CN103379324A - Parallel realization method, device and system for advanced motion vector prediction AMVP - Google Patents
Parallel realization method, device and system for advanced motion vector prediction AMVP Download PDFInfo
- Publication number
- CN103379324A CN103379324A CN2012101118209A CN201210111820A CN103379324A CN 103379324 A CN103379324 A CN 103379324A CN 2012101118209 A CN2012101118209 A CN 2012101118209A CN 201210111820 A CN201210111820 A CN 201210111820A CN 103379324 A CN103379324 A CN 103379324A
- Authority
- CN
- China
- Prior art keywords
- motion
- vector prediction
- vector
- predicting unit
- unit
- 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
Images
Abstract
The embodiment of the invention provides a parallel realization method, a device and a system for advanced motion vector prediction AMVP. The parallel realization method includes that: a motion vector prediction selection set is constructed for a coding unit at a coding terminal; motion vector prediction is carried out in a parallel way by all prediction units of the coding unit through utilizing the constructed motion vector prediction selection set to generate predicted motion vectors respectively; motion estimation is carried out by each of the prediction units of the coding unit according to the predicted motion vector generated by the prediction unit. In the parallel implementation method, the device and the system for advanced motion vector prediction, motion vector prediction is carried out by all the prediction units of the coding unit by utilizing the generated motion vector prediction election set to generate predicted motion vectors respectively, and motion prediction is carried out by each prediction unit in the coding unit according to the predicted motion vector generated by the prediction unit, so that parallel motion prediction of prediction units of a coder terminal is achieved.
Description
Technical field
The present invention relates to multimedia video encoding and decoding technique field, relate in particular to Parallel Implementation method, the Apparatus and system of a kind of advanced motion-vector prediction (Advanced Motion Vector Prediction, AMVP).
Background technology
In the efficient video coding standard, coding unit (Coding Unit, CU) is the elementary cell of Video coding.The size of coding unit can be from 64x64 to 8x8.Wherein 64x64 becomes maximum coding unit (Largest Coding Unit, LCU).In each coding unit, give a forecast, change quantization, entropy coding, reprocessing be with coding.Wherein coding unit is when giving a forecast, and coding unit can be divided into the predicting unit (Prediction Unit, PU) of different sizes, and predicting unit is the elementary cell of prediction.Coding unit greater than 8x8 can be supported 2Nx2N, 2NxN, three kinds of symmetrical predicting unit dividing mode of Nx2N and 2Nx0.5N, 2Nx1.5N, 0.5Nx2N, four kinds of asymmetric predicting unit dividing mode of 1.5Nx2N; The coding unit of 8x8 has 2Nx2N, NxN, and 2NxN, four kinds of symmetrical predicting unit dividing mode of Nx2N, but do not have asymmetrical predicting unit to divide.(when coding unit was 64x64, N was 32.) for frame mode, the predicting unit of 2Nx2N is only arranged greater than the coding unit of 8x8; Also there is the predicting unit of NxN in the coding unit of 8x8.According to different dividing mode, coding unit has 1～4 predicting unit.Coding unit and predicting unit are divided: shown in Fig. 1 (a), for the prior art coding unit is divided schematic diagram, shown in Fig. 1 (b), for the prior art predicting unit is divided schematic diagram.
Coding side for the predicting unit of 2Nx2N, at first checks to merge/skip (Merge/Skip) pattern, then checks other inter-frame mode; For other big or small predicting unit, at first check to merge (Merge) pattern, then check other inter-frame mode.Merging/skip in the inter-frame mode beyond (Merge/Skip) pattern the predicting unit estimation of need to taking exercises.In the process of estimation.In predicting unit is taken exercises the process of estimating, adopt advanced motion-vector prediction mechanism (Advanced Motion Vector Prediction, AMVP); Namely the information of predicting unit is tried to achieve a motion-vector prediction participation in the election collection (MVP Candidate List around the basis, MVPCL), then from participation in the election collection (Candidate List), select a best contender's (Candidate) motion vector (MV) as the starting point of motion search, the motion vector that obtains according to motion search is at last concentrated from participating in the election of and is selected best utilization vector prediction (Motion Vector Prediction, MVP) as final motion-vector prediction.In decoding end, if the coding unit that decodes is inter-frame mode and is non-merging/skip mode, then each predicting unit in the coding unit is made up motion-vector prediction participation in the election collection, then obtain motion vector (Motion Vector according to other information decodings in motion-vector prediction participation in the election collection and the code stream, MV), the compensation of taking exercises at last.Because the structure of motion-vector prediction participation in the election collection is with predicting unit information is relevant on every side, this building process also can only be carried out in serial.
The formation of advanced motion-vector prediction participation in the election collection among the HM6.0: shown in Fig. 2 (a), be the spatial domain contender's schematic diagram in the participation in the election collection formation of the advanced motion-vector prediction of prior art, shown in Fig. 2 (b), be the time domain contender's schematic diagram in the participation in the election collection formation of the advanced motion-vector prediction of prior art.The contender in spatial domain is divided into two classes: the predicting unit (L on the left side
m, L
M+1) and the predicting unit (T of top
-1, T
n, T
N+1).For the predicting unit on the left side, from L
M+1To L
mCheck successively, select first available predicting unit, add the participation in the election collection; For the predicting unit of top, press T
N+1, T
n, T
-1First available predicting unit of sequential search, and its is added the participation in the election collection.For the contender of time domain, by first the available predicting unit of sequential search from the intermediate mass (Co-located Center) of same position to same position bottom right piece (Co-located Right Bottom), it is added participation in the election collect.
From the process of above-mentioned predicting unit coding, can see a problem: current predicting unit can be used the movable information of adjacent predicting unit on every side when encoding, and these movable informations can't obtain before adjacent predicting unit is encoded.So so that predicting unit when coding, must serial finish according to the order of sequence.Equally, decoding end also must be finished when generating the motion-vector prediction reference set according to the order of sequence in serial.So just limited the Parallel Implementation of encoding and decoding.For example in given parallel zone (such as the coding unit level), the contender who has some predicting unit depends on the together movable information of interior other predicting unit in zone, thereby estimation that can not each predicting unit of Parallel Implementation, as shown in Figure 3, advanced motion-vector prediction participation in the election collection formation schematic diagram (only drawn symmetrical predicting unit among Fig. 3 and divided, asymmetrical division is similar) for the prior art predicting unit.Blank region representation is when the predicting unit of pre-treatment, fills its contender of region representation of line anyhow or grey around the current predicting unit.The contender of the region representation top of band, the part of vertical line represents the contender on the left side, the part of grey is illustrated in the contender in the same parallel zone, the contender that the region representation of band also is not encoded before the current predicting unit of coding.To find out, in the predicting unit of interframe 2NxN, the contender of the top of predicting unit 1 just can obtain after must waiting until predicting unit 0 encoding and decoding from Fig. 3, thereby the encoding and decoding of two predicting unit 0,1 can only be carried out in serial; Other predicting unit are divided similar.
A new generation's video encoding standard---efficient video coding standard improved much than coding efficiency H.264/AVC, but its encoder complexity will be far above coding standard in the past.Particularly, now people to the requirement of video from the SD to the high definition, high definition very, the encoding and decoding time has also improved widely.How for codec provides a kind of parallel solution, become a major issue thereby accelerate encoding-decoding process.In the 8th JCT-VC meeting, some are adopted about the parallel scheme that merges/skip (Merge/Skip), comprise H0082, H0240.Yet, the parallel solution of merging/skip is only arranged, can not realize the parallel encoding and decoding of inter prediction unit truly, because the structure of motion-vector prediction participation in the election collection is with predicting unit is relevant in the estimation on every side.
In the prior art, inner in the one and same coding unit, the generation of the motion vector participation in the election collection of a predicting unit will be with the movable information of another predicting unit.As seen from Figure 3, the predicting unit of interframe 2N * N and N * 2N is divided, predicting unit 1 will be used the movable information of predicting unit 0; Predicting unit to interframe N * N is divided, and predicting unit 1 will be used the movable information of predicting unit 0, and predicting unit 2 will be used the movable information of predicting unit 0 and 1, and predicting unit 3 will be used predicting unit 0,1,2 movable information.So so that, a plurality of predicting unit of inside, one and same coding unit can not walk abreast and make up its motion vector participation in the election collection.
Summary of the invention
The embodiment of the invention provides Parallel Implementation method, the Apparatus and system of a kind of advanced motion-vector prediction AMVP, and can walk abreast with a plurality of predicting unit that realize inside, one and same coding unit makes up its motion vector participation in the election collection.
On the one hand, the embodiment of the invention provides the Parallel Implementation method of a kind of advanced motion-vector prediction AMVP, and the Parallel Implementation method of described AMVP comprises:
For a coding unit of coding side, set up same motion-vector prediction participation in the election collection;
All predicting unit in the described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast;
The motion vector of the prediction that each predicting unit in the described coding unit generates according to this predicting unit respectively carries out estimation.
Preferably, in an embodiment of the present invention, the size that the same described motion-vector prediction participation in the election of foundation integrates is as the motion-vector prediction participation in the election collection of the predicting unit of 2N * 2N, and wherein, N=2k, k are the integer more than or equal to 2.
Preferably, in an embodiment of the present invention, for decoding end, in decode procedure, at first carry out the entropy decoding, if the coding unit of decoding adopts interframe encoding mode and is not merging/skip mode, then carry out the derivation of advanced motion-vector prediction participation in the election collection, obtain the motion vector of the prediction of each predicting unit, determine its motion vector MV (motion vector) in conjunction with the motion vector of prediction and the motion vector difference MVD (Motion Vector Difference) of the predicting unit that obtains of decoding, according to this motion vector predicting unit is carried out motion compensation at last.
Preferably, in an embodiment of the present invention, the described derivation of carrying out advanced motion-vector prediction participation in the election collection, the motion vector that obtains predicting comprises: in case decode a coding unit, the parallel motion-vector prediction participation in the election collection that makes up all predicting unit of this coding unit inside.
On the other hand, the embodiment of the invention provides the Parallel Implementation device of a kind of advanced motion-vector prediction AMVP, and the Parallel Implementation device of described AMVP comprises:
Motion-vector prediction participation in the election collection is set up the unit, is used for the coding unit for coding side, sets up same motion-vector prediction participation in the election collection;
Motion vector prediction unit, all predicting unit that are used for described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast;
The motion vector of the prediction that motion estimation unit, each predicting unit that is used for described coding unit generate according to this predicting unit respectively carries out estimation.
Preferably, in an embodiment of the present invention, described motion-vector prediction participation in the election collection is set up size that the same described motion-vector prediction participation in the election of setting up the unit integrate and is collected as the motion-vector prediction participation in the election of the predicting unit of 2N * 2N, wherein, N=2k, k are the integer more than or equal to 2.
On the one hand, the embodiment of the invention provides the Parallel Implementation System of a kind of advanced motion-vector prediction AMVP again, and the Parallel Implementation System of described AMVP comprises:
Encoder is used for the coding unit for coding side, sets up same motion-vector prediction participation in the election collection; All predicting unit in the described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast; The motion vector of the prediction that each predicting unit in the described coding unit generates according to this predicting unit respectively carries out estimation;
Decoder, be used at decode procedure, at first carry out the entropy decoding, if the coding unit of decoding adopts interframe encoding mode and is not merging/skip mode, then carry out the derivation of advanced motion-vector prediction participation in the election collection, obtain the motion vector of the prediction of each predicting unit, determine its motion vector MV in conjunction with the motion vector of prediction and the motion vector difference MVD of the predicting unit that obtains of decoding, according to this motion vector predicting unit is carried out motion compensation at last.
Preferably, in an embodiment of the present invention, the size that the same described motion-vector prediction participation in the election that described encoder is set up integrates is as the motion-vector prediction participation in the election collection of the predicting unit of 2N * 2N, and wherein, N=2k, k are the integer more than or equal to 2.
Preferably, in an embodiment of the present invention, described decoder carries out the derivation of advanced motion-vector prediction participation in the election collection concurrently, the motion vector that obtains predicting comprises: in case decode a coding unit, the parallel motion-vector prediction participation in the election collection that makes up all predicting unit of this coding unit inside.
Technique scheme has following beneficial effect: because same motion-vector prediction participation in the election collection, is set up in employing for a coding unit of coding side; All predicting unit in the described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast; The motion vector of the prediction that each predicting unit in the described coding unit generates according to this predicting unit respectively carries out the technological means of estimation, thereby realized the parallel of encoder-side predicting unit estimation, it is parallel to have realized that further decoding end predicting unit motion vector participation in the election collection makes up.
Description of drawings
In order to be illustrated more clearly in the embodiment of the invention or technical scheme of the prior art, the below will do to introduce simply to the accompanying drawing of required use in embodiment or the description of the Prior Art, apparently, accompanying drawing in the following describes only is some embodiments of the present invention, for those of ordinary skills, under the prerequisite of not paying creative work, can also obtain according to these accompanying drawings other accompanying drawing.
Fig. 1 (a) divides schematic diagram for the prior art coding unit;
Fig. 1 (b) divides schematic diagram for the prior art predicting unit;
Fig. 2 (a) is the spatial domain contender's schematic diagram during the participation in the election collection of the advanced motion-vector prediction of prior art consists of;
Fig. 2 (b) is the time domain contender's schematic diagram during the participation in the election collection of the advanced motion-vector prediction of prior art consists of;
Fig. 3 is that the advanced motion-vector prediction participation in the election collection of prior art predicting unit consists of schematic diagram;
Fig. 4 is the Parallel Implementation method flow diagram of a kind of advanced motion-vector prediction AMVP of the embodiment of the invention;
Fig. 5 is the Parallel Implementation apparatus structure schematic diagram of a kind of advanced motion-vector prediction AMVP of the embodiment of the invention;
Fig. 6 is that the concurrency of original coding end is analyzed schematic diagram;
Fig. 7 is that the coding side concurrency after the embodiment of the invention is improved is analyzed schematic diagram;
Fig. 8 is that the embodiment of the invention 8 * 8 coding units adopt same set of motion-vector prediction participation in the election collection schematic diagram;
Fig. 9 is that the concurrency of original decoded end is analyzed schematic diagram;
Figure 10 is that the decoding end concurrency was analyzed schematic diagram after the embodiment of the invention was improved.
Embodiment
Below in conjunction with the accompanying drawing in the embodiment of the invention, the technical scheme in the embodiment of the invention is clearly and completely described, obviously, described embodiment only is the present invention's part embodiment, rather than whole embodiment.Based on the embodiment among the present invention, those of ordinary skills belong to the scope of protection of the invention not making the every other embodiment that obtains under the creative work prerequisite.
As shown in Figure 4, be the Parallel Implementation method flow diagram of a kind of advanced motion-vector prediction AMVP of the embodiment of the invention, the Parallel Implementation method of described AMVP comprises:
401, for a coding unit of coding side, set up same motion-vector prediction participation in the election collection;
402, all predicting unit in the described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast;
403, the motion vector of the prediction that generates according to this predicting unit respectively of each predicting unit in the described coding unit carries out estimation.
Preferably, the size that the same described motion-vector prediction participation in the election of foundation integrates is as the motion-vector prediction participation in the election collection of the predicting unit of 2N * 2N, and wherein, N=2k, k are the integer more than or equal to 2.
Preferably, for decoding end, in decode procedure, at first carry out the entropy decoding, if the coding unit of decoding adopts interframe encoding mode and is not merging/skip mode, then carry out the derivation of advanced motion-vector prediction participation in the election collection, obtain the motion vector of the prediction of each predicting unit, determine its motion vector MV in conjunction with the motion vector of prediction and the motion vector difference MVD of the predicting unit that obtains of decoding, according to this motion vector predicting unit is carried out motion compensation at last.Preferably, the described derivation of carrying out advanced motion-vector prediction participation in the election collection, the motion vector that obtains predicting comprises: in case decode a coding unit, the parallel motion-vector prediction participation in the election collection that makes up all predicting unit of this coding unit inside.
Corresponding to said method embodiment, as shown in Figure 5, be the Parallel Implementation apparatus structure schematic diagram of a kind of advanced motion-vector prediction AMVP of the embodiment of the invention, the Parallel Implementation device of described AMVP comprises:
Motion-vector prediction participation in the election collection is set up unit 51, is used for the coding unit for coding side, sets up same motion-vector prediction participation in the election collection;
Motion vector prediction unit 52, all predicting unit that are used for described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast;
The motion vector of the prediction that motion estimation unit 53, each predicting unit that is used for described coding unit generate according to this predicting unit respectively carries out estimation.
Preferably, described motion-vector prediction participation in the election collection is set up size that the same described motion-vector prediction participation in the election of setting up unit 51 integrate and is collected as the motion-vector prediction participation in the election of the predicting unit of 2N * 2N, and wherein, N=2k, k are the integer more than or equal to 2.
Again on the one hand, the embodiment of the invention also provides the Parallel Implementation System of a kind of advanced motion-vector prediction AMVP, the Parallel Implementation System of described AMVP comprises: encoder, be used for the coding unit for coding side, and set up same motion-vector prediction participation in the election collection; All predicting unit in the described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast; The motion vector of the prediction that each predicting unit in the described coding unit generates according to this predicting unit respectively carries out estimation; Decoder, be used at decode procedure, at first carry out the entropy decoding, if the coding unit of decoding adopts interframe encoding mode and is not merging/skip mode, then carry out the derivation of advanced motion-vector prediction participation in the election collection, obtain the motion vector of the prediction of each predicting unit, determine its motion vector MV in conjunction with the motion vector of prediction and the motion vector difference MVD of the predicting unit that obtains of decoding, according to this motion vector predicting unit is carried out motion compensation at last.
Preferably, the size that the same described motion-vector prediction participation in the election that described encoder is set up integrates is as the motion-vector prediction participation in the election collection of the predicting unit of 2N * 2N, and wherein, N=2k, k are the integer more than or equal to 2.
Preferably, described decoder carries out the derivation of advanced motion-vector prediction participation in the election collection concurrently, the motion vector that obtains predicting comprises: in case decode a coding unit, the parallel motion-vector prediction participation in the election collection that makes up all predicting unit of this coding unit inside.
Embodiment of the invention said method, Apparatus and system technical scheme have following beneficial effect: because same motion-vector prediction participation in the election collection, is set up in employing for a coding unit of coding side; All predicting unit in the described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast; The motion vector of the prediction that each predicting unit in the described coding unit generates according to this predicting unit respectively carries out the technological means of estimation, thereby realized the parallel of encoder-side predicting unit estimation, it is parallel to have realized that further decoding end predicting unit motion vector participation in the election collection makes up.
As shown in Figure 6, be the concurrency analysis schematic diagram of original coding end, adopt the above Parallelization Scheme of the embodiment of the invention to improve rear difference as shown in Figure 7, for the coding side concurrency after the embodiment of the invention improvement is analyzed schematic diagram.Wherein, MCL refers to merging patterns participation in the election collection (Merge Candidate List), MME is the estimation (Merge Motion Estimation) of merging patterns, AMVPCL is advanced motion-vector prediction participation in the election collection (AMVP Candidate List), and ME is estimation (Motion Estimation).In addition, the dotted portion among Fig. 6 only exists when predicting unit is 4 * 4.Among Fig. 6, it is larger that the longer expression of module changes complexity density.As can be seen from Figure 6, the complexity of estimation is maximum, if can allow the estimation executed in parallel of predicting unit, and the coding rate of turbo encoder to a great extent just.
Coding side for the predicting unit of 2Nx2N, at first checks to merge/skip (Merge/Skip) pattern, then checks estimation (Motion Estimation); For other big or small predicting unit, at first check to merge (Merge) pattern, then check estimation; And then inspection frame mode; Last percent of pass aberration optimizing is selected optimum coding mode.In the 8th meeting of JCT-VC, H0082 and H0240 have been adopted, so that the merging/skip mode of a plurality of predicting unit can executed in parallel.Merging/the skip mode of predicting unit and normal estimation can executed in parallel, because do not have correlation between these two kinds of patterns.But predicting unit is taken exercises in the process of estimating, at first the information of predicting unit is tried to achieve a motion-vector prediction participation in the election collection (MVP Candidate List around the basis, MVPCL), then from participation in the election collection (Candidate List), select a best contender's (Candidate) motion vector (MV) as the starting point of motion search, the motion vector that obtains according to motion search is at last concentrated from participating in the election of and is selected best utilization vector prediction (Motion Vector Prediction, MVP) as final motion-vector prediction.Because the structure of motion-vector prediction participation in the election collection has correlation between the predicting unit, the estimation of predicting unit can only be carried out in serial.
In the scheme after improvement, in certain parallel territory, the derivation of the advanced motion prediction participation in the election collection of adjacent predicting unit no longer interdepends, each predicting unit can be obtained motion-vector prediction participation in the election collection concurrently, thereby can walk abreast to such an extent that find the starting point of own motion search, carry out concurrently motion search, and final calculating concurrently found the solution final motion-vector prediction (MVP).Among Fig. 7, all predicting unit are used a cover motion-vector prediction participation in the election collection in the one and same coding unit, thus only need make up the motion-vector prediction participation in the election collection of 2Nx2N at coding side, thus can realize that predicting unit is parallel.
A coding unit adopts same motion-vector prediction participation in the election collection to refer to all predicting unit in the coding unit are adopted same set of motion-vector prediction participation in the election collection, namely size is the motion-vector prediction participation in the election collection of the predicting unit of 2N * 2N, specifically as shown in Figure 8, adopt same set of motion-vector prediction participation in the election collection schematic diagram for the embodiment of the invention 8 * 8 coding units.For a coding unit, only need to rebuild the motion-vector prediction participation in the election collection that size is the predicting unit of 2N * 2N, and need not every kind of predicting unit is all rebuild once.Do like this Parallel Implementation that not only can realize predicting unit in the coding unit, can also bring for coding side the reduction of complexity.The reduction of complexity is as shown in table 1 below, and wherein N is the reference frame number purpose summation in all reference frame lists.In the original coding device, the piece of a 64x64 if adopt the CU coding of 64x64, only needs 1 CU; To this CU, have 13 predicting unit (1 2Nx2N+2 * (2NxN+Nx2N+0.5Nx2N+1.5Nx2N+2Nx0.5N+2Nx1.5N)=13); To make up the motion-vector prediction reference set one time to each reference frame in each reference frame lists; So be 1*13*N motion-vector prediction reference set.Equally, the piece of a 64x64 if adopt the coding unit of 8x8 to remove coding, then has 64 coding units; The predicting unit of 8x8 has 9 predicting unit (1 2Nx2N+2 * (2NxN+Nx2N)+4 a * NxN)=9); So will make up altogether 64*9*N motion-vector prediction reference set.Adopt after the described scheme of the embodiment of the invention, a coding unit only need to make up a cover motion-vector prediction participation in the election collection, the i.e. motion-vector prediction of the predicting unit of 2Nx2N participation in the election collection.Adopt after the described scheme of the embodiment of the invention, if the coding unit of 8x8 adopts same set of motion-vector prediction participation in the election collection, then the structure number of times of motion-vector prediction participation in the election collection is 1/9 of original coding device.
The reduced complexity table of table 1 embodiment of the invention
Decoding end:
As shown in Figure 9, for the concurrency of original decoded end is analyzed schematic diagram, as shown in figure 10, improve rear decoding end concurrency for the embodiment of the invention and analyze schematic diagram.Dotted portion only just exists when the predicting unit size is 4 * 4.Wherein, AMVPCL is advanced motion-vector prediction participation in the election collection.In the decode procedure, at first carry out the entropy decoding, if the coding unit of decoding adopts interframe encoding mode and is not merging/skip mode, then carry out the derivation of advanced motion-vector prediction participation in the election collection, obtain the motion vector of the prediction of each predicting unit, determine its motion vector MV in conjunction with the motion vector of prediction and the motion vector difference MVD of the predicting unit that obtains of decoding, according to this motion vector predicting unit is carried out motion compensation at last.In the original decoded device, because the participation in the election collection of the predicting unit of inside, one and same coding unit is relevant, so the advanced motion-vector prediction participation in the election collection of each predicting unit must be derived according to the order of sequence.And in improving one's methods, the parallel territory of regulation is inner, and the participation in the election collection of predicting unit is no longer relevant, but also executed in parallel of the derivation of its participation in the election collection.In embodiments of the present invention, in case decode a coding unit, the motion-vector prediction participation in the election of all predicting unit of this coding unit inside collects the structure that can walk abreast.
Experimental result:
The embodiment of the invention realizes at the upper completing technology of HM5.0 (reference software of efficient video coding standard).In experiment, adopt the universal test condition, but DisableInter4x4 is made as at 0 o'clock.Experiment porch is Intel (R) Xeon (R) CPUX5660@2.80GHZ 2.79GHZ 23.9G internal memory.The coding unit of table 2 expression 8x8 adopts a cover motion-vector prediction participation in the election collection than the coding efficiency of original coding device.ClassA～E represent respectively resolution be 2560x1600,1920x1080,832x480,416x240,1280x720 cycle tests wherein the result of Class A only comprise the cycle tests of 8bit.Random Access HE, Random Access LC, Low delay B HE, Low delay B LC, Low delay P HE, the test result when Low delay PLC represents respectively to adopt different coding toolses.Y, U, V represent respectively a luminance component and two chromatic components.From experimental result, adopt embodiment of the invention such scheme, with respect to the original encoding device, when the same-code quality, only have 0.1% with interior code check lifting.So the square embodiment of the invention can both realize the parallel of encoding and decoding on the basis of loss coding performance hardly.
The coding unit of table 28 * 8 adopts the coding efficiency of same motion-vector prediction participation in the election collection
Those skilled in the art can also recognize the various illustrative components, blocks (illustrative logical block) that the embodiment of the invention is listed, the unit, and step can pass through electronic hardware, computer software, or both combinations realize.Be the clear replaceability (interchangeability) of showing hardware and software, above-mentioned various illustrative components (illustrative components), unit and step have been described their function generally.Such function is to realize depending on the designing requirement of specific application and whole system by hardware or software.Those skilled in the art can be for every kind of specific application, and can make ins all sorts of ways realizes described function, but this realization should not be understood to exceed the scope of embodiment of the invention protection.
Various illustrative logical block described in the embodiment of the invention, or the unit can pass through general processor, digital signal processor, application-specific integrated circuit (ASIC) (ASIC), field programmable gate array (FPGA) or other programmable logic device, discrete gate or transistor logic, discrete hardware components, or the design of above-mentioned any combination realizes or operates described function.General processor can be microprocessor, and alternatively, this general processor also can be any traditional processor, controller, microcontroller or state machine.Processor also can realize by the combination of calculation element, for example digital signal processor and microprocessor, multi-microprocessor, Digital Signal Processor Core of one or more microprocessors associatings, or any other similarly configuration realize.
Method described in the embodiment of the invention or the step of algorithm can directly embed hardware, the software module of processor execution or the two combination.Software module can be stored in the storage medium of other arbitrary form in RAM memory, flash memory, ROM memory, eprom memory, eeprom memory, register, hard disk, moveable magnetic disc, CD-ROM or this area.Exemplarily, storage medium can be connected with processor so that processor can be from storage medium reading information, and can deposit write information to storage medium.Alternatively, storage medium can also be integrated in the processor.Processor and storage medium can be arranged among the ASIC, and ASIC can be arranged in the user terminal.Alternatively, processor and storage medium also can be arranged in the different parts in the user terminal.
In one or more exemplary designs, the described above-mentioned functions of the embodiment of the invention can realize in hardware, software, firmware or this three's combination in any.If realize in software, these functions can be stored on the medium with computer-readable, or are transmitted on the medium of computer-readable with one or more instructions or code form.The computer-readable medium comprises the computer storage medium and is convenient to so that allow computer program transfer to other local communication medium from a place.Storage medium can be the useable medium that any general or special computer can access.For example, such computer readable media can include but not limited to RAM, ROM, EEPROM, CD-ROM or other optical disc storage, disk storage or other magnetic storage device, or other anyly can be used for carrying or storage can be read by general or special computer or general or special processor the program code of form with instruction or data structure and other medium.In addition, any connection can suitably be defined as the computer-readable medium, for example, if software is by a coaxial cable, optical fiber computer, twisted-pair feeder, Digital Subscriber Line (DSL) or also being comprised in the defined computer-readable medium with wireless way for transmittings such as infrared, wireless and microwave from a web-site, server or other remote resource.Described video disc (disk) and disk (disc) comprise Zip disk, radium-shine dish, CD, DVD, floppy disk and Blu-ray Disc, and disk is usually with the magnetic duplication data, and video disc carries out the optical reproduction data with laser usually.Above-mentioned combination also can be included in the computer-readable medium.
Above-described embodiment; purpose of the present invention, technical scheme and beneficial effect are further described; institute is understood that; the above only is the specific embodiment of the present invention; the protection range that is not intended to limit the present invention; within the spirit and principles in the present invention all, any modification of making, be equal to replacement, improvement etc., all should be included within protection scope of the present invention.
Claims (9)
1. the Parallel Implementation method of an advanced motion-vector prediction AMVP is characterized in that, the Parallel Implementation method of described AMVP comprises:
For a coding unit of coding side, set up same motion-vector prediction participation in the election collection;
All predicting unit in the described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast;
The motion vector of the prediction that each predicting unit in the described coding unit generates according to this predicting unit respectively carries out estimation.
2. the Parallel Implementation method of AMVP as claimed in claim 1 is characterized in that, the size that the same described motion-vector prediction participation in the election of foundation integrate collects as the motion-vector prediction participation in the election of the predicting unit of 2N * 2N, wherein, and N=2
k, k is the integer more than or equal to 2.
3. the Parallel Implementation method of AMVP as claimed in claim 1 is characterized in that,
For decoding end, in decode procedure, at first carry out the entropy decoding, if the coding unit of decoding adopts interframe encoding mode and is not merging/skip mode, then carry out the derivation of advanced motion-vector prediction participation in the election collection, obtain the motion vector of the prediction of each predicting unit, determine its motion vector MV in conjunction with the motion vector of prediction and the motion vector difference MVD of the predicting unit that obtains of decoding, according to this motion vector predicting unit is carried out motion compensation at last.
4. the Parallel Implementation method of AMVP as claimed in claim 3 is characterized in that, the described derivation of carrying out advanced motion-vector prediction participation in the election collection, and the motion vector that obtains predicting comprises:
In case decode a coding unit, the parallel motion-vector prediction participation in the election collection that makes up all predicting unit of this coding unit inside.
5. the Parallel Implementation device of an advanced motion-vector prediction AMVP is characterized in that, the Parallel Implementation device of described AMVP comprises:
Motion-vector prediction participation in the election collection is set up the unit, is used for the coding unit for coding side, sets up same motion-vector prediction participation in the election collection;
Motion vector prediction unit, all predicting unit that are used for described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast;
The motion vector of the prediction that motion estimation unit, each predicting unit that is used for described coding unit generate according to this predicting unit respectively carries out estimation.
6. the Parallel Implementation device of AMVP as claimed in claim 5 is characterized in that,
Described motion-vector prediction participation in the election collection is set up size that the same described motion-vector prediction participation in the election of setting up the unit integrate and is collected as the motion-vector prediction participation in the election of the predicting unit of 2N * 2N, wherein, and N=2
k, k is the integer more than or equal to 2.
7. the Parallel Implementation System of an advanced motion-vector prediction AMVP is characterized in that, the Parallel Implementation System of described AMVP comprises:
Encoder is used for the coding unit for coding side, sets up same motion-vector prediction participation in the election collection; All predicting unit in the described coding unit all adopt the same described motion-vector prediction participation in the election collection of foundation to carry out concurrently motion-vector prediction, respectively the motion vector of generation forecast; The motion vector of the prediction that each predicting unit in the described coding unit generates according to this predicting unit respectively carries out estimation;
Decoder, be used at decode procedure, at first carry out the entropy decoding, if the coding unit of decoding adopts interframe encoding mode and is not merging/skip mode, then carry out the derivation of advanced motion-vector prediction participation in the election collection, obtain the motion vector of the prediction of each predicting unit, determine its motion vector MV in conjunction with the motion vector of prediction and the motion vector difference MVD of the predicting unit that obtains of decoding, according to this motion vector predicting unit is carried out motion compensation at last.
8. the Parallel Implementation System of AMVP as claimed in claim 7 is characterized in that,
The size that the same described motion-vector prediction participation in the election that described encoder is set up integrates collects as the motion-vector prediction participation in the election of the predicting unit of 2N * 2N, wherein, and N=2
k, k is the integer more than or equal to 2.
9. the Parallel Implementation System of AMVP as claimed in claim 7 is characterized in that,
Described decoder carries out the derivation of advanced motion-vector prediction participation in the election collection concurrently, and the motion vector that obtains predicting comprises: in case decode a coding unit, the parallel motion-vector prediction participation in the election collection that makes up all predicting unit of this coding unit inside.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210111820.9A CN103379324B (en) | 2012-04-16 | 2012-04-16 | A kind of advanced motion-vector prediction AMVP Parallel Implementation method, apparatus and system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201210111820.9A CN103379324B (en) | 2012-04-16 | 2012-04-16 | A kind of advanced motion-vector prediction AMVP Parallel Implementation method, apparatus and system |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN103379324A true CN103379324A (en) | 2013-10-30 |
| CN103379324B CN103379324B (en) | 2017-12-15 |
Family
ID=49463819
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201210111820.9A Expired - Fee Related CN103379324B (en) | 2012-04-16 | 2012-04-16 | A kind of advanced motion-vector prediction AMVP Parallel Implementation method, apparatus and system |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103379324B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107396102A (en) * | 2017-08-30 | 2017-11-24 | 中南大学 | A kind of inter-frame mode fast selecting method and device based on Merge technological movement vectors |
Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101383971A (en) * | 2007-09-06 | 2009-03-11 | 北京中电华大电子设计有限责任公司 | Intra-frame prediction processing method based on image encoding and decoding |
-
2012
- 2012-04-16 CN CN201210111820.9A patent/CN103379324B/en not_active Expired - Fee Related
Patent Citations (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101383971A (en) * | 2007-09-06 | 2009-03-11 | 北京中电华大电子设计有限责任公司 | Intra-frame prediction processing method based on image encoding and decoding |
Non-Patent Citations (2)
| Title |
|---|
| MINHUA ZHOU: "Configurable and CU-group level parallel merge/skip", 《JOINT COLLABORATIVE TEAM ON VIDEO CODING (JCT-VC) OF ITU-T SG16 WP3 AND ISO/IEC JTC1/SC29/WG11 8TH MEETNING：SAN JOSE，CA, USA，1-10 FEBRUARY，2012》 * |
| TOSHIYASU SUGIO，AKAHIRO NISHI: "Non-CE9: Cross-check report of parallel merge candidate derivation for Inter_NxN partition type (JCTVC-H0091) by Panasonic", 《JOINT COLLABORATIVE TEAM ON VIDEO CODING (JCT-VC) OF ITU-T SG16 WP3 AND ISO/IEC JTC1/SC29/WG11 8TH MEETNING：SAN JOSE，CA，USA，1-10 FEBRUARY，2012》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107396102A (en) * | 2017-08-30 | 2017-11-24 | 中南大学 | A kind of inter-frame mode fast selecting method and device based on Merge technological movement vectors |
| CN107396102B (en) * | 2017-08-30 | 2019-10-08 | 中南大学 | A kind of inter-frame mode fast selecting method and device based on Merge technological movement vector |
Also Published As
| Publication number | Publication date |
|---|---|
| CN103379324B (en) | 2017-12-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| JP6728249B2 (en) | Image coding supporting block division and block integration | |
| JP6518274B2 (en) | Video decoding method and video coding method | |
| CN103503461B (en) | The method that picture is compiled code | |
| CN103380622B (en) | Intra prediction mode coding/decoding method and for its equipment | |
| CN102685474B (en) | Encoding and decoding method of prediction modes, encoding and decoding device and network system | |
| CN102948146A (en) | Video encoding method and video encoding apparatus and video decoding method and video decoding apparatus, which perform deblocking filtering based on tree-structure encoding units | |
| CN104012089A (en) | Method and apparatus for setting reference picture index of temporal merging candidate | |
| CN110446038A (en) | Method and apparatus for sharing candidate list | |
| KR20130030181A (en) | Method and apparatus for motion vector encoding/decoding using motion vector predictor | |
| Zhao et al. | Efficient realization of parallel HEVC intra encoding | |
| WO2021247906A1 (en) | Geometric partition mode with motion vector refinement | |
| CN108605123A (en) | The method and apparatus that video is coded and decoded by using prediction | |
| CN103379322A (en) | Parallel implementation method, device and system for advanced motion vector prediction AMVP | |
| CN103379324A (en) | Parallel realization method, device and system for advanced motion vector prediction AMVP | |
| CN110149512A (en) | Inter-prediction accelerated method, control device, electronic device, computer storage medium and equipment | |
| CN103379323A (en) | Parallel implementation method, device and system for advanced motion vector prediction AMVP | |
| CN104768014A (en) | Coding and decoding method | |
| JP2023105072A (en) | Signal transmission of global motion vector in picture header | |
| KR101711894B1 (en) | Method and apparatus for encoding video using coding information in upper depth |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20171215 Termination date: 20210416 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |