motion vector generation unit 260 uses brightness movement vector 261 following calculating first candidate's chroma motion vectors 262.
(MVC1_x，MVC1_y)＝(MV_x/2，MV_y/2) (9)
Then, the first candidate's chroma motion vector 262 that calculates is exported to selected cell.
Figure 17 has shown the operation of the second chroma motion vector generation unit.In the preferred embodiment, brightness movement vector 271 (MV_x MV_y) is transfused to the second chroma motion vector generation unit 270, export second candidate's chroma motion vector 272 (MVC2_x, MVC2_y).Chroma motion vector generation unit 270 uses brightness movement vector 271 following calculating second candidate's chroma motion vectors 272.
(MVC2_x，MVC2_y)＝(MV_x/2，MV_y/2+1/4) (10)
Then, the second candidate's chroma motion vector 272 that calculates is exported to selected cell.
Figure 18 has shown the operation of the 3rd chroma motion vector generation unit.In the preferred embodiment, brightness movement vector 281 (MV_x MV_y) is transfused to the 3rd chroma motion vector generation unit 280, export the 3rd candidate's chroma motion vector 282 (MVC3_x, MVC3_y).Chroma motion vector generation unit 280 uses brightness movement vector 281 following calculating the 3rd candidate's chroma motion vectors 282.
(MVC3_x，MVC3_y)＝(MV_x/2，MV_y/2-1/4) (11)
Then, the 3rd candidate's chroma motion vector 282 that calculates is exported to selected cell.
Figure 19 has shown the operation of a preferred embodiment of selected cell 240 of the present invention.At first, in the preferred embodiment, service condition judgement table 241 is judged the coded fields parity 247 and the reference field parity 248 thereof of motion vector, and exports the selection information 249 of the chroma motion vector generation unit that will select.In the preferred embodiment, if reference field is identical with coded fields, then this condition judgment table 241 is used to export the selection information that first candidate's chroma motion vector 244 is selected in expression.If reference field and coded fields are respectively field, top and field, the end, then condition judgment table 241 is used to export the selection information that the second chrominance vector candidate 245 is selected in expression.If reference field and coded fields are respectively field, the end and field, top, then condition judgment table 241 is used to export the selection information that the 3rd chrominance vector candidate 246 is selected in expression.
In this case, first, second or the 3rd candidate's chroma motion vector 244,245 and 246 are connected respectively to 282 among 272 and Figure 18 among 262 among Figure 16, Figure 17.Then, selector 243 is selected in first, second and the 3rd candidate's chroma motion vector 244,245 and 246 one according to selection information 249, and output (MVC_x is MVC_y) as its chroma motion vector 242.
Figure 20 has shown that of the present invention is respectively to use luminance vector to calculate the operation of chrominance vector under field, the end and the situation of pushing up the field in reference field and coded fields.In example shown in Figure 20, suppose to be used for predicting that (MV_x MV_y) is (0,1) for the brightness movement vector of top coded fields pixel 160.Under this situation, field brightness location of pixels 161 is with prediction luminance pixel 160 at the bottom of the selection reference field.The computational process of the chroma motion vector will be used for predicting top coded fields chroma pixel 162 is described below with reference to Figure 15.
At first, reference field and coded fields are respectively field, the end and field, top among Figure 20.In this case, use condition judgment table 241 shown in Figure 19 is selected the selection information 249 about the 3rd candidate's chroma motion vector.According to equation (11), the 3rd candidate's chroma motion vector is calculated as follows.
(MVC3_x，MVC3_y)＝(MV_x/2，MV_y/2-1/4)
＝(0/2，1/2-1/4)
＝(0，1/4) (12)
Then, export this value as chroma motion vector shown in Figure 19 242.If this vector (0,1/4) puts on top coded fields chroma pixel 162, then reference field chroma pixel position, the end 163 is used as predicted value.Among Figure 20, the vertical positional relationship between the pixel is corresponding to the vertical positional relationship between the actual pixels.Can be clear that from Figure 20 brightness movement vector (0,1) and chroma motion vector (0,1/4) are parallel to each other.Therefore, brightness and the color error ratio between the chromatic component as traditional problem can be resolved in the present invention.
Equally, Figure 21 has shown that of the present invention is respectively the operation of using luminance vector calculating chrominance vector under the situation of top and field, the end in reference field and coded fields.
In example shown in Figure 21, suppose to be used for predicting that (MV_x MV_y) is (0,1) for the brightness movement vector of end coded fields pixel 170.Under this situation, select reference field luminance pixel position, top 171 with prediction luminance pixel 170.The computational process of the chroma motion vector will be used for predicting end coded fields chroma pixel 172 is described below with reference to Figure 15.
At first, reference field and coded fields are respectively field, top and field, the end among Figure 21.In this case, use condition judgment table 241 shown in Figure 19 is selected the selection information 249 about second candidate's chroma motion vector.According to equation (10), candidate's second chroma motion vector is calculated as follows.
(MVC2_x，MVC2_y)＝(MV_x/2，MV_y/2+1/4)
＝(0/2，1/2+1/4)
＝(0，3/4) (13)
Then, export this value as chroma motion vector shown in Figure 19 242.If this vector (0,3/4) puts on end coded fields chroma pixel 172, then top reference field chroma pixel position 173 is used as predicted position.Among Figure 21, the vertical positional relationship between the pixel is corresponding to the vertical positional relationship between the actual pixels.Can be clear that from Figure 21 brightness movement vector (0,1) and chroma motion vector (0,3/4) are parallel to each other.Therefore, brightness and the misalignment between the chromatic component as traditional problem can be resolved in the present invention.
Though in the example shown in Figure 20 and 21, described the prediction of specific vector, in the prediction between the field of other parities, also can realize not having between brightness and the colourity prediction of misalignment by application the preferred embodiment.
When the reference field parity was identical with the coded fields parity, this misalignment did not take place.Therefore, from traditional brightness movement vector shown in Figure 10, select the result of the of the present invention first chroma motion vector generation unit 233 identical, and this result is used as chroma motion vector 232 with 220 configurations of chroma motion vector generation unit.Because in this case, the chroma motion vector that the present invention calculates is with traditional identical, the description of just having omitted the preferred embodiment here.
In another aspect of this invention, equation (9), (10) and (11) change according to the unit of brightness and chroma motion vector.
Figure 22 to 24 has shown another embodiment of the first chroma motion vector generation unit of the present invention, the second chroma motion vector generation unit and the 3rd chroma motion vector generation unit.
As situation about giving a definition in: when the value of brightness movement vector changes " 4 ", brightness movement vector is represented the displacement of a brightness campaign pixel, and when the value of chroma motion vector changes " 8 ", chroma motion vector is represented the displacement of a colourity motion pixel, and chroma motion vector generation unit 260a uses the following calculated candidate first chroma motion vector 262a of brightness movement vector 261a.
(MVC1_x，MVC1_y)＝(MV_x，MV_y) (14)
Then, the first candidate's chroma motion vector 262a that calculates is exported to selected cell.
Chroma motion vector generation unit 270a uses the following calculating second candidate's chroma motion vector 272a of brightness movement vector 271a.
(MVC2_x，MVC2_y)＝(MV_x，MV_y+2) (15)
Then, the second candidate's chroma motion vector 272a that calculates is exported to selected cell.
Chroma motion vector generation unit 280a uses following calculating the 3rd candidate's chroma motion vector 282a of brightness movement vector 281a.
(MVC3_x，MVC3_y)＝(MV_x，MV_y-2) (16)
Then, the 3rd candidate's chroma motion vector 282a that calculates is exported to selected cell.
Though described the preferred embodiment supposes that it has adopted AVC FCD, this is a preferred embodiment, and the coded format of field picture is not limited to this.
According to the present invention, in the different field of parity, also can calculate the chroma motion vector that is parallel to brightness movement vector, therefore can solve the problem that has deviation as reference pixel position between the brightness of traditional problem and the chromatic component.
Claims (2)
1. dynamic image encoding method, this dynamic image encoding method carries out motion compensated prediction and execution encoding process between the field to motion image signal, the field constitutes with field, the end each frame in the described motion image signal and the pixel count of colourity vertical component is different with the pixel count of brightness vertical component by pushing up, it is characterized in that
When the value of vector component is that the luminance component motion vector of movement in vertical direction of a pixel of the unit luminance component of establishing the expression field picture is MVy with 4, the value of vector component is that the chromatic component motion vector of movement in vertical direction of a pixel of the unit chromatic component of establishing the expression field picture is when being MVCy with 8, according to reference field and coded fields be respectively the top or the different situations of field, the end generate the chromatic component motion vector, promptly
When reference field and coded fields all are top or when all being field, the end, MVCy and MVy have the relation such as MVCy=MVy;
When reference field was field, the end for a top coded fields, MVCy and MVy had the relation such as MVCy=MVy+2;
When reference field was the field, top for field, end coded fields, MVCy and MVy had the relation such as MVCy=MVy-2.
2. dynamic image encoding device, this dynamic image encoding device carries out motion compensated prediction and execution encoding process between the field to motion image signal, the field constitutes with field, the end each frame in the described motion image signal and the pixel count of colourity vertical component is different with the pixel count of brightness vertical component by pushing up, it is characterized in that this dynamic image encoding device has:
Judge that reference field and coded fields are respectively the field, top or the unit of field, the end; And
A kind of like this unit, it generates the chromatic component motion vector, is that the luminance component motion vector of movement in vertical direction of a pixel of the unit luminance component of establishing the expression field picture is MVy in the value of vector component with 4, the value of vector component is that the chromatic component motion vector of movement in vertical direction of a pixel of the unit chromatic component of establishing the expression field picture is when being MVCy with 8
When reference field and coded fields all are top or when all being field, the end, MVCy and MVy have the relation such as MVCy=MVy;
When reference field was field, the end for a top coded fields, MVCy and MVy had the relation such as MVCy=MVy+2;
When reference field was the field, top for field, end coded fields, MVCy and MVy had the relation such as MVCy=MVy-2.
Applications Claiming Priority (4)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| JP2002261427 | 2002-09-06 | ||
| JP261427/2002 | 2002-09-06 | ||
| JP289350/2003 | 2003-08-07 | ||
| JP2003289350A JP3791922B2 (en) | 2002-09-06 | 2003-08-07 | Moving picture decoding apparatus and method |
Related Parent Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB031566103A Division CN1324904C (en) | 2002-09-06 | 2003-09-05 | Kinematic image coder and decoder |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101043629A CN101043629A (en) | 2007-09-26 |
| CN101043629B true CN101043629B (en) | 2011-01-26 |
Family
ID=36773690
Family Applications (7)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200710103419 Expired - Lifetime CN101043629B (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding method and devices thereof |
| CN 200510130221 Expired - Lifetime CN100502512C (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding device and method |
| CN 200710103418 Expired - Lifetime CN101043628B (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding and decoding devices |
| CN 200710103420 Expired - Lifetime CN101043630B (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding and decoding devices |
| CN 200710103417 Expired - Lifetime CN101043627B (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding device and motion picture decoding devices |
| CN 200510130222 Expired - Lifetime CN100502513C (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding device and motion picture decoding method |
| CN 200510130223 Pending CN1791225A (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding device and motion picture decoding device |
Family Applications After (6)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200510130221 Expired - Lifetime CN100502512C (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding device and method |
| CN 200710103418 Expired - Lifetime CN101043628B (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding and decoding devices |
| CN 200710103420 Expired - Lifetime CN101043630B (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding and decoding devices |
| CN 200710103417 Expired - Lifetime CN101043627B (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding device and motion picture decoding devices |
| CN 200510130222 Expired - Lifetime CN100502513C (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding device and motion picture decoding method |
| CN 200510130223 Pending CN1791225A (en) | 2002-09-06 | 2003-09-05 | Motion picture encoding device and motion picture decoding device |
Country Status (1)
| Country | Link |
|---|---|
| CN (7) | CN101043629B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101170707B (en) * | 2007-11-08 | 2012-08-08 | 华为技术有限公司 | Image format conversion and recovery method and device |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP0600510A2 (en) * | 1992-12-04 | 1994-06-08 | Sony Corporation | Moving picture decoding device |
| US5506621A (en) * | 1992-08-21 | 1996-04-09 | Canon Kabushiki Kaisha | Image processing method and apparatus |
| CN1140958A (en) * | 1994-09-23 | 1997-01-22 | Lg电子株式会社 | Half-pixel motion compensation controller accepting MPEG2 |
| EP0863675A2 (en) * | 1997-03-07 | 1998-09-09 | General Instrument Corporation | Motion estimation and compensation of video object planes for interlaced digital video |
| US5900910A (en) * | 1991-06-25 | 1999-05-04 | Canon Kabushiki Kaisha | Block matching method and apparatus which can use luminance/chrominance data |
| EP0955607A2 (en) * | 1998-05-07 | 1999-11-10 | Sarnoff Corporation | Method and apparatus for adaptively scaling motion vector information |
| CN1347620A (en) * | 1999-02-18 | 2002-05-01 | 通用仪器公司 | Method and architecture for converting MPE G-2 4:2:2-profile bitstreams into main-profile bitstreams |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US6108039A (en) * | 1996-05-23 | 2000-08-22 | C-Cube Microsystems, Inc. | Low bandwidth, two-candidate motion estimation for interlaced video |
-
2003
- 2003-09-05 CN CN 200710103419 patent/CN101043629B/en not_active Expired - Lifetime
- 2003-09-05 CN CN 200510130221 patent/CN100502512C/en not_active Expired - Lifetime
- 2003-09-05 CN CN 200710103418 patent/CN101043628B/en not_active Expired - Lifetime
- 2003-09-05 CN CN 200710103420 patent/CN101043630B/en not_active Expired - Lifetime
- 2003-09-05 CN CN 200710103417 patent/CN101043627B/en not_active Expired - Lifetime
- 2003-09-05 CN CN 200510130222 patent/CN100502513C/en not_active Expired - Lifetime
- 2003-09-05 CN CN 200510130223 patent/CN1791225A/en active Pending
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US5900910A (en) * | 1991-06-25 | 1999-05-04 | Canon Kabushiki Kaisha | Block matching method and apparatus which can use luminance/chrominance data |
| US5506621A (en) * | 1992-08-21 | 1996-04-09 | Canon Kabushiki Kaisha | Image processing method and apparatus |
| EP0600510A2 (en) * | 1992-12-04 | 1994-06-08 | Sony Corporation | Moving picture decoding device |
| CN1140958A (en) * | 1994-09-23 | 1997-01-22 | Lg电子株式会社 | Half-pixel motion compensation controller accepting MPEG2 |
| EP0863675A2 (en) * | 1997-03-07 | 1998-09-09 | General Instrument Corporation | Motion estimation and compensation of video object planes for interlaced digital video |
| EP0955607A2 (en) * | 1998-05-07 | 1999-11-10 | Sarnoff Corporation | Method and apparatus for adaptively scaling motion vector information |
| CN1347620A (en) * | 1999-02-18 | 2002-05-01 | 通用仪器公司 | Method and architecture for converting MPE G-2 4:2:2-profile bitstreams into main-profile bitstreams |
Non-Patent Citations (1)
| Title |
|---|
| JP特开2001-238228A 2001.08.31 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100502512C (en) | 2009-06-17 |
| CN101043628A (en) | 2007-09-26 |
| CN101043628B (en) | 2010-12-15 |
| CN1784017A (en) | 2006-06-07 |
| CN100502513C (en) | 2009-06-17 |
| CN101043630B (en) | 2013-03-27 |
| CN1791225A (en) | 2006-06-21 |
| CN1801943A (en) | 2006-07-12 |
| CN101043627A (en) | 2007-09-26 |
| CN101043627B (en) | 2010-12-08 |
| CN101043630A (en) | 2007-09-26 |
| CN101043629A (en) | 2007-09-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US9549192B2 (en) | Motion picture encoding device and motion picture decoding device | |
| CN101043629B (en) | Motion picture encoding method and devices thereof | |
| CA2591655C (en) | A device and method for providing inter-field motion compensation for chrominance motion vectors while decoding motion pictures |
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 | ||
| CX01 | Expiry of patent term |
Granted publication date: 20110126 |
|
| CX01 | Expiry of patent term |