CN103888782A - Parallel task partitioning method for HEVC decoder - Google Patents
Parallel task partitioning method for HEVC decoder Download PDFInfo
- Publication number
- CN103888782A CN103888782A CN201410077419.7A CN201410077419A CN103888782A CN 103888782 A CN103888782 A CN 103888782A CN 201410077419 A CN201410077419 A CN 201410077419A CN 103888782 A CN103888782 A CN 103888782A
- Authority
- CN
- China
- Prior art keywords
- decoding
- unit
- control unit
- hevc decoder
- working cell
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 75
- 238000000638 solvent extraction Methods 0.000 title abstract 3
- 230000008569 process Effects 0.000 claims abstract description 26
- 238000012545 processing Methods 0.000 claims abstract description 23
- 238000013139 quantization Methods 0.000 claims abstract description 8
- 230000001360 synchronised effect Effects 0.000 claims abstract description 8
- 230000009466 transformation Effects 0.000 claims abstract description 8
- 230000003139 buffering effect Effects 0.000 claims description 15
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 claims description 8
- 238000001914 filtration Methods 0.000 claims description 7
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 230000008030 elimination Effects 0.000 claims description 6
- 238000003379 elimination reaction Methods 0.000 claims description 6
- 239000000284 extract Substances 0.000 claims description 4
- 238000007599 discharging Methods 0.000 claims description 2
- 230000000153 supplemental effect Effects 0.000 claims description 2
- 238000012546 transfer Methods 0.000 claims description 2
- 239000000872 buffer Substances 0.000 abstract description 9
- 238000012360 testing method Methods 0.000 description 8
- 238000013461 design Methods 0.000 description 6
- 230000007246 mechanism Effects 0.000 description 5
- 230000006399 behavior Effects 0.000 description 3
- 238000001514 detection method Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 230000006870 function Effects 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 2
- 238000005070 sampling Methods 0.000 description 2
- 239000002699 waste material Substances 0.000 description 2
- 241001209435 Actus Species 0.000 description 1
- 241000796385 Choreutis chi Species 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008901 benefit Effects 0.000 description 1
- 239000008280 blood Substances 0.000 description 1
- 210000004369 blood Anatomy 0.000 description 1
- 230000000052 comparative effect Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000010276 construction Methods 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000008520 organization Effects 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 238000011056 performance test Methods 0.000 description 1
- 230000008707 rearrangement Effects 0.000 description 1
- 238000012163 sequencing technique Methods 0.000 description 1
Images
Abstract
The invention discloses a parallel task partitioning method for an HEVC decoder. According to the method, processing of a network abstraction layer unit, decoding of a parameter set and header information and management work of a decoding image buffer area are executed by one control unit, entropy coding, inverse quantization, inverse transformation, intra-frame prediction, inter-frame prediction and deblocking filter required by a decoding and coding tree unit and pixel self-adaptive compensation are executed by a plurality of work units, and a steamline type work mode is adopted in the control unit and the work units. The parallel task partitioning method for the HEVC decoder can ensure load balance better and eliminate dependence on a lock structure in the synchronous process, so that the computing resource utilization rate is improved, and synchronous overhead is saved.
Description
Technical field
The present invention relates to a kind of decoder implementation method of video decode technical field, specifically a kind of decoder parallel task division methods for HEVC standard.
Background technology
In order to adapt to better ultra high-definition video traffic, International Telecommunications Union and International Standards Organization combine and have formulated video compression standard HEVC of new generation.For HEVC decoder, the huge data processing demand that ultra high-definition video brings has exceeded the disposal ability of existing monokaryon CPU, and this has proposed stern challenge to decoding real-time.
At present, the best mode addressing the above problem is parallel processing technique.Decoder is carried out to paralell design and need to solve two class problems, be i.e. task division and synchronous.The first kind of way of task division is that function is divided, and comprises that each functional modules such as entropy decoding, inverse quantization, inverse transformation, infra-frame prediction, inter prediction, block elimination filtering and pixel adaptive equalization are assigned on different processor units by HEVC decoder; The second way of task division is that data are divided, and spatially carries out region division according to certain mode by image frame data to be decoded, then the decoding task of zones of different is distributed to one group of processor and has been walked abreast.The object of decoder synchronous working is the execution sequence between the TU task unit of one group of concurrent working of constraint, guarantees the correctness of decode procedure.
Up to now, the most efficient HEVC decoding framework---the row wave current water method for parallel processing of eclipsed form that industrial quarters and academia announce, at periodical Journal of Signal Processing Systems by authors such as C.Chi, Volume71, Issue3, delivers in the paper of Parallel HEVC Decoding on Multi-and Many-core Architectures by name in pp247-260.The method can be summarized as follows: for decoding process that can not parallel processing, distribute scanning and two processing units of output.Network abstraction layer unit is mainly responsible for detecting, extract and processing in scan process unit, decoding parametric collection categorical data, obtaining slice header information and data to be decoded are divided into the unit that several can parallel processing according to code tree unit behavior unit, is current decoded frame memory allocated region buffer release region in time in decoding image buffering area.Output processing unit needs the image of being responsible for decoding to complete to put into buffering area according to DISPLAY ORDER sequence on the one hand, needs on the other hand timely release to show the shared memory space of image of output.
For process that can parallel processing, the method can be distributed corresponding parallel computation working cell according to the sum of computer available resources.Be responsible for code tree unit behavior unit above-mentioned working cell, carries out parallel decoding according to the data dependence relation in the ranks of row wave current water method regulation.Row wave current water method for parallel processing that it should be noted that eclipsed form is expanded in time-domain row wave current water method under the prerequisite of compatible HEVC standard code stream, and obtained larger can concurrency.The improved principle of time-domain is: if can not continue to provide the code tree cell row that needs decoding for working cell in current decoded frame time, just decoder can start the decode procedure to next frame image.This mechanism can reduce the computing unit time idle because of wait task effectively, contributes to improve overall decoding performance.
Although the row wave current water method for parallel processing of eclipsed form has obtained good real-time decoding speed, it still has the leeway that can promote in performance.By analysis, we are found to be scan process unit and export processing unit corresponding function and distribute two parallel computation unit to have the design drawback of following two aspects:
First drawback is that resource utilization is low.Briefly say, the task of scanning work unit mainly comprises that decoding header and application, release show the storage area in image buffer, and the task of output services unit mainly comprises showing that the data in image buffers region sort and discharge.
We are the computation complexity of analytic solution harbour information first.The document HEVC complexity and implementation analysis.IEEE Transaction on Circuits and Systems for Video Technology2012:22 (12) delivering people such as F.Bossen, has carried out full test to the complexity of HEVC decoder in document Comparative Rate-Distortion-Complexity Analysis of HEVC and AVC Video Codecs.IEEE Transaction on Circuits and Systems for Video Technology2012.22 (12) .1885-1898 that the people such as pp1685-1696. and J.Vanne deliver.From the computational complexity analysis result to each functional module of decoder, entropy decoded portion accounts for 5% to 6% of decoder overall complexity, and Columbus's coding/decoding method complexity of using of decoding header is compared with low and data volume is less for global entropy solution code process, this part decode time only accounts for the very fraction of entropy decoding, accounts for below 1% of overall decode time.Even if the workload of all the other entropy decodings and prediction process of reconstruction makes the task amount that is assigned to each computing unit decrease because of paralell design, for the processor of common 8 cores or 4 cores, each processor at least will be born the computation complexity of 90.4%/8=11.3% or 90.4%/4=22.6%.And this complexity is apparently higher than the complexity of decoding header.
Next the computation complexity of further analysis management buffering area.Mainly comprise the following aspects for the management work that shows image buffer: in buffering area, be 1) current decoded video frames application memory space; 2) due to likely different between decoding order and DISPLAY ORDER, the frame of video of decoding and rebuilding need to be in the enterprising rearrangement order of time sequencing before output; 3) discharge the image that has shown output and be no longer used as reference frame.In these work, decoded frame is reordered and can in linear session, complete (such as chain structure) by efficient data structure; And show application in image buffer or discharge time of the required consumption of memory space also less.Deliver the test result analysis of document according to F.Bossen and J.Vanne, its method complexity accounts for overall below 1%.
To sum up, although header and the buffer information of two decoded frames can be processed in scanning work unit and output services unit simultaneously, because the complexity of itself method is starkly lower than the work complexity of decoding band cutting unit, above-mentioned two computing units there will be the longer idle waiting time in running, and this can cause the waste of computational resource.
Second drawback is to produce synchronization overhead.From the angle of parallel computation, decoded picture display buffer is one group and shares data, and in scanning work unit and output services unit, controls the application of this buffering area and the program of release function all belongs to critical section.Execution sequence between critical section program must be controlled effectively, otherwise just can make mistakes.A simple example is, in the situation that there is no synchronization mechanism, while carrying out memory space releasing operation when the frame of video not re-used being detected in Yin buffering area, scanning work unit, output services unit is probably also because detecting that identical phenomenon will carry out same operation.Not only can waste computational resource for twice to identical operation execution and also very likely make the mistake, therefore need to introduce synchronization mechanisms for these two critical sections.And the cost of introducing synchronization mechanism is exactly the expense of generation and sync correlation, this comprises that waiting for because of a working cell that another working cell completes access to sharing the formula of monopolizing of data cell the resources idle causing, and the synchronization mechanism of use computer software and hardware Resource Supply and the expense that causes.
In sum, we need a kind of HEVC decoder parallel task division methods that can further improve computational resource utilance and save synchronization overhead.
Summary of the invention
The present invention is directed to the deficiency of existing HEVC decoder design method, design a kind of HEVC decoder parallel task division methods, thereby realize the object that improves computational resource utilance and save synchronization overhead.
The present invention is achieved by the following technical solutions, HEVC decoder parallel task division methods of the present invention, stipulate entropy decoding, inverse quantization, inverse transformation, infra-frame prediction, inter prediction, the collaborative work mode of block elimination filtering and pixel adaptive equalization functional module, that is: by the processing to network abstraction layer unit, to the decoding of parameter set and header, and transfer to a control unit to carry out to the management work of decoding image buffering area, decoding and coding is set to the entropy decoding that unit needs, inverse quantization, inverse transformation, infra-frame prediction, inter prediction, block elimination filtering and pixel adaptive equalization process are transferred to several working cell executed in parallel, between above-mentioned control unit and working cell, meet the working method of pipeline system.
In above-mentioned HEVC decoder parallel task division methods, control unit comprises and extracts network abstraction layer unit and remove competition position the work for the treatment of of network abstraction layer unit; The parameter set information of control unit decoding comprises video parameter collection, sequence parameter set, image parameters collection and supplemental enhancement information; The header of control unit decoding comprises that network abstraction layer unit header, access unit define that symbol information, sequence termination message, bit stream termination message, padding data information, band are cut apart header, band is cut apart suffix information; Control unit is included as image memory allocated space to be decoded, the image in buffering area is being the shared space of the needed image of subsequent decoding process according to showing the order sequence of output and not discharging the management work of decoding image buffering area.
In above-mentioned HEVC decoder parallel task division methods, the decoding work of a code tree cell row is responsible in each working cell, and the updating context that wherein entropy is decoded is the regulation to the parallel processing of row wave current water according to HEVC standard.Between working cell, adopt the collaborative work mode synchronous without lock, for the working cell of processing two code tree cell row adjacent on space, in decoding progress, will guarantee that next line at least lags behind two code tree unit of lastrow.
Compared with prior art, the present invention has following beneficial effect:
In above-mentioned HEVC decoder parallel task division methods, between control unit and working cell, meet the working method of pipeline system, working cell is in the time processing current frame data, and control unit can be processed next frame data.Between control unit and working cell, adopt the collaborative work mode synchronous without lock.The present invention's proof load balance better is also removed the dependence of synchronizing process for lock construction, thereby has improved computational resource utilance and saved synchronization overhead.
Accompanying drawing explanation
By reading the detailed description of non-limiting example being done with reference to the following drawings, it is more obvious that other features, objects and advantages of the present invention will become:
Fig. 1 is the HEVC decoder parallel task dividing system frame diagram that the present invention stipulates.
Fig. 2 is the workflow diagram of one embodiment of the invention control unit.
Fig. 3 is the workflow diagram of one embodiment of the invention working cell.
Fig. 4 be between one embodiment of the invention control unit and working cell without lock synchronizing process.
Fig. 5 is the spatially adjacent synchronizing process of two code tree cell row up and down of one embodiment of the invention parallel decoding.
Embodiment
Below in conjunction with specific embodiment, the present invention is described in detail.Following examples will contribute to those skilled in the art further to understand the present invention, but not limit in any form the present invention.It should be pointed out that to those skilled in the art, without departing from the inventive concept of the premise, can also make some distortion and improvement.These all belong to protection scope of the present invention.
The HEVC decoder task division method stipulating for the present invention shown in Fig. 1.Wherein, control unit is responsible for from coded bit stream, extracting and processing network abstraction layer unit, parameter set information and header are wherein resolved, be then stored in task pool according to the subcode stream of code tree row partition, for follow-up decode procedure.In addition, control unit is also in charge of decoding image buffering area, for being about to the image memory allocated space of decoding, and discharges the storage area no longer being used by subsequent decoding process.Except a control unit, the system shown in Fig. 1 also can, according to concrete hardware resource, be distributed several working cells.Between each working cell, mutually coordinate the subcode stream in decoding task pond concurrently.
Shown in Fig. 2 is the workflow of one embodiment of the invention control unit.First judge in header buffering area whether obtained the header that is no more than a two field picture in initial condition, if condition does not meet, again detect this condition; If condition meets, further Sampling network level of abstraction unit in bit stream.If network abstraction layer unit do not detected, repeat the testing to network abstraction layer unit, if network abstraction layer unit detected, extract this unit and remove competition position.For removing the later network abstraction layer unit in competition position, can carry out different parse operation according to its Type Control unit, if network abstraction layer unit is parameter set type, to preserve after corresponding resolution data, control unit comes back to the initial condition of flow process; If being band, network abstraction layer unit cuts apart type, control unit starts to resolve band and cuts apart header, code stream is divided into subcode stream code tree unit behavior unit, that supply row wave current water parallel decoding, the then image memory allocated region for being about to decode in decoding image buffering area, and discharge the shared buffer area of image no longer being used by subsequent decoding process, finally get back to the starting stage of flow process.
Shown in Fig. 3 is the workflow of one embodiment of the invention working cell.Working cell first whether have in Detection task pond can for decoding with a subcode stream that code tree cell row is corresponding, if not, duplicate detection process; If had, further judge whether first code tree unit of current line meets the required data dependency condition of parallel decoding.The data dependency condition here refers to, to current decoded picture frame take code tree unit after unit divides, if present encoding tree unit is in the capable m row of n, if the non-existent words in code tree unit of the capable m+1 row of n-1, the data dependency condition of present encoding tree unit is defaulted as satisfied, if the code tree unit of the capable m+1 row of n-1 exists, whether the decode procedure of investigating this unit completes, if completed, the data dependency condition of present encoding tree unit meets, if do not completed, shows that data dependency condition does not meet.After the data dependency condition of present encoding tree unit meets, start present encoding tree unit to carry out entropy decoding, inverse quantization, inverse transformation, prediction reconstruction and loop filtering process.After said process completes, investigate the position of present encoding tree unit in present encoding tree cell row, if present encoding is set unit in current line low order end, workflow is got back to the initial condition of flow chart shown in Fig. 3; If present encoding tree unit is not at the low order end of one's own profession, investigates object and be set to the code tree unit on the present encoding tree right side, unit, and start to investigate this unit and whether meet data dependency condition.
Shown in Fig. 4 be between one embodiment of the invention control unit and working cell without lock synchronizing process.Synchronizing process between above-mentioned two class unit is by a flag bit control, and this flag bit is used for judging whether current frame pixel decoding completes, and is initialized to true.After control unit is parsed a slice header information, all can investigate this flag bit, if this flag bit is SM set mode, can starts the parsing work of next slice header information and this flag bit is resetted; Otherwise if this flag bit is reset mode, this flag bit of duplicate detection, until this flag bit is done.After last code tree unit decodes of present frame, working cell is set again.
Shown in Fig. 5 is the spatially adjacent synchronizing process of two code tree cell row up and down of one embodiment of the invention parallel decoding.Above-mentioned synchronizing process is the flag bit control corresponding by each code tree unit.In the time that being SM set mode, the flag bit of a code tree unit represents that the decode procedure of this code tree unit completes; If when this flag bit is reset mode, represent that the decode procedure of this code tree unit not yet completes.The picture frame of current decoding is divided take code tree unit as unit, establish code tree unit A and be positioned at the capable m row of n, and code tree unit B is positioned at the capable m+1 row of n-1.In the time judging whether code tree unit A can start decode procedure, need to investigate the flag bit of code tree unit B, if this flag bit is SM set mode, can start code tree unit A and comprise the decoding process of entropy decoding, inverse quantization, inverse transformation, infra-frame prediction, inter prediction, block elimination filtering and pixel adaptive equalization etc., flag bit set then by the flag bit reset of code tree unit B, and to code tree unit A; If the flag bit of code tree unit B is reset mode, the decoding need of work of code tree unit A is postponed, until the flag bit of code tree unit B is set because its decode procedure completes.
Under the test code streams encoding condition shown in the experiment porch shown in table 1, table 2, to testing according to the performance of the designed decoder of method in the present invention, cycle tests is that resolution is 3840x2160, bit-depth is 8, ultra high-definition sequence C actus, Coastguard and foreman that aberration sub sampling pattern is 4:2:0.The performance test results is as shown in table 3, and this test result proves that the method for the present invention's design has good real-time, and this has reflected the effective use to computational resource and the saving to synchronization overhead in decode procedure.
Table 1 experiment porch information
Table 2 test code streams encoding condition
Table 3 ultra high-definition sequential decoding performance
Above specific embodiments of the invention are described.It will be appreciated that, the present invention is not limited to above-mentioned specific implementations, and those skilled in the art can make various distortion or modification within the scope of the claims, and this does not affect flesh and blood of the present invention.
Claims (6)
1. a HEVC decoder parallel task division methods, the method has stipulated entropy decoding, inverse quantization, inverse transformation, infra-frame prediction, inter prediction, the collaborative work mode of block elimination filtering and pixel adaptive equalization functional module, it is characterized in that: described method is by the processing to network abstraction layer unit, to the decoding of parameter set and header, and transfer to a control unit to carry out to the management work of decoding image buffering area, decoding and coding is set to the entropy decoding that unit needs, inverse quantization, inverse transformation, infra-frame prediction, inter prediction, block elimination filtering and pixel adaptive equalization process are transferred to several working cell executed in parallel, between above-mentioned control unit and working cell, meet the working method of pipeline system.
2. a kind of HEVC decoder parallel task division methods according to claim 1, is characterized in that, described control unit comprises and extracts network abstraction layer unit and remove competition position the work for the treatment of of network abstraction layer unit;
The parameter set information of described control unit decoding comprises video parameter collection, sequence parameter set, image parameters collection and supplemental enhancement information;
The header of described control unit decoding comprises that network abstraction layer unit header, access unit define that symbol information, sequence termination message, bit stream termination message, padding data information, band are cut apart header, band is cut apart suffix information;
Described control unit is included as image memory allocated space to be decoded, the image in buffering area is being the shared space of the needed image of subsequent decoding process according to showing the order sequence of output and not discharging the management work of decoding image buffering area.
3. a kind of HEVC decoder parallel task division methods according to claim 1 and 2, it is characterized in that, the decoding work of a code tree cell row is responsible in each described working cell, and wherein the updating context of entropy decoding is according to the regulation of HEVC standard centering row wave current water parallel processing.
4. a kind of HEVC decoder parallel task division methods according to claim 3, it is characterized in that, between described working cell, adopt the collaborative work mode synchronous without lock, and for the working cell of processing two code tree cell row adjacent on space, in decoding progress, to guarantee that next line at least lags behind two code tree unit of lastrow.
5. a kind of HEVC decoder parallel task division methods according to claim 1 and 2, it is characterized in that, between described control unit and working cell, meet the working method of pipeline system, working cell is in the time processing current frame data, and control unit can be processed next frame data.
6. a kind of HEVC decoder parallel task division methods according to claim 1, is characterized in that, adopts the collaborative work mode synchronous without lock between described control unit and working cell.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410077419.7A CN103888782B (en) | 2014-03-04 | 2014-03-04 | Parallel task partitioning method for HEVC decoder |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201410077419.7A CN103888782B (en) | 2014-03-04 | 2014-03-04 | Parallel task partitioning method for HEVC decoder |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103888782A true CN103888782A (en) | 2014-06-25 |
| CN103888782B CN103888782B (en) | 2017-02-08 |
Family
ID=50957470
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201410077419.7A Expired - Fee Related CN103888782B (en) | 2014-03-04 | 2014-03-04 | Parallel task partitioning method for HEVC decoder |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103888782B (en) |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104954808A (en) * | 2015-06-17 | 2015-09-30 | 福州瑞芯微电子有限公司 | Method and system for realizing cycle operation based on CTU (Coding Tree Unit) buffers |
| CN105847840A (en) * | 2015-11-18 | 2016-08-10 | 西安邮电大学 | Parallel structure construction method for high efficiency video coding inverse transform operation |
| CN106658012A (en) * | 2017-01-06 | 2017-05-10 | 华南理工大学 | Parallel pipeline task division method for VP9 decoder |
| CN107548490A (en) * | 2014-12-26 | 2018-01-05 | 英特尔公司 | Humidometer in parallel computation |
| CN111787330A (en) * | 2020-06-16 | 2020-10-16 | 眸芯科技（上海）有限公司 | Coding method supporting decoding compression frame buffer self-adaptive distribution and application |
| CN114125464A (en) * | 2020-08-27 | 2022-03-01 | 扬智科技股份有限公司 | Video decoding method and video decoding device |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120257702A1 (en) * | 2011-04-11 | 2012-10-11 | Matthias Narroschke | Order of deblocking |
| CN103024389A (en) * | 2012-12-24 | 2013-04-03 | 芯原微电子(北京)有限公司 | HEVC (high efficiency video coding) decoding device and method |
| CN103379318A (en) * | 2012-04-12 | 2013-10-30 | 中兴通讯股份有限公司 | Video bitstream coding method and apparatus |
| CN103414895A (en) * | 2013-07-29 | 2013-11-27 | 复旦大学 | Encoder intra-frame prediction device and method applicable to HEVC standards |
-
2014
- 2014-03-04 CN CN201410077419.7A patent/CN103888782B/en not_active Expired - Fee Related
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20120257702A1 (en) * | 2011-04-11 | 2012-10-11 | Matthias Narroschke | Order of deblocking |
| CN103379318A (en) * | 2012-04-12 | 2013-10-30 | 中兴通讯股份有限公司 | Video bitstream coding method and apparatus |
| CN103024389A (en) * | 2012-12-24 | 2013-04-03 | 芯原微电子(北京)有限公司 | HEVC (high efficiency video coding) decoding device and method |
| CN103414895A (en) * | 2013-07-29 | 2013-11-27 | 复旦大学 | Encoder intra-frame prediction device and method applicable to HEVC standards |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107548490A (en) * | 2014-12-26 | 2018-01-05 | 英特尔公司 | Humidometer in parallel computation |
| CN104954808A (en) * | 2015-06-17 | 2015-09-30 | 福州瑞芯微电子有限公司 | Method and system for realizing cycle operation based on CTU (Coding Tree Unit) buffers |
| CN104954808B (en) * | 2015-06-17 | 2018-03-23 | 福州瑞芯微电子股份有限公司 | One kind realizes rotation operating method and system based on code tree unit CTU buffer |
| CN105847840A (en) * | 2015-11-18 | 2016-08-10 | 西安邮电大学 | Parallel structure construction method for high efficiency video coding inverse transform operation |
| CN105847840B (en) * | 2015-11-18 | 2018-12-07 | 西安邮电大学 | A kind of parallel organization building method for efficient video coding inverse transformation operation |
| CN106658012A (en) * | 2017-01-06 | 2017-05-10 | 华南理工大学 | Parallel pipeline task division method for VP9 decoder |
| CN111787330A (en) * | 2020-06-16 | 2020-10-16 | 眸芯科技（上海）有限公司 | Coding method supporting decoding compression frame buffer self-adaptive distribution and application |
| CN111787330B (en) * | 2020-06-16 | 2022-03-08 | 眸芯科技（上海）有限公司 | Coding method supporting decoding compression frame buffer self-adaptive distribution and application |
| CN114125464A (en) * | 2020-08-27 | 2022-03-01 | 扬智科技股份有限公司 | Video decoding method and video decoding device |
| CN114125464B (en) * | 2020-08-27 | 2024-02-06 | 扬智科技股份有限公司 | Video decoding method and video decoding device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103888782B (en) | 2017-02-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103888782A (en) | Parallel task partitioning method for HEVC decoder | |
| CN105992008B (en) | A kind of multi-level multi-task parallel coding/decoding method in multi-core processor platform | |
| CN102098503B (en) | Method and device for decoding image in parallel by multi-core processor | |
| CN108449603B (en) | Based on the multi-level task level of multi-core platform and the parallel HEVC coding/decoding method of data level | |
| US9948941B2 (en) | Circuit, method and video decoder for video decoding | |
| US8660191B2 (en) | Software video decoder display buffer underflow prediction and recovery | |
| CN106454354B (en) | A kind of AVS2 parallel encoding processing system and method | |
| CN102625108B (en) | Multi-core-processor-based H.264 decoding method | |
| CN103297767A (en) | JPEG image decoding method and decoder suitable for multi-core embedded type platform | |
| KR101292668B1 (en) | Video encoding apparatus and method based-on multi-processor | |
| CN110337002A (en) | The multi-level efficient parallel decoding algorithm of one kind HEVC in multi-core processor platform | |
| US8090028B2 (en) | Video deblocking memory utilization | |
| HK1046799A1 (en) | Method and apparatus for video decoding on a multiprocessor system | |
| Sankaraiah et al. | GOP level parallelism on H. 264 video encoder for multicore architecture | |
| CN104521234B (en) | Merge the method for processing video frequency and device for going block processes and sampling adaptive migration processing | |
| CN106851298B (en) | High-efficiency video coding method and device | |
| CN104980764A (en) | Parallel coding/decoding method, device and system based on complexity balance | |
| WO2012097675A1 (en) | Encoding/decoding method and codec based on intra-prediction mode of image block | |
| CN101841722B (en) | Detection method of detection device of filtering boundary strength | |
| CN103997648A (en) | System and method for achieving decompression of JPEG2000 standard images rapidly based on DSPs | |
| Gudumasu et al. | Software-based versatile video coding decoder parallelization | |
| CN112422986A (en) | Hardware decoder pipeline optimization method and application | |
| De Souza et al. | OpenCL parallelization of the HEVC de-quantization and inverse transform for heterogeneous platforms | |
| US8111749B2 (en) | Image encoding or decoding method and device, with parallelization of processing over several processors and coprocessors, corresponding computer-readable storage medium | |
| CN101256668A (en) | Method for implementing video filtering to working balanced multiple nucleus |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20170208 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |