CN104581152A - HEVC intra-frame prediction mode decision accelerating method - Google Patents
HEVC intra-frame prediction mode decision accelerating method Download PDFInfo
- Publication number
- CN104581152A CN104581152A CN201410842187.XA CN201410842187A CN104581152A CN 104581152 A CN104581152 A CN 104581152A CN 201410842187 A CN201410842187 A CN 201410842187A CN 104581152 A CN104581152 A CN 104581152A
- Authority
- CN
- China
- Prior art keywords
- rmd
- alternative mode
- mode
- mpm
- rdo
- 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
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses an HEVC intra-frame prediction mode selection accelerating method which comprises the following steps: firstly, speculating the texture consistence of an adjacent PU and a present PU in a space, if the characteristic of texture consistence is achieved, in a prediction unit that a first mode decided through RMD is in MPM, skipping the RDO process, and directly setting the mode as an RDO optimal mode; secondly, according to the texture characteristics reflected in two previous modes decided through RMD, in the PU, further respectively performing mode decision acceleration on the prediction unit that the first mode decided through RMD is not in MPM according to different situations when judging whether the two previous modes decided through RMD are both DC or Planar modes or both angle modes or not. By adopting the HEVC intra-frame prediction mode selection accelerating method, the texture consistence of the prediction unit and an adjacent encoded prediction unit, and the texture characteristics of the prediction units, which are reflected through RMD and MPM, are sufficiently revealed, and the method has the advantages that the complexity of the RDO prediction mode decision process in intra-frame mode decision is effectively reduced, and the like.
Description
Technical field
The present invention relates to high-performance video coding (High Efficiency Video Coding, HEVC) field, particularly the technology of intra mode decision in coding.
Background technology
Digital video has become a part indispensable in our daily life.Along with the fast development of network and video capture technology, ultra high-definition Video service has become the development trend of future digital video.In order to meet the demand of ultra high-definition Video service better, dynamic image expert group (Moving Picture Experts Group, MPEG) and Video coding combine group (JointCollaborative Team on Video Coding, JCT-VC) combine and be proposed new video compression standard---high efficiency Video coding (High Efficiency Video Coding, HEVC).As ITU-T successor H.264/MPEG-4AVC, HEVC have employed many new coding techniquess, can save the code word of half when identical video quality relative to predecessor.Owing to have employed many new technology, the complexity of encoder increases greatly, makes the realization of real-time coding become difficulty all the more like this, so the complexity reducing encoder is the important research field of HEVC.
The number of intra prediction direction, in order to improve the compression performance of intraframe coding, is increased to 35,0 and 1 patterns and is respectively DC and Planar pattern by HEVC, and remaining 2 ~ 35 pattern is angle mould formula, this considerably increases the computation complexity of intra mode decision.In order to reduce the complexity of this part, the technology that existing HEVC takes is: for a predicting unit (PredictionUnit, PU), what first carry out is simple thick model selection (Rough Mode Decision, RMD).RMD adopts Hadamard transform (Hadamard Transform) to calculate the rate distortion costs of each pattern.RMD by 35 kinds of patterns according to the ascending arrangement of rate distortion costs.Wherein, encoder can select RDO alternative mode according to the size of PU, for 64x64,32x32, the PU unit of 16x16 size, RMD selects 3 patterns of Least-cost as RDO alternative mode, for the PU unit of 8x8 and 4x4 size, RMD selects 8 patterns of Least-cost as RDO alternative mode, and these alternative mode will be stored in alternate list.After RMD terminates, most possible pattern (Most Probable Mode, MPM) optimization model that the PU in above current PU and left side has selected will be checked whether in alternative mode list, if the optimization model of top or left side PU is not in alternative mode list, then added in list, specific practice can list of references 1 (see L.Zhao, L.Zhang, S.Ma, and D.Zhao, " Fast mode decision algorithm for intra prediction in HEVC, " inProc.in Proc.VCIP 2011, pp.1-4.).At ensuing rate-distortion optimization (Rate-distortion optimization, RDO) stage will adopt the discrete cosine/sine transform (Discrete Cosine/Sine Transform, DCT/DST) of encoding and using to all alternative mode calculation rate distortion costs.Namely the minimum pattern of RDO stage rate distortion costs is final optimization model.After determining optimization model, residual error quadtree conversion (Residual Quadtree transform, RQT) can carry out transition coding according to the optimization model selected to PU.Except the existing fast intra mode selection technique of HEVC, also there is some other method.Such as document 2 is (see M.Zhang, C.Zhao, and J.Xu, " An adaptive fast intra mode decision inHEVC; " in Proc.ICIP 2012, pp.221-224.) infer the textural characteristics of current prediction block in the alternative mode distribution selected by RMD.This technology is to 16x16, PU and 8x8 of 32x32 size, the PU of 4x4 size needs two kinds of different processing schemes, and need front four alternative mode to RMD selects to analyze for the PU of 8x8 and 4x4 size, and by judging whether to there is DC or Planar pattern, and first alternative angle mould formula and the whether adjacent textural characteristics of inferring this PU of the second alternative angle mould formula, and the using texture homogeneity of adjacent PU that the relation between this technology alternative mode of not utilizing RMD to select and MPM reflects.
Summary of the invention
The object of the present invention is to provide a kind of HEVC Intra prediction mode selection accelerated method.
For achieving the above object, solution of the present invention is: first emphasis considers the using texture homogeneity of the adjacent PU in space and current PU.Utilize this feature of using texture homogeneity, for the predicting unit of first mode in MPM that RMD chooses, skip RDO process, directly arranging this pattern is RDO optimization model.Secondly, textural characteristics in the PU that the first two pattern that recycling RMD chooses reflects, by judging whether that the first two pattern that RMD chooses is all DC or Planar pattern, or be all angle mould formula, the predicting unit of first mode not in MPM that RMD chooses is divided into flat block further, has the texture block of certain orientation, and the complex texture block three kinds of situations without specific direction carry out model selection acceleration respectively.
The present invention has fully excavated the texture features in the using texture homogeneity of predicting unit that RMD and MPM reflect and adjacent encoded predicting unit and predicting unit, effectively can reduce the complexity of RDO predictive mode selection course in intra mode decision.
The textural characteristics of the adjacent PU of the current prediction unit that the present invention utilizes MPM to reflect, and the textural characteristics of current PU that first alternative mode selecting of RMD reflects, combine MPM with RMD and infer current PU whether texture is consistent with adjacent PU.Conforming rule definition is: if first alternative mode is in MPM, so just thinks that current PU and adjacent PU are that texture is consistent.Have the predicting unit of using texture homogeneity for adjacent PU, RMD first alternative mode is using the optimization model as current prediction unit RDO.
For being judged as the predicting unit inconsistent with adjacent PU texture according to above-mentioned conformance criteria, the first two alternative mode utilizing RMD to select is inferred current PU textural characteristics by the present invention, and predicting unit is divided three classes according to following criterion:
(1) if the first and second alternative mode of current prediction unit RMD gained are all DC or Planar pattern, then this PU is flat block;
(2) if the first and second alternative mode of current prediction unit RMD gained are all angle mould formula, then this PU is the texture block with certain orientation;
(3) if the first and second alternative mode of current prediction unit RMD gained are not all DC or Planar pattern, be also not all angle mould formula, then this PU is the complex texture block not having specific direction.
For first kind predicting unit, the alternative mode of its RDO only includes DC and Planar pattern, and other patterns can be removed from alternate list.For Equations of The Second Kind predicting unit, the alternative mode of its RDO only includes first three alternative mode and the MPM of RMD gained, and other pattern can be removed from alternate list.For the 3rd class predicting unit, the alternative mode of its RDO only includes the first two alternative mode and the MPM of RMD gained, and other patterns can be removed from alternate list.For this reason, the implementation step of technical solution of the present invention is:
Step 1: RMD and MPM carrying out current prediction unit.
Step 2: check the first alternative mode of selecting of RMD whether in MPM.If the first alternative mode that RMD selects is in MPM, then this prediction has similar textural characteristics with adjacent encoded unit, directly this pattern is set to RDO optimization model, skips RDO process.
Step 3: the predicting unit of the first alternative mode not in MPM selected for RMD, checks whether the first two alternative mode that RMD selects is all DC or Planar pattern.If the first two alternative mode that RMD selects is all DC or Planar pattern, then this prediction is flat block, and its RDO only checks DC and Planar two alternative mode, and other angle mould formulas can be removed from alternative mode list.
Step 3: the first alternative mode selected for RMD is not in MPM, and the first two alternative mode that RMD selects is not all the predicting unit of DC or Planar pattern, checks whether the first two alternative mode that RMD selects is all angle mode.If the first two alternative mode that RMD selects is all angle mode, then this predicting unit is the texture block with certain orientation, and RDO will only check first three alternative mode that RMD selects and MPM, and other patterns can be removed from alternative mode list.
Step 4: the first alternative mode selected for RMD is not in MPM, and the first two alternative mode that RMD selects is not all DC or Planar pattern, also be not all the predicting unit of angle mould formula, then this predicting unit is the complex texture block not having specific direction, RDO will only check the first two alternative mode that RMD selects and MPM, and other patterns can be removed from alternative mode list.
Step 5: the optimization model selected by RDO is RQT to current coded unit.
Owing to adopting such scheme, the invention has the beneficial effects as follows:
1, first this discovery make use of the relation of neighboring prediction block that RMD and MPM result reflects and current prediction block using texture homogeneity, and the consistent predicting unit of texture can be made directly to skip RDO process.This technology can reduce the complexity of HEVC intra mode decision process effectively.
2, the predicting unit of this algorithm to all sizes adopts texture features prediction scheme in unified block, only predicting unit just can be divided into three classes by different texture features by the first two RDO alternative mode of Water demand RMD gained, and carries out predictive mode selection acceleration respectively.
3, the present invention is on the basis of existing HEVC intra mode decision scheme, when losing video subjective and objective quality hardly, effectively reduces encoder complexity, and the method is easy to realize.
Accompanying drawing explanation
Fig. 1 is fast schema selection method flow chart in HEVC frame of the present invention.
Embodiment
Below in conjunction with accompanying drawing illustrated embodiment, the present invention is further illustrated.
As shown in Figure 1, an embodiment of the present invention comprises the following steps:
Step 1: based on the test platform HM that HEVC is general, after RMD and MPM of current prediction unit finishes, the most possible pattern examination present encoding block of the adjacent encoder block that the alternative mode list drawn according to RMD and MPM draw and the texture relation of adjacent encoder block.If the first mode in alternative mode list in MPM, then skips RDO, the first alternative mode, as RDO optimization model, directly carries out follow-up RQT process.Otherwise forward step 2 to.First branch in this step corresponding diagram 1, namely the first alternative mode is in most possible pattern.
Step 2: check whether alternative mode list the first two pattern is all 0,1 pattern, i.e. DC and Planar pattern.If so, then RDO is only to 0, and 1 pattern checks, after obtaining optimal prediction modes, then carries out follow-up RQT process.Otherwise forward step 3 to.Second branch in this step corresponding diagram 1, namely the 1st, 2 alternative mode is pattern 0,1.
Whether step 3: check whether alternative mode list the first two pattern is all angle mould formula, namely all in 2 ~ 35 these closed intervals.If so, then RDO only checks first three pattern and the MPM of alternate list, after obtaining optimal prediction modes, then carries out follow-up RQT process.Otherwise forward step 4 to.3rd branch of this step corresponding diagram 1, namely first three alternative mode adds most possible pattern.
Step 4:RDO only checks the first two pattern and the MPM of alternate list, after obtaining optimal prediction modes, then carries out follow-up RQT process.4th branch of this step corresponding diagram 1.
Above-mentioned is can understand and apply the invention for ease of those skilled in the art to the description of embodiment.Person skilled in the art obviously easily can make various amendment to these embodiments, and General Principle described herein is applied in other embodiments and need not through performing creative labour.Therefore, the invention is not restricted to the embodiments described, and those skilled in the art, according to announcement of the present invention, do not depart from improvement that scope makes and amendment all should within protection scope of the present invention.
Claims (7)
1. a HEVC Intra prediction mode selection accelerated method, it is characterized in that: the using texture homogeneity first inferring the adjacent PU in space and current PU, if have this feature of using texture homogeneity, for the predicting unit of first mode in MPM that RMD chooses, skip RDO process, directly arranging this pattern is RDO optimization model.
2. HEVC Intra prediction mode selection accelerated method according to claim 1, is characterized in that: the rule definition of described using texture homogeneity is: if first alternative mode is in MPM, then think that current PU and adjacent PU are that texture is consistent.
3. HEVC Intra prediction mode selection accelerated method according to claim 1, it is characterized in that: the textural characteristics of the adjacent PU of the current prediction unit utilizing MPM to reflect, and the textural characteristics of current PU that first alternative mode selecting of RMD reflects, MPM with RMD is combined and infers current PU whether texture is consistent with adjacent PU.
4. HEVC Intra prediction mode selection accelerated method according to claim 1, it is characterized in that: the textural characteristics in the PU that the first two pattern utilizing RMD to choose reflects, by judging whether that the first two pattern that RMD chooses is all DC or Planar pattern, or be all angle mould formula, the predicting unit of first mode not in MPM chosen RMD is divided into different situations further and carries out model selection acceleration respectively.
5. HEVC Intra prediction mode selection accelerated method according to claim 4, is characterized in that:
For being judged as the predicting unit inconsistent with adjacent PU texture, the first two alternative mode utilizing RMD to select infers current PU textural characteristics, and predicting unit is divided three classes according to following criterion:
(1) if the first and second alternative mode of current prediction unit RMD gained are all DC or Planar pattern, then this PU is flat block;
(2) if the first and second alternative mode of current prediction unit RMD gained are all angle mould formula, then this PU is the texture block with certain orientation;
(3) if the first and second alternative mode of current prediction unit RMD gained are not all DC or Planar pattern, be also not all angle mould formula, then this PU is the complex texture block not having specific direction.
6. HEVC Intra prediction mode selection accelerated method according to claim 5, it is characterized in that: for first kind flat block predicting unit, the alternative mode of its RDO only includes DC and Planar pattern, and other pattern can be removed from alternate list; Equations of The Second Kind is had to the texture block predicting unit of certain orientation, the alternative mode of its RDO only includes first three alternative mode and the MPM of RMD gained, and other pattern can be removed from alternate list; 3rd class is not had to the complex texture block predicting unit of specific direction, the alternative mode of its RDO only includes the first two alternative mode and the MPM of RMD gained, and other pattern can be removed from alternate list.
7. HEVC Intra prediction mode selection accelerated method according to claim 1, is characterized in that: comprise the following steps:
Step 1: RMD and MPM carrying out current prediction unit;
Step 2: check the first alternative mode of selecting of RMD whether in MPM; If the first alternative mode that RMD selects is in MPM, then this prediction has similar textural characteristics with adjacent encoded unit, directly this pattern is set to RDO optimization model, skips RDO process;
Step 3: the predicting unit of the first alternative mode not in MPM selected for RMD, checks whether the first two alternative mode that RMD selects is all DC or Planar pattern; If the first two alternative mode that RMD selects is all DC or Planar pattern, then this predicting unit is flat block, and its RDO only checks DC and Planar two alternative mode, and other angle mould formula can be removed from alternative mode list;
Step 4: the first alternative mode selected for RMD is not in MPM, and the first two alternative mode that RMD selects is not all the predicting unit of DC or Planar pattern, checks whether the first two alternative mode that RMD selects is all angle mode; If the first two alternative mode that RMD selects is all angle mode, then this predicting unit is the texture block with certain orientation, and RDO will only check first three alternative mode that RMD selects and MPM, and other pattern can be removed from alternative mode list;
Step 5: if the first alternative mode that RMD selects is not in MPM, and the first two alternative mode that RMD selects is not all DC or Planar pattern, also be not all the predicting unit of angle mould formula, then this predicting unit is the complex texture block not having specific direction, RDO will only check the first two alternative mode that RMD selects and MPM, and other pattern can be removed from alternative mode list;
Step 6: the optimization model selected by RDO is RQT to current coded unit.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410842187.XA CN104581152A (en) | 2014-12-25 | 2014-12-25 | HEVC intra-frame prediction mode decision accelerating method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410842187.XA CN104581152A (en) | 2014-12-25 | 2014-12-25 | HEVC intra-frame prediction mode decision accelerating method |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN104581152A true CN104581152A (en) | 2015-04-29 |
Family
ID=53096221
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410842187.XA Pending CN104581152A (en) | 2014-12-25 | 2014-12-25 | HEVC intra-frame prediction mode decision accelerating method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN104581152A (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105208387A (en) * | 2015-10-16 | 2015-12-30 | 浙江工业大学 | HEVC intra-frame prediction mode fast selection method |
| WO2017113456A1 (en) * | 2015-12-30 | 2017-07-06 | 邦彦技术股份有限公司 | Hevc-oriented fast intra-frame prediction mode decision method and device |
| CN107396130A (en) * | 2017-07-28 | 2017-11-24 | 天津大学 | HEVC fast intra mode decision making algorithms based on DCT |
| WO2019141007A1 (en) * | 2018-01-16 | 2019-07-25 | 腾讯科技（深圳）有限公司 | Method and device for selecting prediction direction in image encoding, and storage medium |
| CN111757105A (en) * | 2020-06-30 | 2020-10-09 | 北京百度网讯科技有限公司 | Image encoding processing method, apparatus, device and medium |
| CN112118444A (en) * | 2019-06-20 | 2020-12-22 | 杭州海康威视数字技术股份有限公司 | Encoding method and device |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102665079A (en) * | 2012-05-08 | 2012-09-12 | 北方工业大学 | Adaptive fast intra prediction mode decision for high efficiency video coding (HEVC) |
| CN103096069A (en) * | 2011-11-04 | 2013-05-08 | 吴秀美 | Method and apparatus of deriving intra predicion mode |
| CN103929652A (en) * | 2014-04-30 | 2014-07-16 | 西安电子科技大学 | Intra-frame prediction fast mode selecting method based on autoregressive model in video standard |
-
2014
- 2014-12-25 CN CN201410842187.XA patent/CN104581152A/en active Pending
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103096069A (en) * | 2011-11-04 | 2013-05-08 | 吴秀美 | Method and apparatus of deriving intra predicion mode |
| CN102665079A (en) * | 2012-05-08 | 2012-09-12 | 北方工业大学 | Adaptive fast intra prediction mode decision for high efficiency video coding (HEVC) |
| CN103929652A (en) * | 2014-04-30 | 2014-07-16 | 西安电子科技大学 | Intra-frame prediction fast mode selecting method based on autoregressive model in video standard |
Non-Patent Citations (1)
| Title |
|---|
| DONGDONG ZHANG，YOUWEI CHEN，EBROUL IZQUIERDO: "Fast intra mode decision for HEVC based on texture characteristic from RMD and MPM", 《VISUAL COMMUNICATIONS AND IMAGE PROCESSING CONFERENCE, 2014 IEEE》 * |
Cited By (12)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105208387A (en) * | 2015-10-16 | 2015-12-30 | 浙江工业大学 | HEVC intra-frame prediction mode fast selection method |
| CN105208387B (en) * | 2015-10-16 | 2018-03-13 | 浙江工业大学 | A kind of HEVC Adaptive Mode Selection Method for Intra-Prediction |
| WO2017113456A1 (en) * | 2015-12-30 | 2017-07-06 | 邦彦技术股份有限公司 | Hevc-oriented fast intra-frame prediction mode decision method and device |
| CN107396130A (en) * | 2017-07-28 | 2017-11-24 | 天津大学 | HEVC fast intra mode decision making algorithms based on DCT |
| WO2019141007A1 (en) * | 2018-01-16 | 2019-07-25 | 腾讯科技（深圳）有限公司 | Method and device for selecting prediction direction in image encoding, and storage medium |
| US11395002B2 (en) | 2018-01-16 | 2022-07-19 | Tencent Technology (Shenzhen) Company Limited | Prediction direction selection method and apparatus in image encoding, and storage medium |
| CN112118444A (en) * | 2019-06-20 | 2020-12-22 | 杭州海康威视数字技术股份有限公司 | Encoding method and device |
| CN112118444B (en) * | 2019-06-20 | 2022-11-25 | 杭州海康威视数字技术股份有限公司 | Encoding method and device |
| CN111757105A (en) * | 2020-06-30 | 2020-10-09 | 北京百度网讯科技有限公司 | Image encoding processing method, apparatus, device and medium |
| US20210297664A1 (en) * | 2020-06-30 | 2021-09-23 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method for image encoding, electronic device and storage medium |
| US11582449B2 (en) * | 2020-06-30 | 2023-02-14 | Beijing Baidu Netcom Science And Technology Co., Ltd. | Method for image encoding, electronic device and storage medium |
| CN111757105B (en) * | 2020-06-30 | 2023-07-04 | 北京百度网讯科技有限公司 | Image coding processing method, device, equipment and medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| WO2018010492A1 (en) | Rapid decision making method for intra-frame prediction mode in video coding | |
| CN104581152A (en) | HEVC intra-frame prediction mode decision accelerating method | |
| CN101815215B (en) | Selecting method for coding mode and a device thereof | |
| CN103517069A (en) | HEVC intra-frame prediction quick mode selection method based on texture analysis | |
| CN103248895B (en) | A kind of quick mode method of estimation for HEVC intraframe coding | |
| CN105141954A (en) | HEVC interframe coding quick mode selection method | |
| EP2214415A1 (en) | A dual prediction video encoding and decoding method and a device | |
| CN103997645B (en) | Quick HEVC intra-frame coding unit and pattern decision method | |
| CN103561263A (en) | Motion compensation prediction method based on motion vector restraint and weighting motion vector | |
| CN102932642B (en) | Interframe coding quick mode selection method | |
| CN105721878A (en) | Image Processing Device And Method For Intra-Frame Predication In Hevc Video Coding | |
| JP5345220B2 (en) | Encoding and decoding with the exclusion of one or more predetermined predictors | |
| CN104394409B (en) | HEVC predictive mode fast selecting methods based on spatial correlation | |
| CN102217315A (en) | I-frame de-flickering for gop-parallel multi-thread video encoding | |
| CN101888546B (en) | A kind of method of estimation and device | |
| CN103327325A (en) | Intra-frame prediction mode rapid self-adaptation selection method based on HEVC standard | |
| CN103533355B (en) | A kind of HEVC fast encoding method | |
| CN107566846A (en) | Video coding skip mode decision-making technique, device, equipment and storage medium | |
| CN104702959B (en) | A kind of intra-frame prediction method and system of Video coding | |
| CN110351552B (en) | Fast coding method in video coding | |
| CN104853191A (en) | HEVC fast coding method | |
| Ting et al. | Gradient-based PU size selection for HEVC intra prediction | |
| CN105681808A (en) | Rapid decision-making method for SCC interframe coding unit mode | |
| CN102647593A (en) | AVS (Audio Video Standard) intra mode decision method and AVS intra mode decision device | |
| CN110365975A (en) | A kind of AVS2 video encoding and decoding standard prioritization scheme |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| WD01 | Invention patent application deemed withdrawn after publication | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20150429 |