CN104539954B - A kind of Cascading Methods of acceleration HEVC coding rates - Google Patents
A kind of Cascading Methods of acceleration HEVC coding rates Download PDFInfo
- Publication number
- CN104539954B CN104539954B CN201410719598.XA CN201410719598A CN104539954B CN 104539954 B CN104539954 B CN 104539954B CN 201410719598 A CN201410719598 A CN 201410719598A CN 104539954 B CN104539954 B CN 104539954B
- Authority
- CN
- China
- Prior art keywords
- patterns
- hevc
- avc
- pattern
- skip
- 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
Abstract
The present invention discloses a kind of Cascading Methods of acceleration HEVC coding rates, and this method extracts the macro block division result of video first by H.264/AVC precoding is carried out to present frame and HEVC reconstruction frames from pre-encoding results；H.264 then macro block division result is mapped as to HEVC possibility set of patterns by mode map；Rate-distortion optimization search HEVC optimal mode is finally used in the set.Compared with prior art, the present invention innovatively proposes the method for a cascaded encoder to accelerate HEVC cataloged procedures, can greatly simplifie HEVC pattern search process, have good application prospect.
Description
Technical field
The present invention relates to a kind of method that HEVC patterns quickly select field, specifically a kind of Cascade H .264 and HEVC are compiled
Code device come accelerate HEVC encode method.
Background technology
HEVC is the abbreviation of efficient video coding standard, is the video encoding standard of new generation after H.264/AVC, by
Integration and cooperation Video coding group (JCT-VC, the Joint of ITU-T VCEG and ISO/IEC MPEG tissue compositions
Collaborative Team on Video Coding) formulate, issued at the beginning of 2013.HEVC employs block-based mixed
Close coding framework, its mainly include infra-frame prediction, inter prediction, conversion and quantization, entropy code, inverse quantization and inverse transformation and
The technologies such as loop filtering.Compared with traditional coding framework, HEVC employs flexible quadtree coding block division, bulk prediction
With the new technology such as conversion, polytype prediction block and transform block, adaptive motion parameter coding and adaptive loop filter.
These new coding techniques are effectively improved HEVC coding efficiency, but have also been introduced great amount of calculation, are not suitable for real
The application on border.
In HEVC mode selection processes, be exhaustive all possible coding unit, the prediction of the model method of standard
The combination of unit and converter unit, and minimum rate distortion is calculated to select optimal coding mode.This method consumes
Substantial amounts of calculate in those last proofs is not in optimal pattern, so in order to promote the key of HEVC development and practical application
Place, improve its mode selecting method on the premise of the performance of HEVC Video codings is kept and be necessary.It is general next
The improvement for saying model selection is come real using existing time between coding unit, predicting unit and change of scale and spatial redundancy
Existing.The presence of this redundancy has crypticity, it is necessary to constantly go to excavate, and the Optimal improvements of mode selecting method for
HEVC's should actually possess important realistic meaning.
The content of the invention
In view of the shortcomings of the prior art, the present invention provides a kind of Cascading Methods of acceleration HEVC coding rates, using brand-new
Encoder framework to solve the quick select permeability of HEVC patterns, by the method for mode map by H.264/AVC
Macro block division result combines with HEVC possible CU (coding unit) model selection collection, is asked for HEVC fast mode decision
Topic provides a kind of brand-new thinking.
To achieve the above object, the present invention uses following technical scheme：The present invention devises a kind of frame of new encoder
Frame pattern is used for accelerating HEVC coding rate, i.e.,：Present frame and HEVC reconstruction frames are carried out first by H.264/AVC pre-
Coding, the macro block division result of video is extracted from pre-encoding results；Then macro block H.264 is divided by mode map
As a result it is mapped as HEVC possibility set of patterns；Rate-distortion optimization search HEVC optimal mode is finally used in the set.
The Cascading Methods of acceleration HEVC coding rates of the present invention, the method includes the steps of：
Step 1, precoding is carried out to present frame, obtains optimal macro block division；
Precoding is carried out to input frame first with H.264 encoder and obtains the optimal macro block division result of the frame.Pre-
In cataloged procedure, for the framework of encoder using hybrid encoding frame H.264/AVC, input parameter is present frame and HEVC weight
Frame is built, HEVC reconstruction frames will be divided by the reference frame as present frame for estimation and macro block.
H.264/AVC optimal macro block division is mapped as HEVC sets of patterns by step 2.
H.264/AVC the result that the optimal macro block obtained divides is mapped as to HEVC optimal possibility by the method for mapping
Set of patterns.For different size of CU, different mode map strategies is taken to select most suitable CU sets of patterns；
× 64 layer B.64
For 64 × 64 CU, using only SKIP and 2N × 2N predictive mode, because experiment proves to be used only this two kinds
Pattern is little for the performance impact of encoder, but the very big complexity for reducing encoder.
C.32x32 layer
For 32 × 32 layers, CU is divided into five kinds of different types by the type according to four MB in the CU, for every kind of
The CU of type, different mapping relations are all devised to ensure that the possibility set of patterns of mapping is optimal.Five kinds of CU pattern is right with its
The mapping answered is as follows：
A) when being MB (macro block) of Inter16 × 16 comprising 3 or 4 patterns in current CU, only detect current CU's
Inter2N × 2N patterns.
B) when two MB in left side pattern is Inter16 × 16 in current CU, current CU Inter2N × 2N is only detected
With Inter2N × N patterns.
C) when two MB in the right pattern is Inter16 × 16 in current CU, current CU Inter2N × 2N is only detected
With InterN × 2N patterns.
D) when the pattern of four MB in current CU is Intra16 × 16, current CU Intra2N × 2N moulds are only detected
Formula.
E) for other situations, all patterns will be all detected to judge the most suitable predictive modes of current CU.
D.16 × 16 and 8 × 8 layers
For 16 × 16 and 8 × 8 CU, the present invention takes different mapping relations to obtain most suitable PU collection.Wherein
For 16 × 16 and 8 × 8 CU, following mapping relations are taken respectively：
A) CU for 16 × 16, SKIP mode maps H.264/AVC are HEVC SKIP/Merge patterns；16×16
Mode map is 2N × 2N and SKIP/Merge patterns；H.264/AVC 16 × 8 and 8 × 16 correspond to HEVC SKIP/ respectively
Merge, 2N × 2N and corresponding left and right/upper and lower rectangle and asymmetric partition mode；It will be detected for 8 × 8, HEVC all
Pattern except Intra and N × N patterns, for the Intra patterns in H.264/AVC, HEVC will detect all patterns.
B) CU for 8 × 8, SKIP mode maps H.264/AVC are HEVC SKIP/Merge patterns；16 × 16 moulds
Formula is mapped as 2N × 2N and SKIP/Merge patterns；H.264/AVC 16 × 8 and 8 × 16 correspond to HEVC SKIP/ respectively
Merge, 2N × 2N and corresponding left and right/upper and lower rectangle and asymmetric partition mode；It will be detected for 8 × 8, HEVC all
Pattern except Intra patterns, for the Intra patterns in H.264/AVC, HEVC will detect all patterns.
Step 3. carrys out Ergodic Maps different size CU PU sets of patterns using HEVC encoders, is obtained using rate-distortion optimization
Optimal CU and PU patterns.
Compared with prior art, the present invention has following beneficial effect：
The present invention obtains the macro block division result of present frame using precoding is H.264/AVC carried out；Pass through mode map
Method division result H.264/AVC is mapped as to HEVC CU sets of patterns；Finally utilize HEVC rate-distortion optimization slave patterns
Concentrate and select optimal CU and PU (predicting unit) pattern.The needs that the present invention reduces HEVC by the method for mode map are searched
The set of patterns of rope, accelerates HEVC coding rate, while make use of the H.264/AVC encoder of maturation, has good answer
Use prospect.
