CN102427530B - Intra prediction method - Google Patents
Intra prediction method Download PDFInfo
- Publication number
- CN102427530B CN102427530B CN 201110306265 CN201110306265A CN102427530B CN 102427530 B CN102427530 B CN 102427530B CN 201110306265 CN201110306265 CN 201110306265 CN 201110306265 A CN201110306265 A CN 201110306265A CN 102427530 B CN102427530 B CN 102427530B
- Authority
- CN
- China
- Prior art keywords
- intra
- matrix
- predictive mode
- mode
- intra prediction
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Abstract
The invention discloses an intra prediction method, which comprises the following steps of: (1) defining N templates in different directions corresponding to N novel intra prediction modes, wherein N is an arbitrary value between 1 and 9, and searching m image blocks which are most similar to the current image blocks to be predicted in a coded image region in a template matching mode; (2) rearranging the most similar m image blocks which are searched in each template to construct a low-rank observation matrix; (3) performing low-rank matrix filling on observation matrixes obtained in N different prediction modes, and reconstructing unknown element values in the observation matrixes to obtain predicted values of the N current image blocks; (4) performing intra prediction on the current image blocks to be predicted by a standard intra prediction method in H.264; and (5) selecting an optimized predicted value from the predicted values obtained in the steps (3) and (4) through rate-distortion optimization, namely selecting an optimized predicted value from (N+1) predicted values. By the method, spatial correlation in an image can be more effectively utilized.
Description
Technical field
The invention belongs to field of video encoding, relate to a kind of intra-frame prediction method.
Background technology
As the video coding international standard of latest generation, H.264/AVC introduced such as effective means such as multi-direction infra-frame predictions.Spatial redundancy in the intracoded frame of multi-direction infra-frame prediction in reducing image or video sequence plays an important role.H.264/AVC 4 * 4 luminance block and 8 * 8 luminance block are adopted 9 kinds of intra prediction modes, and 16 * 16 luminance block and 8 * 8 chrominance block are adopted 4 kinds of intra prediction modes.Prediction draws according to their encoded adjacent image pieces for the corresponding specific direction of each predictive mode, predicted picture piece.There are 8 kinds with the predictive mode of direction, as shown in Figure 1 in 9 kinds of intra prediction modes for 4 * 4 image blocks.There is not the predictive mode 2 of expression among Fig. 1 namely not with the DC predictive mode of direction.The predicted value of whole 4 * 4 image blocks is the mean value of its adjacent pixel values in the DC predictive mode, and the DC predictive mode is more effective to the smoothed image piece.Infra-frame prediction with direction obtains predicted value by the adjacent pixel values that will encode along the specific direction extrapolation, and this is more effective to the image block that the simple edges profile is arranged.Yet this infra-frame prediction mode based on extrapolation is not utilized global space correlation in the frame, and from the pixel common forecasting inaccuracy of the boundary pixel of having encoded away from.
Many correlative studys have launched to improve infra-frame prediction efficient." Intra Prediction by Template Matching ", Thiow Keng Tan, Choong Seng Boon and Yoshinori Suzuki; 2006 IEEE International Conference on Image Processing, a kind of intra-frame prediction method based on template matches has been proposed among the pp.1693-1696, by take the adjacent encoded pixels of current image block as template in coding region the predicted value of the search encoded image piece acquisition current image block the most similar to it." Intra Prediction by Averaged Template Matching Predictors ", Thiow Keng Tan, Choong Seng Boon and Yoshinori Suzuki; 2007 IEEE Consumer Communications and Networking Conference have done further improvement by the mode that will be averaged by the predicted value that multi-direction and template different size obtain among the pp.405-409 on this basis.At " Improved is Intra Coding Based on Bi-directional Intra Prediction H.264; Directional Transform; and Adaptive Coefficient Scanning ", Yan Ye and M.Karczewicz, 2008 IEEE International Conference on Image Processing, among the pp.2116-2119 by extra bidirectional frame inner estimation mode with the direction transformation of some predictive mode has been improved the effect of infra-frame prediction." Sparse Approximation with Adaptive Dictionary for Image Prediction ", M.Turkan and C.Guillemot, 2009 IEEE International Conference on Image Processing, pp.25-28 has proposed a kind ofly to utilize the method for one group of redundant dictionary to solve the infra-frame prediction problem based on rarefaction representation, and this coding for rest image provides a kind of new thinking.
The matrix fill-in problem is used for recovering unknown element from the sample of a small amount of known elements of an imperfect matrix.This is an ill-conditioning problem, because known sample is far fewer than matrix element, solution is not unique.Yet if matrix to be recovered is low-rank or approximate low-rank, matrix can pass through to separate the minimum optimization problem Exact Reconstruction of order of following NP difficulty:
min rank(A)，s.t.A
ij＝M
ij，(i，j)∈Ω
Wherein A is matrix to be rebuild, and M is observing matrix and M ∈ R
N * n, Ω is the subset that comprises the index of m known observation element.The minimum optimization problem of above-mentioned order can replace with following optimization problem:
min||A||
*，s.t.A
ij＝M
ij，(i，j)∈Ω
Wherein || A||
*Be the nuclear norm of matrix A, the singular value sum of representing matrix.It is because nuclear norm is convex function that nuclear norm is used for the alternative of rank of matrix, can effectively find the solution by the mode of positive semidefinite planning.The optimization problem of rank of matrix minimum that Candes etc. are verified and the protruding optimization problem of nuclear norm minimum have identical unique solution.Recently make it obtain application at numerous areas such as commending system, recognition of face and video denoising in the progress that solves on the low-rank matrix fill-in problem.Meanwhile, the efficient algorithm of the minimum optimization problem of many solution orders such as SVT and FPCA have also proposed.
Intra-frame prediction method in the conventional images Video coding can only utilize the local space correlation, can not effectively utilize the holistic correlation in the image/video frame, therefore can not fully remove spatial redundancy.In addition, existing intra-frame prediction method is to from the pixel value common forecasting inaccuracy of the boundary pixel of having encoded away from.
Summary of the invention
Technology of the present invention is dealt with problems and is: overcome the deficiency that existing infra-prediction techniques exists, a kind of intra-frame prediction method based on the low-rank matrix fill-in is provided.
Of the present invention opening one's minds is: finish infra-frame prediction with the low-rank matrix fill-in, be about to predict in the infra-frame prediction that the unknown pixel value solves as the problem of rebuilding the unknown matrix element in the low-rank matrix fill-in.Because image block has overall similitude in image or the frame of video, in coding region the search image block similar to current image block to be predicted and it is rearranged after obtain the observing matrix of a low-rank or approximate low-rank.Utilize the theoretical element value of rebuilding the unknown in the observing matrix of low-rank matrix fill-in, namely obtain the predicted value to current image block.For compatible with existing coding standard when increasing new intra prediction mode, and then design a kind of new intra prediction mode coding method with the required bit of minimizing coded prediction mode number, thereby reduced code check.
Technical solution of the present invention is: intra-frame prediction method may further comprise the steps:
(1) template of N different directions of definition, corresponding N new intra prediction mode, N is the random natural number between 1～9; Adopt the mode of template matches to search for m the image block the most similar to current image block to be predicted in image coding region, m is the natural number less than 20;
(2) m the most similar image block that each template is searched rearranges, the observing matrix of a low-rank of structure;
(3) observing matrix that N different predictive modes is obtained carries out respectively the low-rank matrix fill-in, rebuilds the unknown element value in the observing matrix, obtains N current image block predicted value;
(4) using H.264, the Plays intra-frame prediction method carries out infra-frame prediction to current image block to be predicted;
(5) the predicted value that the predicted value that obtains from step (3) and step (4) obtain, adopt rate-distortion optimization to select an optimum prediction value; Namely from N+1 predicted value, select an optimum prediction value.
Further, step (1) is specially:
Represent current image block with B, use B
sExpression comprises the larger image block of B and adjacent boundary pixel thereof, B
sSize be n * n, B
sIn only have border pixel values known, the unknown pixel value is value to be predicted; Search obtains m image block the most similar to B, these similar image pieces B as template to utilize the boundary pixel of having encoded in image coding region
iExpression, i=1 ..., m, B
sWith B
iSize is identical.
Further, step (2) is specially:
M image block launches by row or by row, with each image block B
iA n is classified in rearrangement as
2* 1 vector form, it is expressed as
N then
2* (m+1) matrix M is expressed as:
M＝(V
s，V
1，V
2，...，V
m)
V wherein
sExpression is from image block B
sThe vector that obtains.
Further, step (3) is rebuild unknown element value among the observing matrix M by the minimum optimization problem of the nuclear norm of finding the solution following belt restraining:
min||A||
*，s.t.A+E＝M，P
Ω(E)＝0 (1)
Wherein A represents matrix to be rebuild, || A||
*Be the nuclear norm of matrix A, the singular value sum of representing matrix, E represents the compensation to unknown element among the M, Ω has comprised the index of all known elements in the matrix M, P
ΩBe illustrated in the rectangular projection operator in the matrix scope; If (i, j) ∈ is Ω, P then
Ω(X) element of (i, j) position equals X in
Ij, otherwise be zero.
Further, adopt the objective function optimization problem of non-accurate augmentation Lagrange multiplier Algorithm for Solving formula (1), the part Augmented Lagrangian Functions that obtains formula (1) is:
Wherein Y is penalty factor, and μ is arithmetic number, || X||
FBe the F norm of matrix X,
X
*The adjoint matrix of representing matrix X; Classify the reconstructed value of unknown element among the A as image block according to the corresponded manner rearrangement, namely obtain the predicted value to current image block.
Further, described intra-frame prediction method is further comprising the steps of:
In being incorporated into infra-frame prediction H.264/AVC in 9 kinds of intra prediction modes of 4 * 4 image blocks, the N that step (3) is obtained new predictive mode is as the subpattern of former DC predictive mode, that is: originally remained unchanged with 8 intra prediction modes of direction and former DC pattern is replaced by a new mixed mode, wherein new mixed mode comprises N predictive mode and the original DC predictive mode that increases newly; Perhaps
In being incorporated into infra-frame prediction H.264/AVC in 9 kinds of intra prediction modes of 8 * 8 image blocks, the N that step (3) is obtained new predictive mode is as the subpattern of former DC predictive mode, that is: original 8 intra prediction modes except the DC predictive mode remain unchanged and former DC pattern by a new mixed mode replacement, wherein new mixed mode comprises N predictive mode and the original DC predictive mode that increases newly; Perhaps
In being incorporated into infra-frame prediction H.264/AVC in 4 kinds of intra prediction modes of 16 * 16 image blocks, the N that step (3) is obtained new predictive mode is as the subpattern of former DC predictive mode, that is: original 3 intra prediction modes except the DC predictive mode remain unchanged and former DC pattern by a new mixed mode replacement, wherein new mixed mode comprises N predictive mode and the original DC predictive mode that increases newly.
Preferably, N is 7 in the step (1).
The bit number that original 8 intra prediction modes except the DC predictive mode are encoded is 3, and be 6 to the bit number that new mixed mode is encoded, wherein front 3 bits represent the DC predictive mode in 9 kinds of intra prediction modes of 4 * 4 image blocks, and rear 3 bits represent 8 subpatterns; Perhaps
The bit number that original 8 intra prediction modes except the DC predictive mode are encoded is 3, and be 6 to the bit number that new mixed mode is encoded, wherein front 3 bits represent the DC predictive mode in 9 kinds of intra prediction modes of 8 * 8 image blocks, and rear 3 bits represent 8 subpatterns; Perhaps
The bit number that original 3 intra prediction modes except the DC predictive mode are encoded is 2, and be 5 to the bit number that new mixed mode is encoded, wherein front 2 bits represent the DC predictive mode in 4 kinds of intra prediction modes of 16 * 16 image blocks, and rear 3 bits represent 8 subpatterns.
The present invention compared with prior art has following advantage: the present invention proposes to utilize the mode of low-rank matrix fill-in to do infra-frame prediction.If image block is through certain permutatation and reorganization, they can be expressed as the matrix of a low-rank or approximate low-rank.Unknown predicted value can be considered as the element lost in the low-rank matrix.Theoretical according to the low-rank matrix fill-in, predicted value can be by separating mode Exact Reconstruction from limited sample of the minimum optimization problem of order." Exact Matrix Completion via Convex Optimization; " E.J.Candes and B.Recht, The 46th Annual Allerton Conference on Communication, Control and Computing, low-rank characteristic that pp.806-812 is verified has guaranteed and can obtain Exact Reconstruction with high probability.With standard H.264/AVC in the infra-frame prediction of extrapolation compare, the infra-frame prediction mode based on the low-rank matrix fill-in that the present invention proposes can more effectively be utilized the spatial coherence in the image.In the image current block and the strong spatial coherence of coding region can provide more fully than the boundary pixel of corresponding specific direction that information is used for infra-frame prediction.
Description of drawings
Below with reference to accompanying drawing the specific embodiment of the present invention is described.
Fig. 1 is that 4 * 4 image blocks are adjacent encoded pixels and 8 intra prediction modes with direction.
Fig. 2 is according to intra-frame prediction method general frame figure of the present invention.
Fig. 3 is the template of 7 different directions.
Fig. 4 is the observing matrix of structure.
Fig. 5 is the distribution of optimum prediction subpattern in Foreman (QCIF) the sequence mixed mode.
Fig. 6 is the distribution of optimum prediction subpattern in Bus (QCIF) the sequence mixed mode.
Fig. 7 is the R-D curve of Foreman (CIF) video sequence.
Fig. 8 is the R-D curve of Akiyo (CIF) video sequence.
Embodiment
The general frame that proposes a plan as shown in Figure 2.Behind the search image block similar to current image block to be predicted, it is an observing matrix that these image blocks can reorganize in coding region.Most elements in the observing matrix is come own coding region, and the pixel in the corresponding current image block of unknown element wherein.Unknown pixel is rebuild to lay equal stress on by the mode of low-rank matrix fill-in and is arranged as the form of image block, thereby has formed the prediction to current image block.The predicted value and the tradition that are finally obtained by aforesaid way are selected the optimum prediction value based on the predicted value that the infra-frame prediction of extrapolating obtains with rate-distortion optimization (RDO).
Intra-frame prediction method provided by the invention may further comprise the steps:
(1) template of N different directions of definition, corresponding N new intra prediction mode, N is preferably 7, and it is that 7, N also can be the random natural number between 1～9 that present embodiment is got N; Adopt the mode of template matches in image coding region, to search for m the image block the most similar to current image block to be predicted.
The m value is generally the natural number less than 20, and it is 5 that present embodiment is got m.
Represent current image block with B, use B
sExpression comprises the larger image block of B and adjacent boundary pixel thereof, B
sSize be n * n, B
sIn only have border pixel values known, the unknown pixel value is value to be predicted; Search obtains m image block the most similar to B, these similar image pieces B as template to utilize the boundary pixel of having encoded in image coding region
iExpression, i=1 ..., m, B
sWith B
iSize is identical.
(2) m the most similar image block that each template is searched rearranges, the observing matrix of a low-rank of structure.
M image block launches by row or by row, with each image block B
iA n is classified in rearrangement as
2* 1 vector form, it is expressed as
N then
2* (m+1) matrix M is expressed as:
M＝(V
s，V
1，V
2，...，V
m)
V wherein
sExpression is from image block B
sThe vector that obtains.
(3) observing matrix that 7 different predictive modes is obtained carries out respectively the low-rank matrix fill-in, rebuilds the unknown element value in the observing matrix, obtains 7 current image block predicted values.
Rebuild unknown element value among the observing matrix M by the minimum optimization problem of the nuclear norm of finding the solution following belt restraining:
min||A||
*，s.t.A+E＝M，P
Ω(E)＝0 (1)
Wherein A represents matrix to be rebuild, || A||
*Be the nuclear norm of matrix A, the singular value sum of representing matrix, E represents the compensation to unknown element among the M, Ω has comprised the index of all known elements in the matrix M, P
ΩBe illustrated in the rectangular projection operator in the matrix scope; If (i, j) ∈ is Ω, P then
Ω(X) element of (i, j) position equals X in
Ij, otherwise be zero.
Adopt the objective function optimization problem of non-accurate augmentation Lagrange multiplier Algorithm for Solving formula (1), the part Augmented Lagrangian Functions that obtains formula (1) is:
Wherein Y is penalty factor, and μ is arithmetic number, || X||
FBe the F norm of matrix X,
X
*The adjoint matrix of representing matrix X; Classify the reconstructed value of unknown element among the A as image block according to the corresponded manner rearrangement, namely obtain the predicted value to current image block.
(4) using H.264, the Plays intra-frame prediction method carries out infra-frame prediction to current image block to be predicted.
This step is traditional predicted value that obtains based on the infra-frame prediction of direction prediction.
(5) the predicted value that the predicted value that obtains from step (3) and step (4) obtain, adopt rate-distortion optimization to select an optimum prediction value; Namely from 7+1 predicted value, select an optimum prediction value.
The present invention further verifies the performance of the intra-frame prediction method that proposes.To be incorporated into based on the intra-frame prediction method of low-rank matrix fill-in among the reference software JM 11.0 H.264/AVC.Coding parameter arranges as follows:
● class: main class
● test video sequence form: QCIF and CIF
● video sequence coding mode: full I frame
● entropy encryption algorithm: CABAC
● rate-distortion optimization: open
● quantization parameter value: 2227323742
In being incorporated into infra-frame prediction H.264/AVC in 9 kinds of intra prediction modes of 4 * 4 image blocks, 7 new predictive modes that step (3) is obtained are as the subpattern of former DC predictive mode, that is: originally remained unchanged with 8 intra prediction modes of direction and former DC pattern is replaced by a new mixed mode, wherein new mixed mode comprises 7 predictive modes and the original DC predictive mode that increases newly.
The bit number that original 8 intra prediction modes except the DC predictive mode are encoded is 3, and be 6 to the bit number that new mixed mode is encoded, wherein front 3 bits represent the DC predictive mode in 9 kinds of intra prediction modes of 4 * 4 image blocks, and rear 3 bits represent 8 subpatterns.
In the infra-frame prediction of 4 * 4 of brightness, implement, test various video sequence with the validity of verification algorithm.The distribution of optimum subpattern as shown in Figure 5 and Figure 6 in the mixed mode 2 of some video sequence.On average having 50% image block can finally select the subpattern based on the infra-frame prediction of low-rank matrix fill-in to mixed mode is optimization model.This has proved the validity based on the intra-frame prediction method of low-rank matrix fill-in that proposes.
H.264/AVC, the decoded frame that the present invention obtains obtains better visual quality than standard, especially under low code check.The present invention and standard H.264/AVC for the contrast of the reconstructed image objective quality of QCIF and CIF video sequence respectively shown in Table I and Table II, the code check that wherein reduces is to calculate according to BD-Rate.Can find out with standard therefrom and H.264/AVC compare that algorithm of the present invention is on average having 4.21% code check saving to the QCIF video sequence under the reconstruction quality on an equal basis, the CIF video sequence is on average had 5.39% code check saving.Scheme to Foreman (CIF) and the proposition of Akiyo (CIF) video sequence can obtain respectively to save up to 15.85% and 6.16% code check.Corresponding R-D curve as shown in Figure 7 and Figure 8.
Table I the present invention and JM11.0 compare the code check of QCIF video test sequence
Table II the present invention and JM11.0 compare the code check of CIF video test sequence
In addition, in being incorporated into infra-frame prediction H.264/AVC in 9 kinds of intra prediction modes of 8 * 8 image blocks, 7 new predictive modes that step (3) is obtained are as the subpattern of former DC predictive mode, that is: original 8 intra prediction modes except the DC predictive mode remain unchanged and former DC pattern by a new mixed mode replacement, wherein new mixed mode comprises 7 predictive modes and the original DC predictive mode that increases newly.
The bit number that original 8 intra prediction modes except the DC predictive mode are encoded is 3, and be 6 to the bit number that new mixed mode is encoded, wherein front 3 bits represent the DC predictive mode in 9 kinds of intra prediction modes of 8 * 8 image blocks, and rear 3 bits represent 8 subpatterns.
Have again, in being incorporated into infra-frame prediction H.264/AVC in 4 kinds of intra prediction modes of 16 * 16 image blocks, 7 new predictive modes that step (3) is obtained are as the subpattern of former DC predictive mode, that is: original 3 intra prediction modes except the DC predictive mode remain unchanged and former DC pattern by a new mixed mode replacement, wherein new mixed mode comprises 7 predictive modes and the original DC predictive mode that increases newly.
The bit number that original 3 intra prediction modes except the DC predictive mode are encoded is 2, and be 5 to the bit number that new mixed mode is encoded, wherein front 2 bits represent the DC predictive mode in 4 kinds of intra prediction modes of 16 * 16 image blocks, and rear 3 bits represent 8 subpatterns.
The present invention is not limited to the content that claim and above-described embodiment are addressed, so long as any invention of creating out according to design of the present invention all should belong within protection scope of the present invention.
Claims (9)
1. intra-frame prediction method is characterized in that, may further comprise the steps:
(1) template of N different directions of definition, corresponding N new intra prediction mode, N is the random natural number between 1～9; Adopt the mode of template matches to search for m the image block the most similar to current image block to be predicted in image coding region, m is the natural number less than 20;
(2) m the most similar image block that each template is searched rearranges, the observing matrix of a low-rank of structure;
(3) observing matrix that N different predictive modes is obtained carries out respectively the low-rank matrix fill-in, rebuilds the unknown element value in the observing matrix, obtains N current image block predicted value;
(4) using H.264, the Plays intra-frame prediction method carries out infra-frame prediction to current image block to be predicted;
(5) the predicted value that the predicted value that obtains from step (3) and step (4) obtain, adopt rate-distortion optimization to select an optimum prediction value; Namely from N+1 predicted value, select an optimum prediction value.
2. intra-frame prediction method according to claim 1 is characterized in that, step (1) is specially:
Represent current image block with B, use B
sExpression comprises the larger image block of B and adjacent boundary pixel thereof, B
sSize be n * n, B
sIn only have border pixel values known, the unknown pixel value is value to be predicted; Search obtains m image block the most similar to B, these similar image pieces B as template to utilize the boundary pixel of having encoded in image coding region
iExpression, i=1 ..., m, B
sWith B
iSize is identical.
3. intra-frame prediction method according to claim 2 is characterized in that, step (2) is specially:
M image block launches by row or by row, with each image block B
iA n is classified in rearrangement as
2* 1 vector form, it is expressed as
N then
2* (m+1) matrix M is expressed as:
M＝(V
s，V
1，V
2，...，V
m)
V wherein
sExpression is from image block B
sThe vector that obtains.
4. intra-frame prediction method according to claim 3 is characterized in that, step (3) is rebuild unknown element value among the observing matrix M by the minimum optimization problem of the nuclear norm of finding the solution following belt restraining:
min||A||
*，s.t.A+E＝M，P
Ω(E)＝0 (1)
Wherein A represents matrix to be rebuild, || A||
*Be the nuclear norm of matrix A, the singular value sum of representing matrix, E represents the compensation to unknown element among the M, Ω has comprised the index of all known elements in the matrix M, P
ΩBe illustrated in the rectangular projection operator in the matrix scope; If (i, j) ∈ is Ω, P then
Ω(X) element of (i, j) position equals X in
Ij, otherwise be zero.
5. intra-frame prediction method according to claim 4 is characterized in that, adopts the objective function optimization problem of non-accurate augmentation Lagrange multiplier Algorithm for Solving formula (1), and the part Augmented Lagrangian Functions that obtains formula (1) is:
Wherein Y is penalty factor, and μ is arithmetic number, || X||
FBe the F norm of matrix X,
X
*The adjoint matrix of representing matrix X; Classify the reconstructed value of unknown element among the A as image block according to the corresponded manner rearrangement, namely obtain the predicted value to current image block.
6. intra-frame prediction method according to claim 1 is characterized in that, and is further comprising the steps of:
In being incorporated into infra-frame prediction H.264/AVC in 9 kinds of intra prediction modes of 4 * 4 image blocks, the N that step (3) is obtained new predictive mode is as the subpattern of former DC predictive mode, that is: originally remained unchanged with 8 intra prediction modes of direction and former DC pattern is replaced by a new mixed mode, wherein new mixed mode comprises N predictive mode and the original DC predictive mode that increases newly; Perhaps
In being incorporated into infra-frame prediction H.264/AVC in 9 kinds of intra prediction modes of 8 * 8 image blocks, the N that step (3) is obtained new predictive mode is as the subpattern of former DC predictive mode, that is: original 8 intra prediction modes except the DC predictive mode remain unchanged and former DC pattern by a new mixed mode replacement, wherein new mixed mode comprises N predictive mode and the original DC predictive mode that increases newly; Perhaps
In being incorporated into infra-frame prediction H.264/AVC in 4 kinds of intra prediction modes of 16 * 16 image blocks, the N that step (3) is obtained new predictive mode is as the subpattern of former DC predictive mode, that is: original 3 intra prediction modes except the DC predictive mode remain unchanged and former DC pattern by a new mixed mode replacement, wherein new mixed mode comprises N predictive mode and the original DC predictive mode that increases newly.
7. intra-frame prediction method according to claim 6 is characterized in that, N is 7 in the step (1).
8. intra-frame prediction method according to claim 7, it is characterized in that, the bit number that original 8 intra prediction modes except the DC predictive mode are encoded is 3, and be 6 to the bit number that new mixed mode is encoded, wherein front 3 bits represent the DC predictive mode in 9 kinds of intra prediction modes of 4 * 4 image blocks, and rear 3 bits represent 8 subpatterns; Perhaps
The bit number that original 8 intra prediction modes except the DC predictive mode are encoded is 3, and be 6 to the bit number that new mixed mode is encoded, wherein front 3 bits represent the DC predictive mode in 9 kinds of intra prediction modes of 8 * 8 image blocks, and rear 3 bits represent 8 subpatterns; Perhaps
The bit number that original 3 intra prediction modes except the DC predictive mode are encoded is 2, and be 5 to the bit number that new mixed mode is encoded, wherein front 2 bits represent the DC predictive mode in 4 kinds of intra prediction modes of 16 * 16 image blocks, and rear 3 bits represent 8 subpatterns.
9. intra-frame prediction method according to claim 1 is characterized in that, m is the random natural number between 5～10 in the step (1).
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110306265 CN102427530B (en) | 2011-10-11 | 2011-10-11 | Intra prediction method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 201110306265 CN102427530B (en) | 2011-10-11 | 2011-10-11 | Intra prediction method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN102427530A CN102427530A (en) | 2012-04-25 |
| CN102427530B true CN102427530B (en) | 2013-10-23 |
Family
ID=45961468
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 201110306265 Active CN102427530B (en) | 2011-10-11 | 2011-10-11 | Intra prediction method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN102427530B (en) |
Families Citing this family (15)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103903291B (en) * | 2012-12-24 | 2017-05-31 | 阿里巴巴集团控股有限公司 | A kind of method and apparatus of automatic modification picture |
| CN105338351B (en) * | 2014-05-28 | 2019-11-12 | 华为技术有限公司 | Intra prediction coding and decoding, array scanning method and device based on template matching |
| US9800641B2 (en) * | 2015-05-04 | 2017-10-24 | Google Inc. | Pre-fetched encoding for application streaming |
| US20170374369A1 (en) * | 2016-06-24 | 2017-12-28 | Mediatek Inc. | Methods and Apparatuses of Decoder Side Intra Mode Derivation |
| CN107155108B (en) * | 2017-06-19 | 2019-07-12 | 电子科技大学 | A kind of intra-frame prediction method based on brightness change |
| CN111201795B (en) * | 2017-10-09 | 2022-07-26 | 华为技术有限公司 | Memory access window and padding for motion vector modification |
| AU2020256658A1 (en) | 2019-04-12 | 2021-10-28 | Beijing Bytedance Network Technology Co., Ltd. | Most probable mode list construction for matrix-based intra prediction |
| WO2020211807A1 (en) | 2019-04-16 | 2020-10-22 | Beijing Bytedance Network Technology Co., Ltd. | Matrix derivation in intra coding mode |
| WO2020221373A1 (en) | 2019-05-01 | 2020-11-05 | Beijing Bytedance Network Technology Co., Ltd. | Matrix-based intra prediction using filtering |
| WO2020221372A1 (en) | 2019-05-01 | 2020-11-05 | Beijing Bytedance Network Technology Co., Ltd. | Context coding for matrix-based intra prediction |
| WO2020233663A1 (en) | 2019-05-22 | 2020-11-26 | Beijing Bytedance Network Technology Co., Ltd. | Matrix-based intra prediction using upsampling |
| CN114051735A (en) | 2019-05-31 | 2022-02-15 | 北京字节跳动网络技术有限公司 | One-step downsampling process in matrix-based intra prediction |
| EP3963885A4 (en) | 2019-06-05 | 2022-12-14 | Beijing Bytedance Network Technology Co., Ltd. | Context determination for matrix-based intra prediction |
| EP3957075A4 (en) * | 2019-06-17 | 2022-11-02 | Zhejiang Dahua Technology Co., Ltd. | Systems and methods for predicting a coding block |
| KR20220082847A (en) | 2019-10-28 | 2022-06-17 | 베이징 바이트댄스 네트워크 테크놀로지 컴퍼니, 리미티드 | Syntax signaling and parsing based on color components |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1615020A (en) * | 2004-11-10 | 2005-05-11 | 华中科技大学 | Method for pridicting sortable complex in frame |
| CN1662066A (en) * | 2004-02-26 | 2005-08-31 | 中国科学院计算技术研究所 | Method for selecting predicting mode within frame |
| CN101222635A (en) * | 2008-01-24 | 2008-07-16 | 北京工业大学 | Infra-frame prediction method |
| EP2046053A1 (en) * | 2007-10-05 | 2009-04-08 | Thomson Licensing | Method and device for adaptively quantizing parameters for image coding |
-
2011
- 2011-10-11 CN CN 201110306265 patent/CN102427530B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1662066A (en) * | 2004-02-26 | 2005-08-31 | 中国科学院计算技术研究所 | Method for selecting predicting mode within frame |
| CN1615020A (en) * | 2004-11-10 | 2005-05-11 | 华中科技大学 | Method for pridicting sortable complex in frame |
| EP2046053A1 (en) * | 2007-10-05 | 2009-04-08 | Thomson Licensing | Method and device for adaptively quantizing parameters for image coding |
| CN101222635A (en) * | 2008-01-24 | 2008-07-16 | 北京工业大学 | Infra-frame prediction method |
Non-Patent Citations (2)
| Title |
|---|
| 《快速帧内预测方法》;尹宝才等;《北京工业大学学报》;20091031;第35卷(第10期);1423-1427 * |
| 尹宝才等.《快速帧内预测方法》.《北京工业大学学报》.2009,第35卷(第10期),1423-1427. |
Also Published As
| Publication number | Publication date |
|---|---|
| CN102427530A (en) | 2012-04-25 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN102427530B (en) | Intra prediction method | |
| CN104244007B (en) | Image coding method and device and decoding method and device | |
| CN104918056B (en) | The method decoded to intra prediction mode | |
| TW202007166A (en) | Method of video decoding at a decoder and apparatus thereof | |
| JP7085009B2 (en) | Methods and devices for merging multi-sign bit concealment and residual sign prediction | |
| CN110024393A (en) | Low complex degree sign prediction for video coding | |
| CN103202018B (en) | The Video coding that uses the data based on sample to prune | |
| DE112010004109B4 (en) | Method for decoding a bit stream | |
| Tabus et al. | Lossy compression of lenslet images from plenoptic cameras combining sparse predictive coding and JPEG 2000 | |
| CN108141597A (en) | For having the non-separable quadratic transformation of the video coding of recombination | |
| CN104429067B (en) | By predicting that the division of current block carries out method, coding/decoding method, coding and decoding device and its corresponding medium of Video coding | |
| CN107027025B (en) | A kind of light field image compression method based on macro block of pixels adaptive prediction | |
| CN108848387A (en) | The method for deriving reference prediction mode value | |
| CN103338370B (en) | A kind of multi-view depth video fast encoding method | |
| CN104704826B (en) | Two steps quantify and coding method and device | |
| CN103581647A (en) | Depth map sequence fractal coding method based on motion vectors of color video | |
| CN104780379A (en) | Compression method for screen image set | |
| Fang et al. | 3dac: Learning attribute compression for point clouds | |
| Liu et al. | Prior-based quantization bin matching for cloud storage of JPEG images | |
| WO2014058796A1 (en) | Method and apparatus for video coding using reference motion vectors | |
| CN103747272B (en) | Fast transform approach for the remaining quaternary tree coding of HEVC | |
| CN103763558A (en) | Texture image compression method based on image similarities | |
| CN105872549B (en) | Video transcoding method based on block search and orthogonal matching pursuit | |
| US20140029667A1 (en) | Method of coding an image epitome | |
| CN103544717B (en) | A kind of two-stage three dimensional image processing coded method based on SIFT feature |
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 |