CN101056398A - A method and decoding and encoding method for capturing the video difference vector in the multi-video coding process - Google Patents
A method and decoding and encoding method for capturing the video difference vector in the multi-video coding process Download PDFInfo
- Publication number
- CN101056398A CN101056398A CNA2006100709719A CN200610070971A CN101056398A CN 101056398 A CN101056398 A CN 101056398A CN A2006100709719 A CNA2006100709719 A CN A2006100709719A CN 200610070971 A CN200610070971 A CN 200610070971A CN 101056398 A CN101056398 A CN 101056398A
- Authority
- CN
- China
- Prior art keywords
- difference vector
- coding
- looked
- looking
- look
- 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.)
- Pending
Links
- 239000013598 vector Substances 0.000 title claims abstract description 280
- 238000000034 method Methods 0.000 title claims abstract description 105
- 230000001788 irregular Effects 0.000 claims description 5
- 230000015572 biosynthetic process Effects 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000003384 imaging method Methods 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000006073 displacement reaction Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000004513 sizing Methods 0.000 description 1
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention relates to a disparity vector acquiring method during a multi-view encoding process, and an encoding and a decoding method thereof. The invention comprises confirming a disparity vector between two views during the multi-view image encoding process, calculating to obtain a disparity vector between other two views according to the preceding disparity vector between two views and known relative position information of views, then performing multi-view encoding and decoding operations by means of the disparity vectors. The method of the invention, on one hand, primely utilizes the correlation between the disparity vector and the space object deep information, and on the other hand, primely utilizes the direct relation between the disparity vector and the position of each camera. Experiments show that the disparity vector between views can be precisely calculated during the multi-view encoding process, thereby effectively improving the multi-view encoding performance.
Description
Technical field
The present invention relates to the image coding technique in the multimedia field, relate in particular to a kind of technology of definite difference vector.
Background technology
In looking image coding technology field, industry wishes that always the correlation that can effectively utilize between the different views picture encodes to image more.For this reason, proposed the notion of parallax (Disparity), described parallax is meant the displacement of same spatial point between two subpoints on the different image planes.
Parallax then needs to carry out corresponding parallax valuation operation in order to obtain accurately, described parallax valuation (Disparity estimation) is a process of seeking the difference vector of pixel corresponding with present frame in the reference frame or piece under certain criterion, and described difference vector is needs definite parallax value.
At present, because the process of definite described difference vector and the estimating motion process of the definite motion vector in traditional haplopia video coding are similar, therefore, looking in the coding standard formulation of ITU/MPEG tissue advises using traditional haplopia video coding instrument to carry out looking coding more more at present.Wherein, the method for the search that is based on piece equally of the employing of the estimating motion in parallax estimation algorithm and the traditional haplopia video coding realizes.
Yet, cause because motion vector is the motion because of object and video camera, difference vector then is that the skew because of camera position causes, therefore, there are many differences between difference vector and the motion vector, make and adopt definite mode of motion vector can't determine difference vector accurately.
At first, satisfy and be arranged in parallel and proofread and correct if be used to obtain the video camera of image, then difference vector is a scalar, and the motion vector that estimating motion obtains is the vector of two dimension; And for the motion sequence of haplopia, motion-vector field has the correlation of time domain, and the predicted value of difference vector is then irrelevant with time domain; Thereby, because motion vector and difference vector is different, if adopt definite mode of motion vector to determine that difference vector can't obtain the difference vector value of being satisfied with.
And, usually the dynamic range of difference vector is bigger, with size is that the image of 640 * 480 pixels is an example, in stereo pair, horizontal displacement may often occur in the situation of 30-50 pixel, sometimes even can reach 100 more than the pixel, motion vector does not then have so big dynamic range; Therefore,,, then need to establish the hunting zone very big if adopt the estimating motion method in traditional haplopia video coding because the dynamic range of difference vector is bigger, and, can't guarantee to obtain good effect.
Summary of the invention
The purpose of this invention is to provide and a kind of look method and the decoding method that obtains difference vector in the cataloged procedure more, thereby it is more accurate to make the difference vector of difference between looking estimate, performance more that look coding with effective raising.
The objective of the invention is to be achieved through the following technical solutions:
The invention provides and a kind of look the method for obtaining difference vector in the cataloged procedure more, this method is applied to view more than at least three is looked like to handle, and comprises step:
Determine that described many views are as any two first difference vectors between looking in the cataloged procedure;
According to described first difference vector, and each relative position information between looking, calculate two of obtaining that coding needs, second difference vector between looking.
Described method specifically comprises:
In a plurality of looking, determine three and look that three comprise that a coding to be encoded is looked with two references in looking and look, and determine three any two first difference vectors between looking in looking; Afterwards, according to described first difference vector, and three relative position informations between looking, by calculate obtain that coding is looked and in addition any one with reference to second difference vector between looking.
Described any two first difference vectors between looking are:
Coding is looked and two references look in any with reference to the difference vector between looking, perhaps, the difference vector between two references are looked.
Described method also comprises step:
The reference apparent place that described second difference vector is pointed to is being done disparity estimation with reference to the mode of looking according to predetermined as original position in predefined scope, obtain encoding look look with reference between new difference vector value.
Be regularly arranged or irregular alignment between described a plurality of the looking.
Describedly regularly arrangedly be: yi word pattern, cross arrangement or square formation type are arranged.
The present invention also provides a kind of coding method of looking based on difference vector more, and this method is applied to view more than at least three is looked like to handle, and comprising:
Determine that many views are as any two first difference vectors between looking in the cataloged procedure;
According to described first difference vector, and each relative position information between looking, calculate two of obtaining that coding needs, second difference vector between looking;
According to described second difference vector coding is looked the also output encoder result that encodes;
Write in the code stream with described coding result, described first difference vector and behind the coding of the relative position information between respectively looking, perhaps, with described coding result, and first difference vector and second difference vector coding and write code stream.
In the method for the present invention:
One is that coding is looked in the described picture of view more than at least three, and remaining is looked for reference;
Described first difference vector be described coding look and described one of them with reference to the difference vector between looking;
Described second difference vector look for coding and another one with reference to the difference vector between looking.
In the method for the present invention, described according to described second difference vector to coding look encode and the output encoder result be specially:
The predicted value that the respective pixel weighting summation that the reference that respective pixel that the reference that described first difference vector is pointed to is looked and described second difference vector point to is looked obtains encoding and looks;
The original value that coding is looked deducts the predicted value that described coding is looked, the residual values that obtains encoding and look, and this residual values encoded obtain the coding result looked at coding.
In the method for the present invention:
One is that coding is looked in the described picture of view more than at least three, and remaining is looked for reference;
Described first difference vector is the difference vector between wherein two references are looked;
Described second difference vector comprises resolute: the 4th difference vector and the 5th difference vector, wherein:
Described the 4th difference vector is looked for coding and the difference vector of a reference between looking;
Described the 5th difference vector looks for coding and another is with reference to the difference vector between looking.
In the method for the present invention:
Described according to described second difference vector to coding look encode and the output encoder result be specially:
The predicted value that the respective pixel weighting summation that the reference that respective pixel that the reference that described the 4th difference vector is pointed to is looked and described the 5th difference vector point to is looked obtains encoding and looks;
The original value that coding is looked deducts the predicted value that described coding is looked, the residual values that obtains encoding and look, and this residual values encoded obtain the coding result looked at coding.
The present invention also provides a kind of coding/decoding method of looking based on difference vector more, and this method realizes that in decoding end this method is applied to view more than at least three is looked like to handle, and comprises step:
Receive code stream, parsing obtain definite any two between looking first difference vector and respectively look between relative position information, according to described first difference vector and the relative position information between respectively looking calculate and determine and two of coding side corresponding codes needs second difference vector between looking; Perhaps, flow through journey by described decoding end at receiving code and directly obtain described second difference vector;
Utilize described second difference vector that code stream is decoded, the coding that obtains coding side is looked corresponding reconstructed value.
In the method for the present invention:
One is that coding is looked in the described picture of view more than at least three, and remaining is looked for reference;
Described first difference vector be described coding look and described one of them with reference to the difference vector between looking; Described second difference vector look for coding and another one with reference to the difference vector between looking;
And describedly utilize described second difference vector that code stream is decoded, the coding that obtains coding side is looked corresponding reconstructed value and is specially:
The predicted value that the respective pixel weighting summation that the reference that respective pixel that the reference that described first difference vector is pointed to is looked and described second difference vector point to is looked obtains encoding and looks;
The residual values that the coding of decoding acquisition is looked adds the predicted value that described coding is looked, the reconstructed value that obtains encoding and look.
In the method for the present invention:
One is that coding is looked in the described picture of view more than at least three, and remaining is looked for reference;
Described second difference vector comprises resolute: the 4th difference vector and the 5th difference vector, wherein:
Described the 4th difference vector is looked for coding and the difference vector of a reference between looking;
Described the 5th difference vector looks for coding and another is with reference to the difference vector between looking.
Describedly utilize described second difference vector that code stream is decoded, the coding that obtains coding side is looked corresponding reconstructed value and is specially:
The predicted value that the respective pixel weighting summation that the reference that respective pixel that the reference that described the 4th difference vector is pointed to is looked and described the 5th difference vector point to is looked obtains encoding and looks;
The residual values that the coding of decoding acquisition is looked adds the predicted value that described coding is looked, the reconstructed value that obtains encoding and look.
The present invention also provides a kind of and has looked the method for obtaining difference vector in the cataloged procedure more, and this method is applied to view more than at least three is looked like to handle, and comprises step:
Determine the depth information of space object, and described each relative position information between looking;
Utilize the depth information of described space object and each relative position information between looking to calculate to obtain coding to need two difference vectors between looking.
Described method also comprises step:
Pointing to the position with described difference vector is original position, is doing disparity estimation with reference to the mode of looking according to predetermined in predefined scope, obtain encoding look look with reference between new difference vector value.
The present invention also provides a kind of coding method of looking based on difference vector more, and this method is applied to view more than at least three is looked like to handle, and comprises step:
Determine the depth information of space object, and described each relative position information between looking;
Utilize the depth information of described space object and described each relative position information between looking to calculate to obtain coding to need two difference vectors between looking;
Utilizing described difference vector that coding is looked encodes and writes code stream; Simultaneously, also the depth information of space object and described relative position information between respectively looking are encoded and write code stream, perhaps, directly with described difference vector coding and write code stream.
By stating description as can be known, beneficial effect of the present invention is as follows:
Owing to considered position relation between the different cameras, and determine the difference vector used in the cataloged procedure based on described position relation among the present invention, institute is so that the difference vector estimation between different the looking more accurate;
The raising of difference vector accuracy of estimation can effectively improve the effect at many views picture coding;
The raising of difference vector accuracy of estimation makes that the realization of picture coding course is more easy.
Description of drawings
The principle schematic of Fig. 1 for concerning between difference vector and object depth information and the camera position;
Fig. 2 looks for coding and with reference to concerning schematic diagram one between looking;
Fig. 3 looks for coding and with reference to concerning schematic diagram two between looking.
Embodiment
Core of the present invention provides the method for obtaining difference vector in the cataloged procedure, can be by utilizing the positional information between the camera, reach accurate Calculation each look between the purpose of difference vector.
Promptly because under the fixed-site situation of video camera, difference vector only is decided by the degree of depth (distance between space object and the video camera) and the relation of the position between the video camera of object.Therefore, core concept of the present invention is the positional information that makes full use of between the different cameras, difference vector is calculated accurately, thus can be so that the preferable coding parameter in the cataloged procedure be looked in the difference vector conduct more.
As shown in Figure 1, the A among the figure, B, C is the position of any 3 cameras (also can be called three looks) in looking more, c1, c2 and c3 are respectively these 3 cameras distance between any two.O is 3 same space objects that camera is captured.X-axis is an imaging surface.D1, d2, d3 represent that respectively object O looks at A, and B looks with C and looks between any two difference vector.
Be not difficult to find out that from Fig. 13 cameras distance between any two exists proportionate relationship between the difference vector at physics O between any two with corresponding
Though, this is the simplest proportionate relationship just, but as can be seen: exist certain fixing calculated relationship between each difference vector between looking that the position is determined by this simple proportional relation, therefore, relative position information between two the difference vectors of calculate determining by part between looking are looked in conjunction with each just can be determined other any two difference vectors between looking, thereby makes that the procurement process of difference vector is more easy.
For example, still as shown in Figure 1, the present invention can utilize the position relation between each camera, at known one of them difference vector d
iUnder the situation of (i=1,2,3), just can calculate other 2 difference vector values by described proportionate relationship.
Need to prove that based on to the analysis of Fig. 1 as can be seen, the proportionate relationship between difference vector and the camera position is equally applicable to not point-blank situation of camera.At this moment, x and cx have represented a plane, d1, what d2 and d3 represented is the absolute value of difference vector, c1, c2 and c3 then represent the absolute distance between the camera, the direction of each difference vector is then consistent with the direction that camera distributes, at this moment, and d1, d2 and d3 and c1 still exist corresponding fixing calculated relationship between c2 and the c3.Simultaneously, the front is that example describes this fixing calculated relationship to adopt three cameras only, will have corresponding fixing calculated relationship during when the number of camera and greater than 3.
The present invention can be regularly arranged or irregular alignment between described a plurality of the looking in concrete application process, no matter be regularly arranged promptly or all suitable the present invention of a plurality of application scenarioss of looking of irregular alignment.If the regularly arranged mode that adopts then advise adopting yi word pattern, cross arrangement or square formation type to arrange can certainly adopt other regularly arranged mode.
Based on above-mentioned analysis, will describe implementation method of the present invention below.
The method of obtaining difference vector in the cataloged procedure of looking provided by the invention more, after being the difference vector between definite many views are looked as two in the cataloged procedure, difference vector between looking according to described two, and known each relative position information between looking, calculate and obtain two other difference vector between looking; And, all be suitable for the present invention for regularly arranged or irregular alignment between described a plurality of the looking.
Be considered as three and be considered as example with a plurality of, three comprise that a coding to be encoded is looked with two references in looking and look, at this moment, at first, determine three any two difference vectors between looking in looking, can for coding look and two references look in any with reference to the difference vector between looking, also can be the difference vector of two references between looking; Afterwards, according to any two difference vectors between looking of determining, and three relative positions between looking, coding is looked and any one calculates with reference to the difference vector between looking in addition.
Perhaps, in above-mentioned processing procedure, also can utilize known two difference vectors between looking, and the estimated value of the difference vector of each relative position between looking between each is looked calculates, and further determines corresponding difference vector with the position that the estimated value that obtains is pointed to as starting point.To give an example below this implementation will be described:
At coding side, at first, determine the difference vector of many views between looking as two in the cataloged procedure, and the difference vector between looking according to described two, and known each relative position information between looking, calculate and obtain that coding is looked and with reference to the difference vector between looking; Then, the reference apparent place of pointing to the gained difference vector is as original position, promptly as a reference position, doing disparity estimation with reference to the mode of looking within the specific limits according to setting, for example, with the difference vector estimated value that is obtained according to the present invention is the search window central point, in the search window of pre-sizing, carry out the search of similar estimation, just may obtain the littler difference vector of matching error, obtain that a more accurate difference vector value is looked as the coding of determining and with reference to the difference vector between looking, according to this difference vector coding is looked and to be encoded, and with coding result, the coding of determining is looked and with reference to writing in the code stream behind the coding of the difference vector between looking.
In decoding end, receive code stream, resolve and obtain that definite coding is looked and with reference to the difference vector between looking; The coding that utilization is determined is looked and with reference to the difference vector between looking code stream is decoded, and the coding that obtains coding side is looked corresponding reconstructed value.
Need to prove, do at this needs under the situation of further disparity estimation, the relative position information that does not need to transmit again between respectively looking arrives decoding end, this is because of directly the coding of determining being looked and having been passed to decoding end with reference to the difference vector between looking, so do not need to go to calculate it according to positional information again.
Because the difference vector between respectively looking in looking exists certain specific calculated relationship with the relative position information of respectively looking more, therefore, look in the cataloged procedure provided by the invention more, can also utilize the depth information of known space object, and the position between each camera concerns that the difference vector between each is looked calculates; Perhaps, utilize the depth information of known space object, and the position between each camera relation, the estimated value of the difference vector between each is looked is calculated, and to point to the position with the estimated value that obtains be starting point, carries out further disparity estimation.The implementation that this part is described of will giving an example below describes:
At coding side, at first, as shown in Figure 1, determine the depth D of many views as object in the cataloged procedure, the relative position information between looking according to known each calculates and obtains that coding is looked and with reference to the difference vector between looking, for example among Fig. 1, difference vector
Then, the reference apparent place of pointing to the gained difference vector is as initial center, doing disparity estimation with reference to looking within the specific limits, obtain that a more accurate difference vector value is looked as the coding of determining and with reference to the difference vector between looking, according to this difference vector coding is looked and to be encoded, coding result, definite coding are looked and with reference to writing in the code stream behind the coding of the difference vector between looking.
In decoding end, receive code stream, resolve and obtain that definite coding is looked and with reference to the difference vector between looking; The coding that utilization is determined is looked and with reference to the difference vector between looking code stream is decoded, and the coding that obtains coding side is looked corresponding reconstructed value.
Need to prove, in above-mentioned applicating example, if do not do further disparity estimation, then need known object depth information and camera position information are delivered to decoding end, after decoding end is received depth information and camera position information, again difference vector is calculated, afterwards according to the difference vector work of further decoding; Perhaps, the difference vector information direct coding of using in coding side also can be with cataloged procedure writes in the code stream, afterwards, directly can resolve in decoding end and to obtain corresponding difference vector information, and utilize it to carry out decode operation.
The present invention is based on above-mentioned difference vector and obtain mode, a kind of multi-video coding and decoding method based on difference vector also is provided.
At coding side, at first, determine the difference vector of many views between looking as two in the cataloged procedure, and the difference vector between looking according to described two, and known each relative position information between looking, calculate and obtain two other difference vector between looking; Then, according to described two other difference vector between looking coding is looked and to be encoded, with coding result, definite two between looking difference vector and respectively look between the relative position information coding after write in the code stream.
In decoding end, receive code stream, resolve obtain definite two between looking difference vector and respectively look between relative position information; And according to the difference vector of determining and the relative position information between respectively looking calculate and determine two other corresponding difference vector between looking with coding side, utilize described two other difference vector between looking that code stream is decoded again, the coding that obtains coding side is looked corresponding reconstructed value.
The front is analyzed realization principle of the present invention, is elaborated below in conjunction with a kind of specific implementation of accompanying drawing to method of the present invention.
Still reference is shown in Figure 1, A, and B, C is the position of any 3 cameras in looking more, c1, c2 and c3 are respectively these 3 cameras distance between any two.O is 3 same space objects that camera is captured.X-axis is an imaging surface.D1, d2, d3 represent that respectively object O looks at A, and B looks with C and looks between any two difference vector.
In this specific embodiment provided by the invention, as shown in Figure 2, B is considered as current looking of will encoding, and is called coding and looks.A looks and is considered as the reference that B looks with C and looks.The image encoding that B is looked is that unit carries out with the piece, and the size of piece is M * N (M=16,8,4, N=16,8,4), and B view picture has R piece, coded sequence from left to right, from top to bottom.
The implementation procedure of utilizing the method for acquisition difference vector provided by the invention to carry out looking coding specifically comprises following treatment step more:
Step 31: coding is looked and one of them carries out the parallax valuation with reference to looking between 1 (being that A looks or C looks), obtained these 2 difference vectors 1 (being d1 or d2) between looking;
Step 32: the difference vector 1 that step 31 draws, and the relation of the position between each camera, calculate the difference vector 2 (being d2 or d1) that coding is looked and the another one reference is looked between 2 (C looks or A looks);
Step 33: the reference that step 31 gained difference vector 1 is pointed to is looked the reference that respective pixel in 1 and step 32 gained difference vector 2 point to and is looked the predicted value that the respective pixel weighting summation in 2 obtains encoding and looks;
Step 34: the original value that coding is looked deducts the predicted value that coding that step 323 obtains is looked, and the residual values that obtains encoding and look is carried out conversion, coding to residual values, writes code stream; Camera position information and difference vector 1 are encoded and write code stream.
Based on above-mentioned cataloged procedure, corresponding decode procedure specifically comprises:
Step 35: at first resolve encoding code stream, the residual values that obtains encoding and look;
Step 36: resolve encoding code stream, obtain that camera position information and coding are looked and one of them is with reference to difference vector 1 between looking 1;
Step 37: the difference vector 1 that utilizes step 36 to draw, and the relation of the position between each camera, calculate the difference vector 2 that coding is looked and the another one reference is looked between 2;
Step 38: the reference that step 36 gained difference vector 1 is pointed to is looked the reference that respective pixel in 1 and step 37 gained difference vector 2 point to and is looked the predicted value that the respective pixel weighting summation in 2 obtains encoding and looks;
The residual values that step 39 is looked coding adds the predicted value that coding that step 38 obtains is looked, the reconstructed value that obtains encoding and look.
For ease of the present invention there being further understanding, describe below in conjunction with accompanying drawing another embodiment to method provided by the invention.
Be example still with Fig. 1, A among the figure, B, C is the position of any 3 cameras in looking more, c1, c2 and c3 are respectively these 3 cameras distance between any two.O is 3 same space objects that camera is captured.X-axis is an imaging surface.D1, d2, d3 represent that respectively object O looks at A, and B looks with C and looks between any two difference vector.
In this embodiment, as shown in Figure 3, B is considered as current looking of will encoding, and is called coding and looks.A looks and is considered as the reference that B looks with C and looks, and C looks with A and is considered as reference, has finished coding, and A looks and the difference vector d3 of C between looking draws.The image encoding that B is looked is that unit carries out with the piece, and the size of piece is M * N (M=16,8,4, N=16,8,4), and B view picture has R piece.Coded sequence from left to right, from top to bottom.
Utilize the method for acquisition difference vector provided by the invention to realize that the Direct Model of looking in the coding specifically comprises following treatment step more:
Step 51: utilize 2 known references to look difference vector between A and the C, and the relation of the position between each camera, coding is looked and calculated with reference to the difference vector d1 that looks between the A;
Step 52: utilize the difference vector of 2 known references between looking, and the relation of the position between each camera, coding is looked and calculated with reference to the difference vector d2 that looks between the C;
Step 53: the reference that the described difference vector d1 that step 51 is obtained points to is looked the reference that respective pixel among the A and step 52 gained difference vector d2 point to and is looked the predicted value that the respective pixel weighting summation among the C obtains encoding and looks;
Step 54: the original value that coding is looked deducts the predicted value that the coding that obtains in the step 53 is looked, and the residual values that obtains encoding and look is carried out conversion, coding to residual values, writes code stream; Difference vector d3 between camera position information and 2 references are looked encodes and writes code stream.
With the corresponding decode procedure of above-mentioned encoding process process, specifically comprise:
Step 55: at first resolve the encoding code stream that receives, the residual values that obtains encoding and look;
Step 56: resolve described encoding code stream, obtain that camera position information and coding are looked and 2 references look between difference vector;
Step 57: the difference vector between 2 references that utilize step 56 to draw are looked, and the relation of the position between each camera, coding is looked and calculated with reference to the difference vector d1 that looks between the A;
Step 58: the difference vector between 2 references that utilize step 56 to draw are looked, and the relation of the position between each camera, coding is looked and calculated with reference to the difference vector d2 that looks between the C;
Step 59: the reference that the difference vector d1 that calculate to obtain in the step 57 is pointed to is looked the reference that respective pixel among the A and step 58 gained difference vector d2 point to and is looked the predicted value that the respective pixel weighting summation among the C obtains encoding and looks;
Step 510: the residual values that coding is looked adds the predicted value that coding that step 59 obtains is looked, the reconstructed value that obtains encoding and look.
In sum, the present invention can utilize the relative position information between each camera that is used for obtaining image to calculate well and obtain to look the difference vector information that cataloged procedure needs more, thereby improve the performance of looking coding more, and simplify the processing procedure of looking coding more.
The above; only for the preferable embodiment of the present invention, but protection scope of the present invention is not limited thereto, and anyly is familiar with those skilled in the art in the technical scope that the present invention discloses; the variation that can expect easily or replacement all should be encompassed within protection scope of the present invention.Therefore, protection scope of the present invention should be as the criterion with the protection range of claim.
Claims (18)
1, a kind ofly look the method for obtaining difference vector in the cataloged procedure, this method is applied to view more than at least three is looked like to handle, and it is characterized in that, comprises step more:
Determine that described many views are as any two first difference vectors between looking in the cataloged procedure;
According to described first difference vector, and each relative position information between looking, calculate two of obtaining that coding needs, second difference vector between looking.
2, method according to claim 1 is characterized in that, described method specifically comprises:
In a plurality of looking, determine three and look that three comprise that a coding to be encoded is looked with two references in looking and look, and determine three any two first difference vectors between looking in looking; Afterwards, according to described first difference vector, and three relative position informations between looking, by calculate obtain that coding is looked and in addition any one with reference to second difference vector between looking.
3, method according to claim 2 is characterized in that, described any two first difference vectors between looking are:
Coding is looked and two references look in any with reference to the difference vector between looking, perhaps, the difference vector between two references are looked.
4, method according to claim 2 is characterized in that, described method also comprises step:
The reference apparent place that described second difference vector is pointed to is being done disparity estimation with reference to the mode of looking according to predetermined as original position in predefined scope, obtain encoding look look with reference between new difference vector value.
5, according to claim 1,2,3 or 4 described methods, it is characterized in that, be regularly arranged or irregular alignment between described a plurality of the looking.
6, method according to claim 5 is characterized in that, describedly regularly arrangedly is: yi word pattern, cross arrangement or square formation type are arranged.
7, a kind of coding method of looking based on difference vector more, this method is applied to view more than at least three is looked like to handle, and it is characterized in that, comprising:
Determine that many views are as any two first difference vectors between looking in the cataloged procedure;
According to described first difference vector, and each relative position information between looking, calculate two of obtaining that coding needs, second difference vector between looking;
According to described second difference vector coding is looked the also output encoder result that encodes;
Write in the code stream with described coding result, described first difference vector and behind the coding of the relative position information between respectively looking, perhaps, with described coding result, and first difference vector and second difference vector coding and write code stream.
8, method according to claim 7 is characterized in that:
One is that coding is looked in the described picture of view more than at least three, and remaining is looked for reference;
Described first difference vector be described coding look and described one of them with reference to the difference vector between looking;
Described second difference vector look for coding and another one with reference to the difference vector between looking.
9, according to claim 7 or 8 described methods, it is characterized in that:
Described according to described second difference vector to coding look encode and the output encoder result be specially:
The predicted value that the respective pixel weighting summation that the reference that respective pixel that the reference that described first difference vector is pointed to is looked and described second difference vector point to is looked obtains encoding and looks;
The original value that coding is looked deducts the predicted value that described coding is looked, the residual values that obtains encoding and look, and this residual values encoded obtain the coding result looked at coding.
10, method according to claim 7 is characterized in that:
One is that coding is looked in the described picture of view more than at least three, and remaining is looked for reference;
Described first difference vector is the difference vector between wherein two references are looked;
Described second difference vector comprises resolute: the 4th difference vector and the 5th difference vector, wherein:
Described the 4th difference vector is looked for coding and the difference vector of a reference between looking;
Described the 5th difference vector looks for coding and another is with reference to the difference vector between looking.
11, according to claim 7 or 10 described methods, it is characterized in that:
Described according to described second difference vector to coding look encode and the output encoder result be specially:
The predicted value that the respective pixel weighting summation that the reference that respective pixel that the reference that described the 4th difference vector is pointed to is looked and described the 5th difference vector point to is looked obtains encoding and looks;
The original value that coding is looked deducts the predicted value that described coding is looked, the residual values that obtains encoding and look, and this residual values encoded obtain the coding result looked at coding.
12, a kind of coding/decoding method of looking based on difference vector more, this method realizes that in decoding end this method is applied to view more than at least three is looked like to handle, and it is characterized in that, comprises step:
Receive code stream, parsing obtain definite any two between looking first difference vector and respectively look between relative position information, according to described first difference vector and the relative position information between respectively looking calculate and determine and two of coding side corresponding codes needs second difference vector between looking; Perhaps, flow through journey by described decoding end at receiving code and directly obtain described second difference vector;
Utilize described second difference vector that code stream is decoded, the coding that obtains coding side is looked corresponding reconstructed value.
13, method according to claim 12 is characterized in that:
One is that coding is looked in the described picture of view more than at least three, and remaining is looked for reference;
Described first difference vector be described coding look and described one of them with reference to the difference vector between looking; Described second difference vector look for coding and another one with reference to the difference vector between looking;
And describedly utilize described second difference vector that code stream is decoded, the coding that obtains coding side is looked corresponding reconstructed value and is specially:
The predicted value that the respective pixel weighting summation that the reference that respective pixel that the reference that described first difference vector is pointed to is looked and described second difference vector point to is looked obtains encoding and looks;
The residual values that the coding of decoding acquisition is looked adds the predicted value that described coding is looked, the reconstructed value that obtains encoding and look.
14, method according to claim 12 is characterized in that:
One is that coding is looked in the described picture of view more than at least three, and remaining is looked for reference;
Described second difference vector comprises resolute: the 4th difference vector and the 5th difference vector, wherein:
Described the 4th difference vector is looked for coding and the difference vector of a reference between looking;
Described the 5th difference vector looks for coding and another is with reference to the difference vector between looking.
15, according to claim 12 or 14 described methods, it is characterized in that: describedly utilize described second difference vector that code stream is decoded, the coding that obtains coding side is looked corresponding reconstructed value and is specially:
The predicted value that the respective pixel weighting summation that the reference that respective pixel that the reference that described the 4th difference vector is pointed to is looked and described the 5th difference vector point to is looked obtains encoding and looks;
The residual values that the coding of decoding acquisition is looked adds the predicted value that described coding is looked, the reconstructed value that obtains encoding and look.
16, a kind ofly look the method for obtaining difference vector in the cataloged procedure, this method is applied to view more than at least three is looked like to handle, and it is characterized in that, comprises step more:
Determine the depth information of space object, and described each relative position information between looking;
Utilize the depth information of described space object and each relative position information between looking to calculate to obtain coding to need two difference vectors between looking.
17, method according to claim 16 is characterized in that, described method also comprises step:
Pointing to the position with described difference vector is original position, is doing disparity estimation with reference to the mode of looking according to predetermined in predefined scope, obtain encoding look look with reference between new difference vector value.
18, a kind of coding method of looking based on difference vector more, this method is applied to view more than at least three is looked like to handle, and it is characterized in that, comprises step:
Determine the depth information of space object, and described each relative position information between looking;
Utilize the depth information of described space object and described each relative position information between looking to calculate to obtain coding to need two difference vectors between looking;
Utilizing described difference vector that coding is looked encodes and writes code stream; Simultaneously, also the depth information of space object and described relative position information between respectively looking are encoded and write code stream, perhaps, directly with described difference vector coding and write code stream.
Priority Applications (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2006100709719A CN101056398A (en) | 2006-03-29 | 2006-03-29 | A method and decoding and encoding method for capturing the video difference vector in the multi-video coding process |
| PCT/CN2007/001030 WO2007110000A1 (en) | 2006-03-29 | 2007-03-29 | A method and device of obtaining disparity vector and its multi-view encoding-decoding |
| US12/239,226 US8594193B2 (en) | 2006-03-29 | 2008-09-26 | Method and device for multi-view encoding/decoding using disparity vector |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNA2006100709719A CN101056398A (en) | 2006-03-29 | 2006-03-29 | A method and decoding and encoding method for capturing the video difference vector in the multi-video coding process |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN101056398A true CN101056398A (en) | 2007-10-17 |
Family
ID=38795970
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNA2006100709719A Pending CN101056398A (en) | 2006-03-29 | 2006-03-29 | A method and decoding and encoding method for capturing the video difference vector in the multi-video coding process |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101056398A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2009089785A1 (en) * | 2008-01-11 | 2009-07-23 | Huawei Technologies Co., Ltd. | Image processing method, encoding/decoding method and apparatus |
| CN101483770B (en) * | 2008-01-08 | 2011-03-16 | 华为技术有限公司 | Method and apparatus for encoding and decoding |
| WO2013075515A1 (en) * | 2011-11-23 | 2013-05-30 | 华为技术有限公司 | Coding and decoding method, device, encoder, and decoder for multi-view video |
| WO2014005548A1 (en) * | 2012-07-05 | 2014-01-09 | Mediatek Inc. | Method and apparatus of unified disparity vector derivation for 3d video coding |
| WO2014106346A1 (en) * | 2013-01-07 | 2014-07-10 | Mediatek Singapore Pte. Ltd. | Method of signalling additional collocated picture for 3dvc |
| CN105308970A (en) * | 2013-04-05 | 2016-02-03 | 三星电子株式会社 | Method and apparatus for encoding and decoding video with respect to position of integer pixel |
| CN105340273A (en) * | 2013-04-05 | 2016-02-17 | 三星电子株式会社 | Method for predicting disparity vector for interlayer video decoding and encoding apparatus and method |
| CN105453561A (en) * | 2013-08-13 | 2016-03-30 | 联发科技股份有限公司 | Method of deriving default disparity vector in 3D and multiview video coding |
| CN106210740A (en) * | 2014-06-04 | 2016-12-07 | 寰发股份有限公司 | Use method and the device thereof of encoder motion prediction depth data between view |
| US9729849B2 (en) | 2012-04-19 | 2017-08-08 | Intel Corporation | 3D video coding including depth based disparity vector calibration |
-
2006
- 2006-03-29 CN CNA2006100709719A patent/CN101056398A/en active Pending
Cited By (21)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101483770B (en) * | 2008-01-08 | 2011-03-16 | 华为技术有限公司 | Method and apparatus for encoding and decoding |
| CN101483765B (en) * | 2008-01-11 | 2011-09-21 | 华为技术有限公司 | Image processing method, method and apparatus for encoding and decoding |
| WO2009089785A1 (en) * | 2008-01-11 | 2009-07-23 | Huawei Technologies Co., Ltd. | Image processing method, encoding/decoding method and apparatus |
| CN103139569B (en) * | 2011-11-23 | 2016-08-10 | 华为技术有限公司 | The coding of multi-view point video, coding/decoding method, device and codec |
| WO2013075515A1 (en) * | 2011-11-23 | 2013-05-30 | 华为技术有限公司 | Coding and decoding method, device, encoder, and decoder for multi-view video |
| CN103139569A (en) * | 2011-11-23 | 2013-06-05 | 华为技术有限公司 | Multi-viewpoint video coding and decoding method and device and coder and decoder |
| US9467676B2 (en) | 2011-11-23 | 2016-10-11 | Huawei Technologies Co., Ltd. | Multi-view video coding and decoding methods and apparatuses, coder, and decoder |
| US9860514B2 (en) | 2012-04-19 | 2018-01-02 | Intel Corporation | 3D video coding including depth based disparity vector calibration |
| US9729849B2 (en) | 2012-04-19 | 2017-08-08 | Intel Corporation | 3D video coding including depth based disparity vector calibration |
| WO2014005548A1 (en) * | 2012-07-05 | 2014-01-09 | Mediatek Inc. | Method and apparatus of unified disparity vector derivation for 3d video coding |
| US9843820B2 (en) | 2012-07-05 | 2017-12-12 | Mediatek Inc | Method and apparatus of unified disparity vector derivation for 3D video coding |
| WO2014106346A1 (en) * | 2013-01-07 | 2014-07-10 | Mediatek Singapore Pte. Ltd. | Method of signalling additional collocated picture for 3dvc |
| CN105340273A (en) * | 2013-04-05 | 2016-02-17 | 三星电子株式会社 | Method for predicting disparity vector for interlayer video decoding and encoding apparatus and method |
| CN105308970A (en) * | 2013-04-05 | 2016-02-03 | 三星电子株式会社 | Method and apparatus for encoding and decoding video with respect to position of integer pixel |
| US9894377B2 (en) | 2013-04-05 | 2018-02-13 | Samsung Electronics Co., Ltd. | Method for predicting disparity vector for interlayer video decoding and encoding apparatus and method |
| CN105340273B (en) * | 2013-04-05 | 2018-09-11 | 三星电子株式会社 | For predicting for the method for the decoded difference vector of cross-layer video and coding method and equipment |
| CN105308970B (en) * | 2013-04-05 | 2018-11-23 | 三星电子株式会社 | The method and apparatus that video is coded and decoded for the position of integer pixel |
| US10469866B2 (en) | 2013-04-05 | 2019-11-05 | Samsung Electronics Co., Ltd. | Method and apparatus for encoding and decoding video with respect to position of integer pixel |
| CN105453561A (en) * | 2013-08-13 | 2016-03-30 | 联发科技股份有限公司 | Method of deriving default disparity vector in 3D and multiview video coding |
| CN105453561B (en) * | 2013-08-13 | 2018-10-26 | 寰发股份有限公司 | The method of export acquiescence disparity vector in three-dimensional and multi-view video coding |
| CN106210740A (en) * | 2014-06-04 | 2016-12-07 | 寰发股份有限公司 | Use method and the device thereof of encoder motion prediction depth data between view |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101056398A (en) | A method and decoding and encoding method for capturing the video difference vector in the multi-video coding process | |
| US8594193B2 (en) | Method and device for multi-view encoding/decoding using disparity vector | |
| TWI500308B (en) | A method and an apparatus for encoding or decoding a video signal | |
| CN1497981A (en) | Method and device for code and decode motion vector | |
| CN1968419A (en) | Image encoding method and apparatus and image decoding method and apparatus using characteristics of the human visual system | |
| CN1874519A (en) | Intraprediction method and apparatus using video symmetry and encoding and decoding method and apparatus | |
| CN1882085A (en) | Method and apparatus for encoding video signal with improved compression efficiency | |
| CN1929613A (en) | Apparatus and method for image encoding and decoding and recording medium | |
| CN1742488A (en) | Method and apparatus for encoding and decoding stereoscopic video | |
| CN1675848A (en) | Method and apparatus for performing high quality fast predictive motion search | |
| JP2009509407A (en) | Naturally shaped regions for motion compensation | |
| CN1236624C (en) | Quick full pixel movement evaluation method of multiple kinds of modules | |
| CN1738423A (en) | Method for controlling video code bit rate | |
| CN1885948A (en) | Motion vector space prediction method for video coding | |
| CN1756365A (en) | Method and apparatus based on Code And Decode data in the frame mode of multiple scaaning | |
| CN100341334C (en) | Multi-reference frame rapid movement estimation method based on effective coverage | |
| CN1144469C (en) | Motion-vector coding method | |
| CN1285211C (en) | Device and method for enhancing reproduction image quality | |
| CN1767653A (en) | Bit rate control method | |
| CN1595986A (en) | Method and apparatus for 3-d subband video coding | |
| CN1568015A (en) | Multiple visual-angle video coding-decoding prediction compensation method and apparatus | |
| CN101076830A (en) | Data processing apparatus and method | |
| CN1320830C (en) | Noise estimating method and equipment, and method and equipment for coding video by it | |
| CN101056408A (en) | Method and apparatus for determining a threshold for skip mode in video coding | |
| CN101043633A (en) | Quantization matrix selecting method, apparatus of coding and coding/decoding method and system |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C12 | Rejection of a patent application after its publication | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20071017 |