Brief description of the drawings
The detailed description made by reading with reference to the following drawings to non-limiting example, further feature of the invention,
Objects and advantages will become more apparent upon：
Fig. 1 is four kinds of different CU (32 × 32) type schematic diagrames；
Fig. 2 is to use to be used for the cascade encoder block diagram H.264/AVC for accelerating HEVC coding rates；
Fig. 3 is the illustration in implementation process of the present invention.
Embodiment
With reference to specific embodiment, the present invention is described in detail.Following examples will be helpful to the technology of this area
Personnel further understand the present invention, but the invention is not limited in any way.It should be pointed out that the ordinary skill to this area
For personnel, without departing from the inventive concept of the premise, various modifications and improvements can be made.These belong to the present invention
Protection domain.
Accelerate the application of HEVC coding rates to describe here in connection with the present invention, specifically say the cascade that the present invention is built
The framework of encoder, the H.264/AVC mode map between HEVC are applied in HEVC cataloged procedure, and specific block diagram is as schemed
Shown in 1：The step of introducing first using precoding is H.264/AVC carried out, then different CU will be discussed in detail on basis herein
The mode map method of size, the system of selection of last HEVC optimization models.
Step 1, precoding is carried out to present frame, obtains optimal macro block division；
HEVC and coding framework H.264/AVC are hybrid encoding frame, and H.264/AVC by development for many years it
After have been obtained for significant progress, many outstanding work have been done in terms of codec optimization.So in the present invention
In we first with H.264 encoder to input frame carry out precoding obtain the optimal macro block division result of the frame.Prelisting
During code, for the framework of encoder using hybrid encoding frame H.264/AVC, input parameter is present frame and HEVC reconstruction
Frame, HEVC reconstruction frames will be by the reference frame as present frame.
H.264/AVC optimal macro block division is mapped as HEVC sets of patterns by step 2.；
HEVC, using the code tree unit (CTU) based on adative quadtree structure, instead of to improve code efficiency
H.264 the coding unit based on macro block.In HEVC, CTU size is 64 × 64, and can adaptively iteratively
Four points are carried out, until SCU (smallest coding unit), SCU size is 8 × 8.And in H.264/AVC, macro block
(MB) size can be 16 × 16,8 × 8,4 × 4.In H.264/AVC, inter prediction encoding only supports 7 kinds of macroblock partitions
Pattern：16 × 16,16 × 8,8 × 16,8 × 8,8 × 4,4 × 8 and 4 × 4, and HEVC not only supports these symmetrically to split mould
Formula, referred to as 2N × 2N, 2N × N and N × 2N, but also asymmetrical Fractionation regimen is supported, it is referred to as：2N×nU、2N×nD、nL×
2N and nR × 2N.Compared to the coding unit for H.264 fixing macroblock size, divisions of the HEVC to coded image is more flexible, so as to
More preferable video image quality is provided, this otherness also directly results in the high complexity for H.264 arriving HEVC transcodings.
In the present invention, the result that the optimal macro block H.264/AVC obtained divides is mapped as pair by the method for mapping
In the possible optimization model collection of HEVC, for the CU of different layers, different mapping methods is selected to obtain optimal result；
× 64 layer A.64
For 64 × 64 CU, SKIP and 2N × 2N predictive mode is used only in we, because experiment proves that this is used only
Both of which is little for the performance impact of encoder, but the very big complexity for reducing encoder.
B.32x32 layer
For 32 × 32 layers, CU is divided into five kinds of different types by us according to the type of four MB in the CU, for
Each type of CU, we can have different mapping relations to ensure the optimal of effect.Five kinds of CU pattern and its corresponding to
Map as follows：
A) when being the MB of Inter16 × 16 comprising 3 or 4 patterns in current CU, we only detect current CU's
Inter2N × 2N patterns.
B) when two MB in left side pattern is Inter16 × 16 in current CU, we only detect current CU Inter2N
× 2N and Inter2N × N patterns.
C) when two MB in the right pattern is Inter16 × 16 in current CU, we only detect current CU Inter2N
× 2N and InterN × 2N patterns.
D) when the pattern of four MB in current CU is Intra16 × 16, we only detect current CU Intra2N ×
2N patterns.
E) for other situations, all patterns will be all detected to judge the most suitable predictive modes of current CU.
C.16 × 16 and 8 × 8 layers
For 16 × 16 and 8 × 8 CU, the present invention takes different mapping relations to obtain most suitable PU collection.Specifically
Mapped mode as shown in table one and table two：
Table one：Mapping table (16 × 16) between MB and CU
Table two：Mapping table (8 × 8) between MB and CU
Step 3. carrys out Ergodic Maps different size after HEVC optimal possible set of patterns is obtained, using HEVC encoders
CU PU sets of patterns, the rate distortion costs of different mode are calculated, optimal CU and PU patterns are determined using rate-distortion optimization.
Following table three is operation result of the present invention in X265.X265 be current HEVC it is exsertile increase income encoder it
One, x265 support HEVC standard and deep optimization have been carried out in many aspects completely.Simultaneous selection x264 is as H.264/AVC
Encoder carries out the precoding of present frame, and X264 is the classic H.264/AVC encoder of increasing income of current performance.
Cycle tests collection is handled using the mode map method of the cascaded encoder of the present invention, experimental result shows this hair
Bright method can map out optimal possibility set of patterns, also there is good lifting in efficiency, accelerate HEVC coding
Speed.Table three gives the method for the present invention for the performance of different cycle tests and the change of time.
Table three：Experimental result based on X265 and X264
, can be effectively using macro block division result H.264/AVC, and by macro block when the inventive method is used for HEVC and encoded
Division result is mapped as the possibility set of patterns of the CU model selections available for HEVC, reduces calculative CU and PU patterns, reaches
Purpose is encoded to HEVC is accelerated.
Described above is only the preferred embodiment of the present invention, and protection scope of the present invention is not only limited to above-mentioned implementation
Example, all technical schemes belonged under thinking of the present invention belong to the protection category of the present invention.It should be pointed out that for the art
Technical staff for, some improvements and modifications without departing from the principles of the present invention, these improvements and modifications also all should
It is considered as protection scope of the present invention.
Claims (1)
1. a kind of Cascading Methods of acceleration HEVC coding rates, it is characterised in that methods described comprises the following steps：
Step 1, precoding is carried out to input frame, obtains optimal macro block division result；
Precoding is carried out to input frame first with H.264 encoder and obtains the optimal macro block division result of the frame, in precoding
During, for the framework of encoder using hybrid encoding frame H.264/AVC, input parameter is present frame and HEVC reconstruction
Frame, HEVC reconstruction frames will be by the reference frame as present frame；
Step 2, optimal macro block division H.264/AVC is mapped as HEVC sets of patterns；
H.264/AVC the result that the optimal macro block obtained divides is mapped as to HEVC optimal possible pattern by the method for mapping
Collection；For different size of CU, different mapped modes is taken respectively to obtain HEVC set of patterns respectively：
× 64 layer A.64
For 64 × 64 CU, using only SKIP and 2N × 2N predictive mode；
B.32x32 layer
For 32 × 32 CU, CU is divided into five kinds of different types by the type according to four MB in the CU, for every species
The CU of type, different mapping relations are designed to ensure that the set mapped out best suits the feature of the CU；
C.16 × 16 and 8 × 8 layers
For 16 × 16 and 8 × 8 CU, different mapping relations are taken respectively to obtain optimal PU sets of patterns；
Step 3, Ergodic Maps different size CU PU sets of patterns are carried out using HEVC encoders, are obtained using rate-distortion optimization optimal
CU and PU patterns；
In B described in step 2：Five kinds of CU pattern and mapping corresponding to it are as follows：
A) when being the MB of Inter16 × 16 comprising 3 or 4 patterns in current CU, current CU Inter2N × 2N is only detected
Pattern；
B) when two MB in left side pattern is Inter16 × 16 in current CU, only detect current CU Inter2N × 2N and
Inter2N × N patterns；
C) when two MB in the right pattern is Inter16 × 16 in current CU, only detect current CU Inter2N × 2N and
InterN × 2N patterns；
D) when the pattern of four MB in current CU is Intra16 × 16, current CU Intra2N × 2N patterns are only detected；
E) for other situations, all patterns will be all detected to judge the most suitable predictive modes of current CU；
In C described in step 2：
For 16 × 16 and 8 × 8 CU, different mapping relations are taken to obtain most suitable PU collection, specific mapped mode
For：
A) CU for 16 × 16, SKIP mode maps H.264/AVC are HEVC SKIP/Merge patterns；16 × 16 patterns
It is mapped as 2N × 2N and SKIP/Merge patterns；H.264/AVC 16 × 8 mode maps are HEVC SKIP/Merge, 2Nx2N
The dissymmetric mode up and down of pattern, 2NxN pattern symmetrical above and below, 2NxnU and 2NxnD compositions；H.264/AVC 8 × 16 patterns are reflected
Penetrate the left-right asymmetry of SKIP/Merge, 2Nx2N pattern for HEVC, the symmetrical patterns of Nx2N, nRx2N and nLx2N compositions
Pattern；All patterns will be detected except Intra and 16 × 16 patterns for 8 × 8, HEVC, in H.264/AVC
Intra patterns, HEVC will detect all patterns；
B) CU for 8 × 8, SKIP mode maps H.264/AVC are HEVC SKIP/Merge patterns；16 × 16 patterns are reflected
Penetrate as 2N × 2N and SKIP/Merge patterns；H.264/AVC 16 × 8 mode maps are HEVC SKIP/Merge, 2Nx2N mould
The dissymmetric mode up and down of formula, 2NxN pattern symmetrical above and below, 2NxnU and 2NxnD compositions；H.264/AVC 8 × 16 mode maps
For the left-right asymmetry mould of HEVC SKIP/Merge, 2Nx2N pattern, the symmetrical patterns of Nx2N, nRx2N and nLx2N composition
Formula；All patterns will be detected except Intra patterns for 8 × 8, HEVC, for the Intra patterns in H.264/AVC,
HEVC will detect all patterns.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410719598.XA CN104539954B (en) | 2014-12-01 | 2014-12-01 | A kind of Cascading Methods of acceleration HEVC coding rates |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410719598.XA CN104539954B (en) | 2014-12-01 | 2014-12-01 | A kind of Cascading Methods of acceleration HEVC coding rates |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN104539954A CN104539954A (en) | 2015-04-22 |
| CN104539954B true CN104539954B (en) | 2017-12-15 |
Family
ID=52855408
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410719598.XA Active CN104539954B (en) | 2014-12-01 | 2014-12-01 | A kind of Cascading Methods of acceleration HEVC coding rates |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104539954B (en) |
Families Citing this family (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105872564B (en) * | 2016-05-27 | 2019-02-19 | 武汉拓宝科技股份有限公司 | H.264/AVC the intra prediction mode determining method and system of H.265/HEVC transcoding are arrived |
| CN108012154B (en) * | 2016-10-28 | 2020-07-03 | 北京金山云网络技术有限公司 | Video coding method and device |
| CN107690069B (en) * | 2017-08-28 | 2021-01-01 | 中国科学院深圳先进技术研究院 | Data-driven cascade video coding method |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103248893A (en) * | 2013-05-10 | 2013-08-14 | 四川大学 | Method for quick inter-frame transcoding from H. 264/AVC standard to HEVC standard and transcoder thereof |
| CN103442228A (en) * | 2013-08-19 | 2013-12-11 | 四川大学 | Quick frame inner transcoding method from H.264/AVC standard to HEVC standard and transcoder thereof |
| CN103475880A (en) * | 2013-09-11 | 2013-12-25 | 浙江大学 | Method for low-complexity video transcoding from H.264 to HEVC based on statistic analysis |
| CN103491334A (en) * | 2013-09-11 | 2014-01-01 | 浙江大学 | Video transcode method from H264 to HEVC based on region feature analysis |
| CN103796023A (en) * | 2014-03-05 | 2014-05-14 | 武汉拓宝电子系统有限公司 | H.265/HEVC transcoding method and transcoder based on H.264/AVC macro block structure and texture |
-
2014
- 2014-12-01 CN CN201410719598.XA patent/CN104539954B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103248893A (en) * | 2013-05-10 | 2013-08-14 | 四川大学 | Method for quick inter-frame transcoding from H. 264/AVC standard to HEVC standard and transcoder thereof |
| CN103442228A (en) * | 2013-08-19 | 2013-12-11 | 四川大学 | Quick frame inner transcoding method from H.264/AVC standard to HEVC standard and transcoder thereof |
| CN103475880A (en) * | 2013-09-11 | 2013-12-25 | 浙江大学 | Method for low-complexity video transcoding from H.264 to HEVC based on statistic analysis |
| CN103491334A (en) * | 2013-09-11 | 2014-01-01 | 浙江大学 | Video transcode method from H264 to HEVC based on region feature analysis |
| CN103796023A (en) * | 2014-03-05 | 2014-05-14 | 武汉拓宝电子系统有限公司 | H.265/HEVC transcoding method and transcoder based on H.264/AVC macro block structure and texture |
Also Published As
| Publication number | Publication date |
|---|---|
| CN104539954A (en) | 2015-04-22 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN104038764B (en) | A kind of H.264 arrive video transcoding method H.265 and transcoder | |
| CN106254882B (en) | The method for generating quantization block | |
| CN107979756A (en) | Method for video coding and device used in a kind of video coding system | |
| CN103096067B (en) | The method for producing reconstructed blocks | |
| CN104023234B (en) | Fast inter-frame prediction method applicable to high efficiency video coding (HEVC) | |
| CN104967847B (en) | The coding/decoding method of video data | |
| CN103763570B (en) | Rapid HEVC intra-frame prediction method based on SATD | |
| CN109792516A (en) | Method and device in image and coding and decoding video for coloration encoding and decoding in frame | |
| CN104754357B (en) | Intraframe coding optimization method and device based on convolutional neural networks | |
| CN108347616A (en) | A kind of depth prediction approach and device based on optional time domain motion-vector prediction | |
| CN104125469B (en) | A kind of fast encoding method for HEVC | |
| CN109845261A (en) | In image and coding and decoding video in frame coloration encoding and decoding method and device | |
| CN110087087A (en) | VVC interframe encode unit prediction mode shifts to an earlier date decision and block divides and shifts to an earlier date terminating method | |
| CN103561263A (en) | Motion compensation prediction method based on motion vector restraint and weighting motion vector | |
| CN107360419B (en) | A kind of movement forward sight video interprediction encoding method based on perspective model | |
| CN104023233A (en) | Fast inter-frame prediction method of HEVC (High Efficiency Video Coding) | |
| CN104639940B (en) | A kind of quick HEVC method for choosing frame inner forecast mode | |
| CN103348681A (en) | Method and device for determining reference unit | |
| CN104539970B (en) | A kind of 3D HEVC interframe encodes merging patterns high-speed decision method | |
| CN105120290B (en) | A kind of deep video fast encoding method | |
| CN104539954B (en) | A kind of Cascading Methods of acceleration HEVC coding rates | |
| CN107888929A (en) | Video coding coding/decoding method, equipment and generation and the method for stored bits stream | |
| CN105681808A (en) | Rapid decision-making method for SCC interframe coding unit mode | |
| CN104811729B (en) | A kind of video multi-reference frame coding method | |
| CN104883566B (en) | The fast algorithm that a kind of intra prediction block size suitable for HEVC standard divides |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |