CN105847839A - Multi-view video coding strong filtering realization method for array structure - Google Patents
Multi-view video coding strong filtering realization method for array structure Download PDFInfo
- Publication number
- CN105847839A CN105847839A CN201510789499.3A CN201510789499A CN105847839A CN 105847839 A CN105847839 A CN 105847839A CN 201510789499 A CN201510789499 A CN 201510789499A CN 105847839 A CN105847839 A CN 105847839A
- Authority
- CN
- China
- Prior art keywords
- border
- filtering
- subgroup
- data
- block
- 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
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention relates to a multi-view video coding strong filtering realization method for an array structure. Specific to a deblocking filtering algorithm in an MVC multi-view video coding process, parallel analysis is carried out on deblocking filtering. On the basis that each macro block is composed of 16 4x4 blocks, strong filtering is carried out by using a 2x2 two-dimensional processing element array; when vertical boundary filtering of macro blocks is carried out, the 2x2 two-dimensional processing element array carries vertical boundary filtering at the same time. After vertical filtering is finished, the 2x2 two-dimensional processing element array carries out horizontal boundary filtering at the same time. According to the method, block effects resulting from conversion and quantization on the images are removed, the edges of the blocks are smooth, more data loading time is effectively saved, the coding and decoding time is shortened, and the coding and decoding processes are accelerated.
Description
Technical field
The present invention relates to technical field of video coding, particularly multiple view video coding (Multi-view Video
Coding, be called for short MVC) in deblocking effect strong filtering algorithm parallelization.
Technical background
The coding (Multi-view Video Coding is called for short MVC) of multi-view point video is long-range by International Telecommunication Union
The Video Coding Experts group (VCEG) of communication standardization tissue (ITU-T) and International Organization for Standardization/International Electrotechnical Commission
(ISO/IEC) extension to the most standardized technology (i.e. AVC) of the joint video team of dynamic image expert group composition.
MVC is that the multiple viewpoints in a video scene provide a kind of compact representation, the most multiple synchronization cameras.3-
Stereo pairing video visual for D is an important special case of MVC.This standard uses inter-view prediction to improve compressed capability,
Also support original time and spatial prediction simultaneously.MVC inherits excellent properties H.264: high-quality code efficiency, freely
Coding structure, good network compatibility, MVC also add the gradability viewpoint gradability of time simultaneously, and illumination is mended
Repay, the coding tools such as interview prediction, be highly suitable for the compressed encoding of three-dimensional video-frequency.
MVC uses block-based dct transform, quantization, motion compensation, and these technology all can inevitably be drawn at block boundary
Entering fast effect, have a strong impact on supervisor's mass of image, therefore, MVC standard uses Deblocking loop filter to remove blocking effect,
To improve subjective quality and the code efficiency of image.
After block elimination filtering is positioned at reconstruction in whole video decoding process, the macro block completing each decoding is swept by grating
Retouching order to be filtered, image boundary need not filter.And the operand of block elimination filtering constitutes about decoder and calculates the 1/3 of total amount.
Make one of bottleneck in decoder design.In addition, compared with single-view video.Multi-view point video is in different points of view
Capture Same Scene, but also to present the third dimension of 3D to spectators and to be subject to.Huge data volume required for this technology becomes
For restricting its another bottleneck wide variety of.Carry out block elimination filtering the most fast and effectively to reducing its computation complexity very
Important.
Summary of the invention
It is an object of the invention to for the problems referred to above, propose a kind of multiple view video coding for array structure and filter by force
Implementation method.The present invention can greatly reduce the complexity that Video coding calculates under the clear condition not reducing code efficiency.
For achieving the above object, the technology used in the present invention method is as follows: a kind of multiple views for array structure regards
Frequency encodes strong filter achieving method, according to the data dependence between block, at 2 × 2 two-dimensional process element array of adjacent interconnected
(PE0-PE3) on, the macro block (1-16) to 16 × 16 carries out the parallel processing that deblocking effect filters by force.
Use adjacent interconnected 2 × 2 two-dimensional process element array (PE0-PE3) load respectively 44 × 4 block boundaries (V0, V1,
V2, V3, H0, H1, H2, H3) 4 groups of both sides, often 4 subgroup totally 16 subgroups of group, 8 pixels of every subgroup (p3, p2, p1, p0,
Q0, q1, q2, q3), the deblocking effect being sequentially completed vertical boundary and 16 groups of pixels of horizontal boundary in a parallel fashion filters by force.
The filtering computational methods of its macro block vertical boundary: Processor Array (PE0-PE3) vertical to luminance component Y respectively
4 groups, border (V0) both sides, often 4 subgroup totally 16 subgroups, 8 pixels (p3, p2, p1, p0, q0, q1, q2, q3) of every subgroup of group
Carrying out data loading, wait data recordation to complete, carry out parallel filtering computing according to the data loaded respectively, computing to be filtered is complete
Cheng Hou, the filtering to the border (V0) of luminance component calculates complete;
Processor Array (PE0-PE1) is respectively to 2 groups of vertical boundary (V0) both sides of chromatic component U, often 4 subgroups of group totally 8
Subgroup, 8 pixels (p3, p2, p1, p0, q0, q1, q2, q3) of every subgroup carry out data loading, meanwhile, Processor Array
(PE2-PE3) respectively to 2 groups of vertical boundary (V0) both sides of chromatic component V, often 4 subgroup totally 8 subgroups, every subgroups 8 of group
Pixel (p3, p2, p1, p0, q0, q1, q2, q3) carries out data loading, waits data recordation to complete, respectively according to described number
According to carrying out parallel filtering calculating, after computing to be filtered completes, the filtering to the border (V0) of chromatic component calculates complete, whole
During 2 × 2 two-dimensional process element array (PE0-PE1) for BORDER PROCESSING to chromatic component U, and 2 × 2 two-dimensional process unit battle arrays
Row (PE2-PE3) are for the BORDER PROCESSING to chromatic component V, and immobilize;
When processing vertical boundary (V1) of luminance component Y, due to left to border (V1) in border (V0) processing procedure
The data of side (i.e. on the right side of the V0 of border) luminance block (1,5,9,13) load, so only need to load the brightness on right side, border (V1)
Block, and 4 groups of border (V1) both sides, often 4 subgroup totally 16 subgroups, 8 pixels of every subgroup of group are carried out parallel filtering calculating,
After computing to be filtered completes, the filtering to the border (V1) of luminance component calculates complete;
When processing vertical boundary (V2) of luminance component Y, due to left to border (V2) in border (V1) processing procedure
Side, the i.e. data of 4 block (2,6,10, the 14) pixels in border (V1) right side load, so it is right only need to load border (V2)
4 blocks (3,7,11,15) of side, and border (V2) both sides 16 subgroup, 8 pixels of every subgroup are carried out parallel filtering calculating,
After computing to be filtered completes, the filtering to the border (V2) of luminance component calculates complete;
Equally when processing chromatic component U, V respective vertical boundary (V1), only need to load 2 of right side, its border (V1) respectively
The data of block (2,4), have loaded etc. data, and 8 subgroups of boundaries on either side, 8 pixels of every subgroup are carried out also by corresponding PE
Row filtering calculates, and after calculating to be filtered completes, the border (V1) just completing component U, V filters complete;
In like manner, the border (V3) of residual components Y is filtered calculating according to the processing mode of border (V1);It is emphasized that
Whole filter sequence performs according to the order of defined in standard.
The filtering computational methods on macroblock level border: the filtering of horizontal boundary needs to use the filtered data of vertical boundary
Information, so just can carry out horizontal boundary filtering after needing the vertical boundary filtering of whole macro block to be done;Processing elements (PE0-PE3)
Respectively to 4 groups of vertical boundary (H0) both sides of luminance component Y, often 4 subgroup totally 16 subgroups, 8 pixels of every subgroup of group
(p3, p2, p1, p0, q0, q1, q2, q3) carries out data loading, waits data recordation to complete, and carries out according to the data loaded respectively
Parallel filtering computing, after computing to be filtered completes, the filtering to the border (H0) of luminance component calculates complete;
Processing elements (PE0-PE1) respectively to 2 groups of vertical boundary (H0) both sides of chromatic component U, often 4 subgroup totally 8 subgroups of group,
Every 8 pixels (p3, p2, p1, p0, q0, q1, q2, q3) of subgroup carry out data loading, meanwhile, processing elements (PE02-
PE03) respectively to 2 groups of vertical boundary (H0) both sides of chromatic component V, often 4 subgroup totally 8 subgroups, 8 pixels of every subgroup of group
Point (p3, p2, p1, p0, q0, q1, q2, q3) carries out data loading, waits data recordation to complete, and enters according to described data respectively
Row parallel filtering calculates, and after computing to be filtered completes, the filtering to the border (H0) of chromatic component calculates complete, in whole process
In 2 × 2 two-dimensional process element array (PE0-PE1) for BORDER PROCESSING to chromatic component U, and 2 × 2 two-dimensional process element array
(PE2-PE3) for the BORDER PROCESSING to chromatic component V, and immobilize;
When process luminance component Y vertical boundary (H1) time, due in border (H0) processing procedure on border (H1)
The data of side (i.e. on the downside of the H0 of border) luminance block (1,2,3,4) load, so only need to load the brightness of downside, border (H1)
Block (5,6,7,8), and 4 groups of border (H1) both sides, often 4 subgroup totally 16 subgroups, 8 pixels of every subgroup of group are carried out parallel
Filtering calculates, and after computing to be filtered completes, the filtering to the border (H1) of luminance component calculates complete;
When process luminance component Y vertical boundary (H2) time, due in border (H1) processing procedure on border (H2)
The data of side (i.e. on the downside of the H1 of border) luminance block (5,6,7,8) load, so only need to load the brightness of downside, border (H1)
Block (9,10,11,12), and 4 groups of border (H1) both sides, often 4 subgroup totally 16 subgroups, 8 pixels of every subgroup of group are carried out
Parallel filtering calculates, and after computing to be filtered completes, the filtering to the border (H1) of luminance component calculates complete;
Equally when processing chromatic component U, V respective vertical boundary (H1), only need to load 2 blocks of its (H1) downside respectively
The data of the pixel of (3,4), have loaded etc. data, the corresponding processing elements (PE) 8 subgroups to boundaries on either side, every subgroup
8 pixels carry out parallel filtering calculating, and after calculating to be filtered completes, the border H1 filtering just completing component U, V is complete;With
Reason, remaining border of luminance component Y (H3) is filtered calculating according to the processing mode of border (H1).
In like manner, the horizontal boundary (H3) of residue luminance component Y is filtered according to the processing mode of its horizontal boundary (H1)
Calculate;
The present invention proposes a kind of strong filter achieving method of the multiple view video coding for parallel organization, effectively saves more
Many data load times, accelerate coding and decoding process.Further, carry out block elimination filtering fast and effectively, also reduce calculating multiple
Miscellaneous degree.
Accompanying drawing explanation
Fig. 1 is 2 × 2 two-dimensional process element array of adjacent interconnected.
Fig. 2 is the macro block of 16 × 16.
Fig. 3 is the definition of block boundary.
Fig. 4 is the filter sequence of luminance component Y.
Fig. 5 is the filter sequence of chromatic component U.
Fig. 6 is the filter sequence of chromatic component V.
Detailed description of the invention
2 × 2 two-dimensional process element array PE0-PE3(Fig. 1), these 4 processing units interconnect the two of composition 2 × 2 by neighbour
Dimension Processor Array.
Macro block 1-16(Fig. 2), English Macroblock, is a basic conception in video coding technique.Compile at video
In Ma, a coded image is commonly divided into some macro blocks composition, and a macro block is by a luminance pixel block and two added
Chroma pixel block forms.In general, luminance block is the block of pixels of 16 × 16 sizes, and the size of two colourity block of image pixels
Depending on the sample format of its image, such as: for YUV420 sampled images, chrominance block is the block of pixels of 8 × 8 sizes.Each
In image, some macro blocks are arranged sheet of form, and video coding algorithm is in units of macro block, and macro block encodes one by one, group
It is made into continuous print video code flow.
In MVC image be the macro block with 16 × 16 sizes as unit, each macro block is made up of 16 4 × 4 size block,
It is likely to occur blocking effect, so needing each border of macro block is filtered between each two adjacent block.For under yuv format
Macro block, its macro block is made up of luminance component Y and chromatic component U/V.Macroblock boundaries (Fig. 3) is divided into vertical boundary and horizontal sides
Boundary, left side is luminance block (data block that luminance component is corresponding), and right side is chrominance block (data block that chromatic component is corresponding).V0-
V3 is the vertical boundary of luminance block, and H0-H3 is the horizontal boundary of luminance block, and pixel to be filtered is macroblock boundaries often row both sides
8 pixels (p3, p2, p1, p0, q0, q1, q2, q3).In like manner, V0, the V1 on right side are the vertical boundaries of chrominance block, H1, H2
Being the horizontal boundary of chrominance block, same, pixel to be filtered is 8 pixels of Mei Hang both sides, chrominance block border.Deblocking is filtered
Ripple algorithm is exactly to be filtered calculating to 8 pixels of macroblock boundaries often row every time.
The step that implements of the present invention is:
A: luminance component Y is filtered (Fig. 4, in figure, a-l is filter sequence), on the left of PE0 loading processing on the left of luminance block R0(
4 × 4 adjacent data blocks, belong to another macro block) and the first row of right side luminance block 1, the second row, the third line, fourth line pixel;
4 × 4 data blocks adjacent on the left of luminance block R1(on the left of PE1 loading processing, belong to another macro block) and the first of right side luminance block 5
Row, the second row, the third line, fourth line pixel;4 × 4 data blocks adjacent on the left of luminance block R2(on the left of PE2 loading processing, belong to
Another macro block) and the first row of right side luminance block 9, the second row, the third line, fourth line pixel;Luminance block on the left of PE3 loading processing
4 × 4 data blocks adjacent on the left of R3(, belong to another macro block) and the first row of right side luminance block 13, the second row, the third line,
Four row pixels.After having loaded etc. data, PE0-PE3 carries out parallel filtering computing according to the data loaded respectively, to be filtered
After computing completes, the filtering to the vertical boundary V0 of luminance component calculates complete.
B: to chromatic component U (Fig. 5, in figure, b-k is filter sequence), V(Fig. 6, in figure, b-k is filter sequence) it is filtered
Process.The most also being 4 PE parallel processings, PE0-PE1 processes the vertical boundary of chromatic component U;PE2-PE3 processes colourity and divides
The vertical boundary of amount V.And in whole filtering calculates, for processing U, the PE of V component immobilizes, the advantage of do so
It is: the data process to chromatic component U, V is completely unrelated, can separately process, increase degree of parallelism;And process colourity and divide
Having data to be correlated with between 2 corresponding PE of amount U or V, such as PE0 may use any one data in PE1, this
Sample, it is necessary to make PE0-PE1, PE2-PE3 fix one U of process or the data of V component, communicates between data to facilitate.
The chrominance block processing procedure of component U: with PE0 process on the left of chrominance block R0 and the first row of right chroma block 1, second
Row, the third line, fourth line pixel;PE1 process on the left of chrominance block R1 and the first row of right chroma block 3, the second row, the third line,
Fourth line pixel.The chrominance block processing procedure of component V: PE2 process on the left of chrominance block R0 and the first row of right chroma block 1, the
Two row, the third line, fourth line pixel;PE3 process on the left of chrominance block R1 and the first row of right chroma block 3, the second row, the 3rd
Row, fourth line pixel.It is all parallel that PE0-PE3 filtering calculates, and now, their colourity filtering flag position chromaFlag is
It is 1, and need not recalculate BS value, but the BS value of corresponding brightness component directly is replicated.Treat that 4 PE have calculated
Cheng Hou, just completes the filtering of each self-corresponding vertical boundary V0 of chromatic component U, V.
C: for luminance block (1,5,9,13), use data the most processed separately for PE0-PE3 in a, it is not necessary to again
Load.So now, PE0-PE3 only need to load right side block 2, the pixel of 6,10,14 respectively.PE0 loads right side respectively and works as
The first row of front piece 2, the second row, the third line, fourth line pixel;PE1 load on the right side of the first row of current block 6, the second row, the 3rd
Row, fourth line pixel;The first row of current block 10, the second row, the third line, fourth line pixel on the right side of PE2 loading;PE3 loads the right side
The first row of side current block 14, the second row, the third line, fourth line pixel.After having loaded etc. data, PE0-PE3 basis respectively
The data loaded are filtered computing, and after computing to be filtered completes, the filtering to the vertical boundary V1 of luminance component has calculated
Finish.
D: in like manner, PE0-PE3 load the data corresponding with processing vertical boundary V2 according to the mode identical with c, and advance
Row filtering calculates.After 4 PE have filtered, the filtering just completing vertical boundary V2 calculates.
E: now need the vertical boundary V1 of chromatic component U, V is filtered.Now, the chrominance block 1,3 of V, U component needs
The data wanted are the data in b after PE0-PE3 process, so there being these data in PE0-PE3 the most respectively.The most only need
PE0-PE1 loads each row of data of U component chrominance block 2,4 respectively, and PE2-PE3 loads each row of data of the block 2,4 of V i.e. respectively
Can.Now, after data loaded, 4 PE just can carry out parallel filtering calculating, and the operation between them is independent of each other.Treat 4
Individual PE calculates complete, i.e. completes the filtering of the vertical boundary V1 of U, V.
F: same, according to the processing mode identical with c, loads the vertical boundary V3 of luminance component Y respectively by PE0-PE3
Corresponding data, are filtered calculating parallel.After 4 PE have filtered, the filtering just completing vertical boundary V3 calculates.
G: now, the vertical filtering to macro block is complete, and needs to be filtered the horizontal boundary of macro block.PE0 process
Adjacent 4 × 4 data blocks on the upside of luminance block T0(of upside, belong to another macro block) and the first row of downside luminance block 1, secondary series,
3rd row, the 4th row pixel;4 × 4 data blocks adjacent on the upside of luminance block T1(on the upside of PE1 process, belong to another macro block) and under
The first row of side luminance block 2, secondary series, the 3rd row, the 4th row pixel;PE2 process on the upside of on the upside of luminance block T2(adjacent 4 × 4
Data block, belongs to another macro block) and the first row of downside luminance block 3, secondary series, the 3rd row, the 4th row pixel;In PE3 process
4 × 4 data blocks adjacent on the upside of luminance block T3(of side, belong to another macro block) and the first row of downside luminance block 4, secondary series,
Three row, the 4th row pixel.
Now, when processing brightness border H0, PE0 only need to load the data of luminance block T0 and can calculate, and
PE1, PE2, PE3 need to load luminance block T1, the data of T2, T3 and luminance block 2 respectively, block 3, the data of block 4 just can be carried out also
Row calculates.Luminance block 2, block 3, the value of block 4 all leave in PE0, so these numerical value can be read by shared depositor.When
The desired data of PE0-PE3 can be carried out concurrent working after being ready to, and then completes the filter to luminance component horizontal boundary H0
Ripple.
The vertical boundary filtering of h: chromatic component U, V is complete.Need its horizontal boundary is filtered.Identical,
Still it is filtered processing to chromatic component U with PE0-PE1;Chromatic component V is filtered processing by PE2-PE3.Now, PE0-
PE1 only need to load the T0 first row data to the 4th row, other 4 number corresponding needed for then obtaining from PE0-PE1 respectively
According to.And PE2-PE3 need to load the T0 first row data to the 4th row respectively, and from PE0-PE1, obtain other 4 numbers
According to.For processing the PE2-PE3 of chromatic component V, it is also such that its data load.After DSR, just can hold parallel
Row filtering, completes the filtering of the horizontal boundary H0 to chromatic component U, V.
I: with PE0-PE3 parallel processing H1.Same, with PE0 process luminance block 1 and the first row of block 5, secondary series, the
Three row, the 4th row pixel;Luminance block 2 and the first row of block 6, secondary series, the 3rd row, the 4th row pixel is processed with PE1;With
PE2 processes luminance block 3 and the first row of block 7, secondary series, the 3rd row, the 4th row pixel;Luminance block 4 and block 8 is processed with PE3
First row, secondary series, the 3rd row, the 4th row pixel.And now PE0-PE3 to have luminance block 1,2,3,4 the most respectively every
The pixel of row, so only need to obtain the pixel of luminance block 5,6,7,8 each column.When, after DSR, carrying out parallel
Filtering, thus complete the filtering to luminance component horizontal boundary H1.
J: luminance component horizontal boundary H2 is filtered.Its data stand-by mode is identical with i, same, with PE0 process
Luminance block 5 and the first row of block 9, secondary series, the 3rd row, the 4th row pixel;Luminance block 6 and the first of block 10 is processed with PE1
Row, secondary series, the 3rd row, the 4th row pixel;With PE2 process luminance block 7 and the first row of block 11, secondary series, the 3rd row, the
Four row pixels;Luminance block 8 and the first row of block 12, secondary series, the 3rd row, the 4th row pixel is processed with PE3.Now,
PE0-PE3 has the pixel of luminance block 5,6,7,8 each column the most respectively, so only need to obtain the pixel of block 9,10,11,12 each column
?.When, after DSR, being filtered parallel, thus complete the filtering to luminance component horizontal boundary H2.
K: the horizontal boundary H1 of chromatic component U, V is filtered.The of the chrominance block 1 of component U and block 3 is processed with PE0
String, secondary series, the 3rd row, the 4th row pixel;With PE1 process the chrominance block 2 of component U and the first row of block 4, secondary series,
3rd row, the 4th row pixel;Chrominance block 1 and the first row of block 3, secondary series, the 3rd row, the 4th row of component V are processed with PE2
Pixel;Chrominance block 2 and the first row of block 4, secondary series, the 3rd row, the 4th row pixel of component V is processed with PE3.Now
PE0-PE3 has had the pixel of chrominance block 1,2 each column, so only need to obtain the pixel of block 3,4 row.Work as DSR
After, can be filtered parallel, thus complete the filtering to chromatic component horizontal boundary H1.
L: luminance component horizontal boundary H3 is filtered.Its data stand-by mode is identical with j, same, with PE0 process
Luminance block 9 and the first row of block 13, secondary series, the 3rd row, the 4th row pixel;The of luminance block 10 and block 14 is processed with PE1
String, secondary series, the 3rd row, the 4th row pixel;With PE2 process luminance block 11 and the first row of block 15, secondary series, the 3rd
Row, the 4th row pixel;Luminance block 12 and the first row of block 16, secondary series, the 3rd row, the 4th row pixel is processed with PE3.This
Time, PE0-PE3 has the pixel of block 9,10,11,12 each column the most respectively, so only need to obtain block 13,14,15,16 each column
Pixel.When, after DSR, being filtered parallel, thus complete the filter to luminance component horizontal boundary H3
Ripple.
Now, deblocking effect filters by force and completes.
It is last that it is noted that obviously above-mentioned enforcement is only for clearly demonstrating example of the present invention, but not
Therefore the restriction to the scope of the claims of the present invention can be interpreted as.It should be noted that come for those of ordinary skill in the art
Say, without departing from the inventive concept of the premise, it is also possible to making deformation and improve, these broadly fall into the protection model of the present invention
Enclose.Therefore, the protection domain of patent of the present invention should be as the criterion with claims.
Claims (4)
1. the strong filter achieving method of multiple view video coding for array structure, it is characterised in that: between block
Data dependence, is carried out the macro block (1-16) of 16 × 16 on 2 × 2 two-dimensional process element array (PE0-PE3) of adjacent interconnected
The parallel processing that deblocking effect filters by force.
A kind of strong filter achieving method of the multiple view video coding for array structure the most according to claim 1, it is special
Levy and be: use adjacent interconnected 2 × 2 two-dimensional process element array (PE0-PE3) load respectively 44 × 4 block boundaries (V0, V1,
V2, V3, H0, H1, H2, H3) 4 groups, both sides, often 4 subgroup totally 16 subgroups of group, 8 pixels of every subgroup (p3, p2, p1, p0, q0,
Q1, q2, q3), the deblocking effect being sequentially completed vertical boundary and horizontal boundary 16 sub-group of pixels point in a parallel fashion filters by force.
A kind of strong filter achieving method of the multiple view video coding for array structure the most according to claim 2, it is special
Levy and be: the filtering computational methods of its macro block vertical boundary: 2 × 2 two-dimensional process element array (PE0-PE3) are respectively to luminance component
4 groups of vertical boundary (V0) both sides of Y, often 4 subgroup totally 16 subgroups of group, 8 pixels of every subgroup (p3, p2, p1, p0, q0,
Q1, q2, q3) carry out data loading, wait data recordation to complete, carry out parallel filtering computing according to the data loaded respectively, treat
After filtering operation completes, the filtering to the vertical boundary (V0) of luminance component Y calculates complete;2 × 2 two-dimensional process element array
(PE0-PE1) respectively to 2 groups of vertical boundary (V0) both sides of chromatic component U, often 4 subgroup totally 8 subgroups, every subgroups 8 of group
Pixel (p3, p2, p1, p0, q0, q1, q2, q3) carries out data loading, meanwhile, 2 × 2 two-dimensional process element array (PE2-
PE3) respectively to 2 groups of vertical boundary (V0) both sides of chromatic component V, often 4 subgroup totally 8 subgroups, 8 pixels of every subgroup of group
(p3, p2, p1, p0, q0, q1, q2, q3) carries out data loading, waits data recordation to complete, and carries out according to described data respectively
Parallel filtering calculates, and after computing to be filtered completes, the filtering to the border (V0) of chromatic component calculates complete, during whole
2 × 2 two-dimensional process element array (PE0-PE1) are for the BORDER PROCESSING to chromatic component U, and 2 × 2 two-dimensional process element array
(PE2-PE3) for the BORDER PROCESSING to chromatic component V, and immobilize;When the vertical boundary (V1) processing luminance component Y
Time, due in border (V0) processing procedure to border (V1) left side, i.e. 4 blocks (1,5,9,13) in border (V0) right side as
The data of vegetarian refreshments load, so only need to load 4 blocks (2,6,10,14) on right side, border (V1), and to border (V1) two
Side 16 subgroup, 8 pixels of every subgroup carry out parallel filtering calculating, the border after computing to be filtered completes, to luminance component
(V1) filtering calculates complete;When processing vertical boundary (V2) of luminance component Y, due in border (V1) processing procedure
To border (V2) left side, i.e. the data of 4 block (2,6,10, the 14) pixels in border (V1) right side load, so only
4 blocks (3,7,11,15) on right side, border (V2) need to be loaded, and border (V2) both sides 16 subgroup, 8 pixels of every subgroup are clicked on
Row parallel filtering calculates, and after computing to be filtered completes, the filtering to the border (V2) of luminance component calculates complete;Equally at place
During reason chromatic component U, V respective vertical boundary (V1), only need to load the number of 2 blocks (2,4) on right side, its border (V1) respectively
According to, loading etc. data, corresponding PE carries out parallel filtering calculating to 8 subgroups of boundaries on either side, 8 pixels of every subgroup,
After calculating to be filtered completes, the border (V1) just completing component U, V filters complete;In like manner, remaining border of luminance component Y
(V3) it is filtered calculating according to the processing mode of border (V1).
A kind of strong filter achieving method of the multiple view video coding for array structure the most according to claim 2, it is special
Levy and be: the filtering computational methods on macroblock level border: the filtering of horizontal boundary needs to use the filtered data of vertical boundary
Information, so just can carry out horizontal boundary filtering after needing the vertical boundary filtering of whole macro block to be done;2 × 2 two-dimensional process units
Array (PE0-PE3) is respectively to 4 groups of vertical boundary (H0) both sides of luminance component Y, often 4 subgroup totally 16 subgroups of group, often son
Organize 8 pixels (p3, p2, p1, p0, q0, q1, q2, q3) and carry out data loading, wait data recordation to complete, respectively according to added
The data carried carry out parallel filtering computing, and after computing to be filtered completes, the filtering to the border (H0) of luminance component Y has calculated
Finish;2 × 2 two-dimensional process element array (PE0-PE1) are respectively to 2 groups of vertical boundary (H0) both sides of chromatic component U, often group 4
Subgroup totally 8 subgroups, 8 pixels (p3, p2, p1, p0, q0, q1, q2, q3) of every subgroup carry out data loading, meanwhile, 2 ×
2 two-dimensional process element array (PE2-PE3) are respectively to 2 groups of vertical boundary (H0) both sides of chromatic component V, often 4 subgroups of group totally 8
Subgroup, 8 pixels (p3, p2, p1, p0, q0, q1, q2, q3) of every subgroup carry out data loading, wait data recordation to complete, respectively
Parallel filtering calculating is carried out, the filtering after computing to be filtered completes, to the border (H0) of chromatic component according to described data
Calculating complete, during whole, 2 × 2 two-dimensional process element array (PE0-PE1) are for the BORDER PROCESSING to chromatic component U, and 2
× 2 two-dimensional process element array (PE2-PE3) are for the BORDER PROCESSING to chromatic component V, and immobilize;When processing luminance component
During vertical boundary (H1) of Y, due in border (H0) processing procedure to border (H1) upside, i.e. border (H0) downside, 4
The data of individual block (1,2,3,4) load, so only need to load the pixel of 4 blocks (5,6,7,8) of downside, border (H1)
Point, and border (H1) both sides 16 subgroup, 8 pixels of every subgroup are carried out parallel filtering calculating, after computing to be filtered completes, right
The filtering on the border (H1) of luminance component calculates complete;When processing vertical boundary (H2) of luminance component Y, due on border
(H1) to border (H2) upside in processing procedure, i.e. border (H1) downside, the data of 4 blocks (5,6,7,8) load,
So the pixel of 4 blocks (9,10,11,12) of downside, border (H1) only need to be loaded, and to border (H1) both sides 16 subgroup, every
8 pixels of subgroup carry out parallel filtering calculating, and after computing to be filtered completes, the filtering to the border (H1) of luminance component calculates
Complete;Equally when processing chromatic component U, V respective vertical boundary (H1), only need to load 2 blocks of its (H1) downside respectively
The data of the pixel of (3,4), have loaded etc. data, the corresponding processing elements (PE) 8 subgroups to boundaries on either side, every subgroup
8 pixels carry out parallel filtering calculating, and after calculating to be filtered completes, the border H1 filtering just completing component U, V is complete;With
Reason, remaining border of luminance component Y (H3) is filtered calculating according to the processing mode of border (H1).
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510789499.3A CN105847839B (en) | 2015-11-17 | 2015-11-17 | A kind of strong filter achieving method of multiple view video coding for array structure |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201510789499.3A CN105847839B (en) | 2015-11-17 | 2015-11-17 | A kind of strong filter achieving method of multiple view video coding for array structure |
Publications (2)
Publication Number | Publication Date |
---|---|
CN105847839A true CN105847839A (en) | 2016-08-10 |
CN105847839B CN105847839B (en) | 2018-11-16 |
Family
ID=56580454
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201510789499.3A Active CN105847839B (en) | 2015-11-17 | 2015-11-17 | A kind of strong filter achieving method of multiple view video coding for array structure |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN105847839B (en) |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2008067500A2 (en) * | 2006-11-29 | 2008-06-05 | Novafora, Inc. | Parallel deblocking filter for h.264 video codec |
CN101304526A (en) * | 2007-05-09 | 2008-11-12 | 安凯(广州)软件技术有限公司 | Parallel processing method for de-locking filtering in video encoding/decoding system |
CN101951519A (en) * | 2010-10-12 | 2011-01-19 | 西安电子科技大学 | High speed de-blocking effect filtering method |
CN102572430A (en) * | 2011-12-29 | 2012-07-11 | 东南大学 | Method for implementing H.264 deblocking filter algorithm based on reconfigurable technique |
CN103957425A (en) * | 2014-04-23 | 2014-07-30 | 华为技术有限公司 | Image processing method and system |
-
2015
- 2015-11-17 CN CN201510789499.3A patent/CN105847839B/en active Active
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
WO2008067500A2 (en) * | 2006-11-29 | 2008-06-05 | Novafora, Inc. | Parallel deblocking filter for h.264 video codec |
CN101304526A (en) * | 2007-05-09 | 2008-11-12 | 安凯(广州)软件技术有限公司 | Parallel processing method for de-locking filtering in video encoding/decoding system |
CN101951519A (en) * | 2010-10-12 | 2011-01-19 | 西安电子科技大学 | High speed de-blocking effect filtering method |
CN102572430A (en) * | 2011-12-29 | 2012-07-11 | 东南大学 | Method for implementing H.264 deblocking filter algorithm based on reconfigurable technique |
CN103957425A (en) * | 2014-04-23 | 2014-07-30 | 华为技术有限公司 | Image processing method and system |
Also Published As
Publication number | Publication date |
---|---|
CN105847839B (en) | 2018-11-16 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
KR102254599B1 (en) | Method predicting view synthesis in multi-view video coding and method for constituting merge candidate list by using same | |
US8923407B2 (en) | Sample adaptive offset (SAO) parameter signaling | |
CN104412587B (en) | Method and apparatus used for 3d video coding or decoding system | |
CN106464894B (en) | Method for processing video frequency and device | |
DE112012001609B9 (en) | Intra prediction methods, encoders and decoders for using the same | |
CN105898335A (en) | Method and apparatus of loop filters for efficient hardware implementation | |
CN105191317B (en) | The predictive interpretation of depth look-up table in view and across view | |
CN106797468B (en) | Efficient video coding (HEVC) screen content encodes improved pallet mode in (SCC) | |
CN102685532B (en) | Coding method for free view point four-dimensional space video coding system | |
TW202007166A (en) | Method of video decoding at a decoder and apparatus thereof | |
CN100571391C (en) | The method for parallel processing that is used for the video coding and decoding system de-blocking filter | |
CN108293119A (en) | Modification in video coding for the transformation coefficient of non-square converter unit | |
CN104350752B (en) | The device of intra-loop filtering for the lossless coding mode in high-performance video coding | |
CN110089113A (en) | Image coding/decoding method, equipment and the recording medium for stored bits stream | |
JP2017512440A (en) | Improved depth recognition for stereo video | |
CN105874788B (en) | The simplification decoded to the piecewise DC of larger prediction block in 3D video codings | |
WO2008072847A1 (en) | Derivation process of boundary filtering strength, and deblocking filtering method and apparatus using the derivation process | |
CN106464907A (en) | Apparatus and method for eliminating redundancy of view synthesis prediction candidate in merge mode | |
EP2846544A1 (en) | Method and apparatus for encoding multi-view images, and method and apparatus for decoding multi-view images | |
CN100581265C (en) | Processing method for multi-view point video | |
US11838639B2 (en) | Image data encoding/decoding method and apparatus | |
GB2459568A (en) | Overlap transform and de-blocking of decompressed video signal using an edge filter on sub-blocks for upper and left macroblocks | |
CN105245905B (en) | A kind of strong filter achieving method of multiple view video coding for parallel organization | |
CN106105195A (en) | Method and device for the single sample mode of Video coding | |
CN102710949A (en) | Visual sensation-based stereo video coding method |
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 |