CN101360236B - Wyner-ziv video encoding and decoding method - Google Patents
Wyner-ziv video encoding and decoding method Download PDFInfo
- Publication number
- CN101360236B CN101360236B CN 200810063342 CN200810063342A CN101360236B CN 101360236 B CN101360236 B CN 101360236B CN 200810063342 CN200810063342 CN 200810063342 CN 200810063342 A CN200810063342 A CN 200810063342A CN 101360236 B CN101360236 B CN 101360236B
- Authority
- CN
- China
- Prior art keywords
- new images
- image
- pixel
- frame
- odd
- 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.)
- Expired - Fee Related
Links
Images
Landscapes
- Compression Or Coding Systems Of Tv Signals (AREA)
Abstract
The invention discloses a Wyner-Ziv video coding method. Reorganization is carried out on two adjacent front and back frame images on a time domain by utilizing an image reorganization method on a time axis to cause a key frame image and a Wyner-Ziv frame image after reorganized to respectively comprise partial information of the adjacent front and back frame images on the original input video signal time domain, enhance the correlation between the key frame image and a Wyner-Ziv frame image which are reorganized, cause a decoding end to get side information of the Wyner-Ziv frame image with higher quality by utilizing data interpolation of the prior key frame image, and enhance the coding compression performance of the Wyner-Ziv frame image, thereby reaching the purpose of enhancing the rate-distortion performance of coding and decoding of a Wyner-Ziv video; in addition, although the method increases processing such as image reorganization, inverse reorganization and image interpolation, motion estimation interpolation with big calculation amount is removed, therefore, corresponding to the prior Wyner-Ziv Codec, the increase of the coding and decoding complexity is limited.
Description
Technical field
The present invention relates to a kind of video coding and decoding technology, especially relate to a kind of Wyner-ziv video coding-decoding method.
Background technology
The Wyner-ziv video coding and decoding technology is a kind of emerging coding and decoding video mode that proposes in the world in recent years, and it can satisfy the video coding compression requirement of the low encoder complexity of some wireless terminal devices, low-power consumption, and therefore comparatively application prospects is arranged.Yet compression efficiency is low, reconstructed image quality is not high is the problem that current Wyner-ziv video coding and decoding technology is faced.Its main cause is: to carry out estimation at coding side different with the conventional video coding method, the Wyner-ziv method for video coding is implemented estimation in decoding end, owing to can not obtain the primitive frame of present frame in the estimation of decoding end, thereby the accuracy of estimation will be affected, make thereafter only by estimation interpolation (MEI, Motion Estimation Interpolation) side information and the original Wyner-ziv frame of the Wyner-ziv frame that goes out of technology interpolation differ bigger, decoding end has to ask coding side to send a large amount of check bits, thereby causes the distortion performance of Wyner-ziv video coding-decoding method not good.
Summary of the invention
Technical problem to be solved by this invention provides a kind of Wyner-ziv video coding-decoding method, it is by utilizing the image reorganization method on the time shaft, key frame images after the raising reorganization and the correlation between the Wyner-ziv two field picture make decoding end can utilize the data interpolating of existing key frame images to obtain the side information of the higher Wyner-ziv two field picture of quality.
The present invention solves the problems of the technologies described above the technical scheme that is adopted: a kind of Wyner-ziv video coding-decoding method, and it may further comprise the steps:
A. at coding side, front and back adjacent on the time domain two two field pictures are defined as previous moment image and back one time chart picture respectively, the previous moment image is designated as F
T-1, back one time chart looks like to be designated as F
tWith F
T-1Each locations of pixels information decomposition according to himself becomes F
T-1 1And F
T-1 2Two parts are with F
tEach locations of pixels information decomposition according to himself becomes F
t 1And F
t 2Two parts, again according to the image reorganization method with F
T-1 1And F
t 1, F
T-1 2And F
t 2Be reassembled as two frame new images respectively, with F
T-1 1And F
t 1The new images that reorganization obtains is designated as I
N-1, with F
T-1 2And F
t 2The new images that reorganization obtains is designated as I
nDefinition I
N-1Be key frame images and I
nBe the Wyner-ziv two field picture, with key frame images I
N-1With Wyner-ziv two field picture I
nSend into the Wyner-ziv encoder respectively, to key frame images I
N-1Carry out intraframe coding, to Wyner-ziv two field picture I
nCarry out Wyner-ziv frame coding, with Wyner-ziv two field picture I
nThe check bit that obtains behind the coding sends to decoding end when being buffered in coding side for the decoding end request; With the key frame images I behind the coding
N-1Give decoding end by Network Transmission;
B. in decoding end, with the key frame images I behind the coding that receives
N-1Obtain key frame images I by the Wyner-ziv decoder decode
N-1Decipher reestablishment image
Right then
Carry out low-pass filtering and obtain filtered reconstructed image
Right according to the image reorganization method that coding side adopted again
Carry out the bob-weight component from obtaining being included in previous moment image F
T-1In
Be included in the back one time chart as F
tIn
Adopt known image adaptive interpolation method right
With
Carrying out adaptive-interpolation respectively obtains and previous moment image F
T-1With the back one time chart as F
tThe identical F ' of resolution
T-1And F '
tAgain according to image reorganization method that coding side adopted with F '
T-1Resolve into F '
T-1 1And F '
T-1 2Two parts are with F '
tResolve into F '
t 1And F '
t 2Two parts are with F '
T-1 2And F '
t 2Be reassembled as I '
n, definition I '
nBe Wyner-ziv two field picture I
nSide information; Utilize key frame images I
N-1Decipher reestablishment image
Side information I '
nAnd decoding end obtains Wyner-ziv two field picture I to the check bit that the coding side request sends by the Wyner-ziv decoder decode
nDecipher reestablishment image
According to image reorganization method that coding side adopted to key frame images I
N-1Decipher reestablishment image
With Wyner-ziv two field picture I
nDecipher reestablishment image
Carry out the bob-weight component from obtaining previous moment image F
T-1Decipher reestablishment image
With the back one time chart as F
tDecipher reestablishment image
Described image reorganization method adopts any pattern in row image reorganization pattern, row image reorganization pattern, row interlaced video reorganization pattern, column interleaving image reorganization pattern, the pixel interleaving image reorganization pattern.
Described capable image reorganization pattern is a kind of in the following two class modes: 1) get previous moment image F
T-1With the back one time chart as F
tThe odd-numbered line pixel form a frame new images, get F
T-1And F
tThe even number line pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the last field of this new images, then take from F
tThe data of frame occupy the following field of this new images, if take from F
T-1The data of frame occupy the following field of this new images, then take from F
tThe data of frame occupy the last field of this new images; 2) get previous moment image F
T-1The odd-numbered line pixel and the back one time chart as F
tThe even number line pixel form a frame new images, get F
tOdd-numbered line pixel and F
T-1The even number line pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the last field of this new images, then take from F
tThe data of frame occupy the following field of this new images, if take from F
T-1The data of frame occupy the following field of this new images, then take from F
tThe data of frame occupy the last field of this new images.
Described row image reorganization pattern is a kind of in the following two class modes: 1) get previous moment image F
T-1With the back one time chart as F
tThe odd column pixel form a frame new images, get F
T-1And F
tThe even column pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the left field of this new images, then take from F
tThe data of frame occupy the right field of this new images, if take from F
T-1The data of frame occupy the right field of this new images, then take from F
tThe data of frame occupy the left field of this new images; 2) get previous moment image F
T-1The odd column pixel and the back one time chart as F
tThe even column pixel form a frame new images, get F
tOdd column pixel and F
T-1The even column pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the left field of this new images, then take from F
tThe data of frame occupy the right field of this new images, if take from F
T-1The data of frame occupy the right field of this new images, then take from F
tThe data of frame occupy the left field of this new images.
Described capable interlaced video reorganization pattern is: get previous moment image F
T-1With the back one time chart as F
tThe odd-numbered line pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe even number line pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1And F
tThe even number line pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe odd-numbered line pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1Odd-numbered line pixel and F
tThe even number line pixel, and with row alternately location mode form new images I
N-1, get F
tOdd-numbered line pixel and F
T-1The even number line pixel, and with row alternately location mode form new images I
nPerhaps, get F
tOdd-numbered line pixel and F
T-1The even number line pixel, and with row alternately location mode form new images I
N-1, get F
T-1Odd-numbered line pixel and F
tThe even number line pixel, and with row alternately location mode form new images I
n
Described column interleaving image reorganization pattern is: get previous moment image F
T-1With the back one time chart as F
tThe odd column pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe even column pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1And F
tThe even column pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe odd column pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1Odd column pixel and F
tThe even column pixel, and with row alternately location mode form new images I
N-1, get F
tOdd column pixel and F
T-1The even column pixel, and with row alternately location mode form new images I
nPerhaps, get F
tOdd column pixel and F
T-1The even column pixel, and with row alternately location mode form new images I
N-1, get F
T-1Odd column pixel and F
tThe even column pixel, and with row alternately location mode form new images I
n
Described pixel interleaving image reorganization pattern is: get previous moment image F
T-1Odd-numbered line in odd column pixel and the even column pixel in the even number line and get the back one time chart as F
tOdd-numbered line in even column pixel and the odd column pixel in the even number line form new images I
N-1, get F
T-1Odd-numbered line in even column pixel and the odd column pixel in the even number line and get F
tOdd-numbered line in odd column pixel and the even column pixel in the even number line form new images I
nPerhaps, get F
T-1Odd-numbered line in even column pixel and the odd column pixel in the even number line and get F
tOdd-numbered line in odd column pixel and the even column pixel in the even number line form new images I
N-1, get F
T-1Odd-numbered line in odd column pixel and the even column pixel in the even number line and get F
tOdd-numbered line in even column pixel and the odd column pixel in the even number line form new images I
n
Adaptive interpolation method among the described step b is: adopt known edge detection method to treat interpolation image and carry out rim detection and obtain edge direction; The pixel that defines in the current interpolation image of handling is a current pixel, when current pixel belongs to interpolation edge of image zone, according to edge direction this current pixel is carried out linear interpolation; When current pixel belongs to the non-fringe region of interpolation image, directly this current pixel is carried out bilinear interpolation or cube interpolation.
Compared with prior art, the invention has the advantages that the image reorganization method of utilizing on the time shaft recombinates to front and back adjacent on the time domain two two field pictures, the partial information of adjacent front and back two two field pictures on key frame images after the feasible reorganization and each self-contained former incoming video signal time domain of Wyner-Ziv two field picture, key frame images after the reorganization and correlation between the Wyner-ziv two field picture have been improved, make decoding end can utilize the data interpolating of existing key frame images to obtain the side information of the higher Wyner-ziv two field picture of quality, improve Wyner-Ziv two field picture encoding compression performance, thereby reach the purpose of the distortion performance that improves the Wyner-ziv coding and decoding video; In addition, though the inventive method has increased processing such as image reorganization, bob-weight group and image interpolation in decoding end, but removed the great estimation interpolation of amount of calculation from, therefore very limited with respect to its encoding and decoding complexity increase of original Wyner-ziv codec.
Description of drawings
Fig. 1 is the FB(flow block) of the inventive method;
Fig. 2 a is previous moment image F
T-1Schematic diagram;
Fig. 2 b is that back one time chart is as F
tSchematic diagram;
Fig. 3 a is for adopting wherein a kind of key frame images I that the reorganization of image reorganization pattern obtains that goes
N-1Schematic diagram;
Fig. 3 b is for adopting wherein a kind of Wyner-ziv two field picture I that the reorganization of image reorganization pattern obtains that goes
nSchematic diagram;
The key frame images I of Fig. 4 a for adopting the reorganization of another kind of row image reorganization pattern to obtain
N-1Schematic diagram;
The Wyner-ziv two field picture I of Fig. 4 b for adopting the reorganization of another kind of row image reorganization pattern to obtain
nSchematic diagram;
Fig. 5 a is for adopting wherein a kind of key frame images I that the reorganization of interlaced video reorganization pattern obtains that goes
N-1Schematic diagram;
Fig. 5 b is for adopting wherein a kind of Wyner-ziv two field picture I that the reorganization of interlaced video reorganization pattern obtains that goes
nSchematic diagram;
The key frame images I of Fig. 6 a for adopting the reorganization of another kind of row interlaced video reorganization pattern to obtain
N-1Schematic diagram;
The Wyner-ziv two field picture I of Fig. 6 b for adopting the reorganization of another kind of row interlaced video reorganization pattern to obtain
nSchematic diagram;
The key frame images I of Fig. 7 a for adopting the reorganization of pixel interleaving image reorganization pattern to obtain
N-1Schematic diagram;
The Wyner-ziv two field picture I of Fig. 7 b for adopting the reorganization of pixel interleaving image reorganization pattern to obtain
nSchematic diagram;
Fig. 8 a is the two field picture of the cycle tests " Foreman " of QCIF form in previous moment;
Fig. 8 b is the two field picture of the cycle tests " Foreman " of QCIF form in a back moment;
The new images of Fig. 8 c for adopting row image reorganization pattern to obtain to front and back two two field pictures adjacent on the time domain shown in Fig. 8 a and Fig. 8 b;
The new images of Fig. 8 d for adopting row interlaced video reorganization pattern to obtain to front and back two two field pictures adjacent on the time domain shown in Fig. 8 a and Fig. 8 b;
The new images of Fig. 8 e for adopting row image reorganization pattern to obtain to front and back two two field pictures adjacent on the time domain shown in Fig. 8 a and Fig. 8 b;
The new images of Fig. 8 f for adopting column interleaving image reorganization pattern to obtain to front and back two two field pictures adjacent on the time domain shown in Fig. 8 a and Fig. 8 b;
Fig. 9 is the concrete block diagram of FB(flow block) shown in Figure 1;
Figure 10 is for adopting the inventive method and the Wyner-ziv coding method of traditional 8 * 8 estimation interpolation generation side informations, the H.264 inner frame coding method and the comparison diagram of the interframe encode method distortion performance of encoding H.264 respectively to the cycle tests " Coastguard " of QCIF form;
Figure 11 is for adopting the inventive method and the Wyner-ziv coding method of traditional 8 * 8 estimation interpolation generation side informations, the H.264 inner frame coding method and the comparison diagram of the interframe encode method distortion performance of encoding H.264 respectively to the cycle tests " Foreman " of QCIF form.
Embodiment
Embodiment describes in further detail the present invention below in conjunction with accompanying drawing.
A kind of Wyner-ziv video coding-decoding method, as shown in Figure 1, it may further comprise the steps:
A. at coding side, at first, front and back adjacent on the time domain two two field pictures are defined as previous moment image and back one time chart picture respectively, the previous moment image is designated as F
T-1, back one time chart looks like to be designated as F
tWith previous moment image F
T-1Each locations of pixels information decomposition according to himself becomes F
T-1 1And F
T-1 2Two parts are incited somebody to action back one time chart as F
tEach locations of pixels information decomposition according to himself becomes F
t 1And F
t 2Two parts, again according to the image reorganization method with F
T-1 1And F
t 1, F
T-1 2And F
t 2Be reassembled as two frame new images respectively, with F
T-1 1And F
t 1The new images that reorganization obtains is designated as I
N-1, with F
T-1 2And F
t 2The new images that reorganization obtains is designated as I
nThen, definition I
N-1Be key frame images and I
nBe the Wyner-ziv two field picture, with key frame images I
N-1With Wyner-ziv two field picture I
nSend into the Wyner-ziv encoder respectively and encode, promptly to key frame images I
N-1Carry out intraframe coding, to Wyner-ziv two field picture I
nCarry out Wyner-ziv frame coding; With Wyner-ziv two field picture I
nThe check bit that obtains behind the coding sends to decoding end when being buffered in coding side for the decoding end request; At last, with the key frame images I behind the coding
N-1Give decoding end by Network Transmission.In this specific embodiment, to key frame images I
N-1Carry out intraframe coding and can adopt existing arbitrary frame intra coding method, as inner frame coding method H.264; To Wyner-ziv two field picture I
nCarry out Wyner-ziv frame coding and also can adopt existing any Wyner-ziv frame coding method.
Here, the image reorganization method can adopt any pattern in capable image reorganization pattern, row image reorganization pattern, row interlaced video reorganization pattern, column interleaving image reorganization pattern, the pixel interleaving image reorganization pattern.Below with the previous moment image F shown in Fig. 2 a
T-1With the back time chart shown in Fig. 2 b as F
tFor example is described in detail previous moment image F to the various patterns that the image reorganization method adopts
T-1With the back one time chart as F
tAdjacent on time domain.
Row image reorganization pattern is a kind of in the following two class modes: 1) get previous moment image F
T-1With the back one time chart as F
tThe odd-numbered line pixel form a frame new images, get F
T-1And F
tThe even number line pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the last field of this new images, then take from F
tThe data of frame occupy the following field of this new images, if take from F
T-1The data of frame occupy the following field of this new images, then take from F
tThe data of frame occupy the last field of this new images; 2) get previous moment image F
T-1The odd-numbered line pixel and the back one time chart as F
tThe even number line pixel form a frame new images, get F
tOdd-numbered line pixel and F
T-1The even number line pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the last field of this new images, then take from F
tThe data of frame occupy the following field of this new images, if take from F
T-1The data of frame occupy the following field of this new images, then take from F
tThe data of frame occupy the last field of this new images.Fig. 3 a and Fig. 3 b are respectively the F shown in Fig. 2 a
T-1With the F shown in Fig. 2 b
tAdopt the new images I that obtains after the reorganization of row image reorganization pattern
N-1And I
nSchematic diagram, the new images I after the reorganization shown in Fig. 3 a wherein
N-1Middle the first half is F
T-1Middle odd-numbered line pixel, the latter half is F
tMiddle odd-numbered line pixel, the new images I after the reorganization shown in Fig. 3 b
nMiddle the first half is F
T-1Middle even number line pixel, the latter half is F
tMiddle even number line pixel.Under the image reorganization of the being expert at pattern, also can shown in Fig. 4 a and Fig. 4 b, get F
T-1The odd-numbered line pixel and the F of frame
tThe even number line pixel of frame reassembles into I
N-1And I
nIn a frame, get F
tThe odd-numbered line pixel and the F of frame
T-1The even number line pixel of frame reassembles into I
N-1And I
nIn another frame.
Row image reorganization pattern and row image reorganization pattern similarity are only carried out image reorganization with the unit of classifying as rather than with behavior unit.Row image reorganization pattern is a kind of in the following two class modes: 1) get previous moment image F
T-1With the back one time chart as F
tThe odd column pixel form a frame new images, get F
T-1And F
tThe even column pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the left field of this new images, then take from F
tThe data of frame occupy the right field of this new images, if take from F
T-1The data of frame occupy the right field of this new images, then take from F
tThe data of frame occupy the left field of this new images; 2) get previous moment image F
T-1The odd column pixel and the back one time chart as F
tThe even column pixel form a frame new images, get F
tOdd column pixel and F
T-1The even column pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the left field of this new images, then take from F
tThe data of frame occupy the right field of this new images, if take from F
T-1The data of frame occupy the right field of this new images, then take from F
tThe data of frame occupy the left field of this new images.
Row interlaced video reorganization pattern is: get previous moment image F
T-1With the back one time chart as F
tThe odd-numbered line pixel, and with row alternately location mode form new images I
N-1, new images I
N-1Shown in Fig. 5 a, get F
T-1And F
tThe even number line pixel, and with row alternately location mode form new images I
n, new images I
nShown in Fig. 5 b; Perhaps, get F
T-1And F
tThe even number line pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe odd-numbered line pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1Odd-numbered line pixel and F
tThe even number line pixel, and with row alternately location mode form new images I
N-1, new images I
N-1Shown in Fig. 6 a, get F
tOdd-numbered line pixel and F
T-1The even number line pixel, and with row alternately location mode form new images I
n, new images I
nShown in Fig. 6 b; Perhaps, get F
tOdd-numbered line pixel and F
T-1The even number line pixel, and with row alternately location mode form new images I
N-1, get F
T-1Odd-numbered line pixel and F
tThe even number line pixel, and with row alternately location mode form new images I
nRow alternately location mode is: at new images I
N-1Or new images I
nIn, be F as if what deposit when previous row
T-1In delegation, then next line will be deposited F
tIn delegation; As if what deposit when previous row is F
tIn delegation, then next line will be deposited F
T-1In delegation.
Column interleaving image reorganization pattern and row interlaced video reorganization pattern similarity are only carried out image reorganization with the unit of classifying as rather than with behavior unit.Column interleaving image reorganization pattern is: get previous moment image F
T-1With the back one time chart as F
tThe odd column pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe even column pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1And F
tThe even column pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe odd column pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1Odd column pixel and F
tThe even column pixel, and with row alternately location mode form new images I
N-1, get F
tOdd column pixel and F
T-1The even column pixel, and with row alternately location mode form new images I
nPerhaps, get F
tOdd column pixel and F
T-1The even column pixel, and with row alternately location mode form new images I
N-1, get F
T-1Odd column pixel and F
tThe even column pixel, and with row alternately location mode form new images I
nRow alternately location mode are: at new images I
N-1Or new images I
nIn, be F as if what deposit when previous column
T-1In one row, then next column will be deposited F
tIn one row; As if what deposit when previous column is F
tIn one row, then next column will be deposited F
T-1In one row.
Pixel interleaving image reorganization pattern is: get previous moment image F
T-1Odd-numbered line in odd column pixel and the even column pixel in the even number line and get the back one time chart as F
tOdd-numbered line in even column pixel and the odd column pixel in the even number line form new images I
N-1, new images I
N-1Shown in Fig. 7 a, get F
T-1Odd-numbered line in even column pixel and the odd column pixel in the even number line and get F
tOdd-numbered line in odd column pixel and the even column pixel in the even number line form new images I
n, new images I
nShown in Fig. 7 b; Perhaps, get F
T-1Odd-numbered line in even column pixel and the odd column pixel in the even number line and get F
tOdd-numbered line in odd column pixel and the even column pixel in the even number line form new images I
N-1, get F
T-1Odd-numbered line in odd column pixel and the even column pixel in the even number line and get F
tOdd-numbered line in even column pixel and the odd column pixel in the even number line form new images I
nAt the new images I shown in Fig. 7 a and 7b
N-1And I
nIn, coordinate is that (x, pixel y) is taken from F in new images
T-1In coordinate be (x, during y) pixel, in the new images coordinate be (x-1, y), (x+1, y), (x, y-1), (x, pixel y+1) is taken from F respectively
tIn coordinate be (x-1, y), (x+1, y), (x, y-1), (x, pixel y+1); Equally, coordinate is that (x, pixel y) is taken from F in new images
tIn coordinate be (x, during y) pixel, in the new images coordinate be (x-1, y), (x+1, y), (x, y-1), (x, pixel y+1) is taken from F respectively
T-1In coordinate be (x-1, y), (x+1, y), (x, y-1), (x, pixel y+1).
Fig. 8 a and Fig. 8 b have provided front and back two two field pictures of the cycle tests " Foreman " of QCIF (Quarter common intermediate format, standardization video image format commonly used) form respectively.Fig. 8 c, Fig. 8 d, Fig. 8 e and Fig. 8 f have provided the image that adopts row image reorganization pattern, row interlaced video reorganization pattern, row image reorganization pattern and column interleaving image reorganization pattern to recombinate and obtain to the image shown in Fig. 8 a and Fig. 8 b respectively.
B. in decoding end, with the key frame images I behind the coding that receives
N-1Obtain key frame images I by the Wyner-ziv decoder decode
N-1Decipher reestablishment image
Right then
Carry out low-pass filtering and obtain filtered reconstructed image
Again according to the image reorganization method to filtered reconstructed image
Carry out the bob-weight component from obtaining being included in previous moment image F
T-1In
Be included in the back one time chart as F
tIn
Adopt known image adaptive interpolation method right
With
Carry out respectively that adaptive-interpolation obtains and F
T-1And F
tThe identical F ' of resolution
T-1And F '
tAgain according to the image reorganization method with F '
T-1Resolve into F '
T-1 1And F '
T-1 2Two parts are with F '
tResolve into F '
t 1And F '
t 2Two parts are with F '
T-1 2And F '
t 2Be reassembled as I '
n, definition I '
nBe Wyner-ziv two field picture I
nSide information; Utilize key frame images I
N-1Decipher reestablishment image
Side information I '
nAnd decoding end obtains Wyner-ziv two field picture I to the check bit that the coding side request sends by the Wyner-ziv decoder decode
nDecipher reestablishment image
According to the image reorganization method to key frame images I
N-1Decipher reestablishment image
With Wyner-ziv two field picture I
nDecipher reestablishment image
Carry out the bob-weight component from obtaining previous moment image F
T-1Decipher reestablishment image
With the back one time chart as F
tDecipher reestablishment image
Right in above-mentioned steps b
With
Can adopt existing arbitrary image interpolation algorithm to carry out interpolation, consider final interpolation effect, the present invention adopts known image adaptive interpolation method.The image adaptive interpolation method of Cai Yonging is in the present embodiment: adopt known edge detection method to treat interpolation image and carry out rim detection and obtain edge direction; The pixel that defines in the current interpolation image of handling is a current pixel, when current pixel belongs to interpolation edge of image zone, be that current pixel is when having edge pixel in the N * N neighborhood that is the center as edge pixel or with this current pixel, according to edge direction this current pixel is carried out linear interpolation, in the sampling point that participates in linear interpolation, reject inappropriate sampling point, can effectively avoid the edge blurry phenomenon; When current pixel belongs to the non-fringe region of interpolation image, directly this current pixel is carried out bilinear interpolation or cube interpolation; Wherein, N is a window size, and the size of window size N is relevant with the size of interpolation image, General N desirable 3 or 5 or 7 etc.When being carried out interpolation, the pixel of non-fringe region also can adopt other to be applicable to the interpolation method that smooth region interpolation and amount of calculation are less relatively.The inventive method adopts different interpolation methods to the pixel of fringe region and the pixel of non-fringe region, helps keeping when guaranteeing interpolation speed the clear of image border texture.
Edge detection method can be the edge detection algorithm of the maturation of any known, for example Sobel operator, Prewitt operator, Roberts operator, Canny operator etc.
In the present embodiment, the forming process of the check bit among the above-mentioned steps b as shown in Figure 9, at coding side, to the coding after key frame images I
N-1Decode and obtain key frame images I
N-1Decipher reestablishment image
With Wyner-ziv two field picture I
nSubtract each other and obtain residual image D; Obtain quantization parameter S behind residual image D process DCT (Discrete CosineTransform, the discrete cosine transform) transform and quantization, to each bit plane employing Turbo encoder encodes formation check bit of S.Check bit is stored in the buffering area of Wyner-ziv encoder usually.And in decoding end, F '
T-1 2And F '
t 2Be reassembled as side information I '
nAfter, with side information I '
nWith key frame images I
N-1Decipher reestablishment image
Subtract each other and obtain residual image D ', obtain quantization parameter S ' after D ' process dct transform and the quantification, each bit plane of S ' is sent into the Turbo decoder, the check bit that this Turbo decoder sends according to S ' and coding side begins to carry out iterative decoding from the highest order plane of S ', be reduced to pre-provisioning request until the decoding bit error rate, when residual image D and residual image D ' comparison near the time, the check bit that decoding end request coding side sends is less, on the contrary, the check bit that needs is more, so that obtain the reconstructed value of quantization parameter S comparatively accurately
Then the Turbo decoder decode is obtained
Carry out the residual image that inverse quantization and anti-dct transform obtain rebuilding
With the residual image of rebuilding
With key frame images I
N-1Decipher reestablishment image
Addition obtains Wyner-ziv two field picture I
nDecipher reestablishment image
At last to key frame images I
N-1Decipher reestablishment image
With Wyner-ziv two field picture I
nDecipher reestablishment image
Carry out the bob-weight group, finally obtain previous moment image F
T-1Decipher reestablishment image
With the back one time chart as F
tDecipher reestablishment image
Below for adopting the inventive method that cycle tests " Coastguard " and " Foreman " of QCIF form carried out the result that Wyner-ziv encodes, image sets GOP in the experiment (Group OfPicture) length is 2, frame per second is 10fps, and the coding frame number is 100 frames.
Figure 10 and Figure 11 have provided respectively and have adopted the inventive method (the image reorganization method adopts row image reorganization pattern), 8 * 8 traditional estimation interpolation to generate the Wyner-ziv coding method of side informations, H.264 inner frame coding method and the H.264 comparison of the interframe encode method distortion performance of encoding to cycle tests " Coastguard " and " Foreman ".H.264I, H.264P represent H.264 inner frame coding method and H.264 the distortion performance curve of interframe encode method when GOP=2 among Figure 10 and Figure 11 respectively, TWZC-8 * 8Block then represents to use 8 * 8 traditional estimation interpolation to generate the distortion performance curve of the Wyner-ziv coding method of side information.Experimental result shows that the inventive method and the Wyner-ziv coding method phase ratio distortion performance of using 8 * 8 traditional estimation interpolation to generate side information improve about 0.5dB, at inner frame coding method H.264 with H.264 between the distortion performance of interframe encode method, particularly for cycle tests " Coastguard ", the inventive method only differs about 0.1dB with the distortion performance of interframe encode method H.264.This shows that side information generation method of the present invention is quite effective, it has overcome the defective of Wyner-ziv coding method in the side information generative process that 8 * 8 traditional estimation interpolation generate side information effectively, has improved side information and has generated quality.Thereby Wyner-ziv video coding-decoding method of the present invention is compared whole distortion performance with the Wyner-ziv coding method that generates side information based on 8 * 8 traditional estimation interpolation and is improved a lot.
Claims (8)
1. Wyner-ziv video coding-decoding method is characterized in that it may further comprise the steps:
A. at coding side, front and back adjacent on the time domain two two field pictures are defined as previous moment image and back one time chart picture respectively, the previous moment image is designated as F
T-1, back one time chart looks like to be designated as F
tWith F
T-1Each locations of pixels information decomposition according to himself becomes F
T-1 1And F
T-1 2Two parts are with F
tEach locations of pixels information decomposition according to himself becomes F
t 1And F
t 2Two parts, again according to the image reorganization method with F
T-1 1And F
t 1, F
T-1 2And F
t 2Be reassembled as two frame new images respectively, with F
T-1 1And F
t 1The new images that reorganization obtains is designated as I
N-1, with F
T-1 2And F
t 2The new images that reorganization obtains is designated as I
nDefinition I
N-1Be key frame images and I
nBe the Wyner-ziv two field picture, with key frame images I
N-1With Wyner-ziv two field picture I
nSend into the Wyner-ziv encoder respectively, to key frame images I
N-1Carry out intraframe coding, to Wyner-ziv two field picture I
nCarry out Wyner-ziv frame coding, with Wyner-ziv two field picture I
nThe check bit that obtains behind the coding sends to decoding end when being buffered in coding side for the decoding end request; With the key frame images I behind the coding
N-1Give decoding end by Network Transmission;
B. in decoding end, with the key frame images I behind the coding that receives
N-1Obtain key frame images I by the Wyner-ziv decoder decode
N-1Decipher reestablishment image
Right then
Carry out low-pass filtering and obtain filtered reconstructed image
Right according to the image reorganization method that coding side adopted again
Carry out the bob-weight component from obtaining being included in previous moment image F
T-1In
Be included in the back one time chart as F
tIn
Adopt known image adaptive interpolation method right
With
Carrying out adaptive-interpolation respectively obtains and previous moment image F
T-1With the back one time chart as F
tThe identical F ' of resolution
T-1And F '
tAgain according to image reorganization method that coding side adopted with F '
T-1Resolve into F '
T-1 1And F '
T-1 2Two parts are with F '
tResolve into F '
t 1And F '
t 2Two parts are with F '
T-1 2And F '
t 2Be reassembled as I '
n, definition I '
nBe Wyner-ziv two field picture I
nSide information; Utilize key frame images I
N-1Decipher reestablishment image
Side information I '
nAnd decoding end obtains Wyner-ziv two field picture I to the check bit that the coding side request sends by the Wyner-ziv decoder decode
nDecipher reestablishment image
According to image reorganization method that coding side adopted to key frame images I
N-1Decipher reestablishment image
With Wyner-ziv two field picture I
nDecipher reestablishment image
Carry out the bob-weight component from obtaining previous moment image F
T-1Decipher reestablishment image
With the back one time chart as F
tDecipher reestablishment image
2. a kind of Wyner-ziv video coding-decoding method according to claim 1 is characterized in that described image reorganization method adopts any pattern in row image reorganization pattern, row image reorganization pattern, row interlaced video reorganization pattern, column interleaving image reorganization pattern, the pixel interleaving image reorganization pattern.
3. a kind of Wyner-ziv video coding-decoding method according to claim 2 is characterized in that described capable image reorganization pattern is a kind of in the following two class modes: 1) get previous moment image F
T-1With the back one time chart as F
tThe odd-numbered line pixel form a frame new images, get F
T-1And F
tThe even number line pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the last field of this new images, then take from F
tThe data of frame occupy the following field of this new images, if take from F
T-1The data of frame occupy the following field of this new images, then take from F
tThe data of frame occupy the last field of this new images; 2) get previous moment image F
T-1The odd-numbered line pixel and the back one time chart as F
tThe even number line pixel form a frame new images, get F
tOdd-numbered line pixel and F
T-1The even number line pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the last field of this new images, then take from F
tThe data of frame occupy the following field of this new images, if take from F
T-1The data of frame occupy the following field of this new images, then take from F
tThe data of frame occupy the last field of this new images.
4. a kind of Wyner-ziv video coding-decoding method according to claim 2 is characterized in that described row image reorganization pattern is a kind of in the following two class modes: 1) get previous moment image F
T-1With the back one time chart as F
tThe odd column pixel form a frame new images, get F
T-1And F
tThe even column pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the left field of this new images, then take from F
tThe data of frame occupy the right field of this new images, if take from F
T-1The data of frame occupy the right field of this new images, then take from F
tThe data of frame occupy the left field of this new images; 2) get previous moment image F
T-1The odd column pixel and the back one time chart as F
tThe even column pixel form a frame new images, get F
tOdd column pixel and F
T-1The even column pixel form another frame new images, the frame in this two frames new images is designated as I
N-1, another frame is designated as I
n, the new images I after the reorganization
N-1And I
nIn, if take from F
T-1The data of frame occupy the left field of this new images, then take from F
tThe data of frame occupy the right field of this new images, if take from F
T-1The data of frame occupy the right field of this new images, then take from F
tThe data of frame occupy the left field of this new images.
5. a kind of Wyner-ziv video coding-decoding method according to claim 2 is characterized in that described capable interlaced video reorganization pattern is: get previous moment image F
T-1With the back one time chart as F
tThe odd-numbered line pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe even number line pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1And F
tThe even number line pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe odd-numbered line pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1Odd-numbered line pixel and F
tThe even number line pixel, and with row alternately location mode form new images I
N-1, get F
tOdd-numbered line pixel and F
T-1The even number line pixel, and with row alternately location mode form new images I
nPerhaps, get F
tOdd-numbered line pixel and F
T-1The even number line pixel, and with row alternately location mode form new images I
N-1, get F
T-1Odd-numbered line pixel and F
tThe even number line pixel, and with row alternately location mode form new images I
n
6. a kind of Wyner-ziv video coding-decoding method according to claim 2 is characterized in that described column interleaving image reorganization pattern is: get previous moment image F
T-1With the back one time chart as F
tThe odd column pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe even column pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1And F
tThe even column pixel, and with row alternately location mode form new images I
N-1, get F
T-1And F
tThe odd column pixel, and with row alternately location mode form new images I
nPerhaps, get F
T-1Odd column pixel and F
tThe even column pixel, and with row alternately location mode form new images I
N-1, get F
tOdd column pixel and F
T-1The even column pixel, and with row alternately location mode form new images I
nPerhaps, get F
tOdd column pixel and F
T-1The even column pixel, and with row alternately location mode form new images I
N-1, get F
T-1Odd column pixel and F
tThe even column pixel, and with row alternately location mode form new images I
n
7. a kind of Wyner-ziv video coding-decoding method according to claim 2 is characterized in that described pixel interleaving image reorganization pattern is: get previous moment image F
T-1Odd-numbered line in odd column pixel and the even column pixel in the even number line and get the back one time chart as F
tOdd-numbered line in even column pixel and the odd column pixel in the even number line form new images I
N-1, get F
T-1Odd-numbered line in even column pixel and the odd column pixel in the even number line and get F
tOdd-numbered line in odd column pixel and the even column pixel in the even number line form new images I
nPerhaps, get F
T-1Odd-numbered line in even column pixel and the odd column pixel in the even number line and get F
tOdd-numbered line in odd column pixel and the even column pixel in the even number line form new images I
N-1, get F
T-1Odd-numbered line in odd column pixel and the even column pixel in the even number line and get F
tOdd-numbered line in even column pixel and the odd column pixel in the even number line form new images I
n
8. a kind of Wyner-ziv video coding-decoding method according to claim 1 is characterized in that the adaptive interpolation method among the described step b is: adopt known edge detection method to treat interpolation image and carry out rim detection and obtain edge direction; The pixel that defines in the current interpolation image of handling is a current pixel, when current pixel belongs to interpolation edge of image zone, according to edge direction this current pixel is carried out linear interpolation; When current pixel belongs to the non-fringe region of interpolation image, directly this current pixel is carried out bilinear interpolation or cube interpolation.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200810063342 CN101360236B (en) | 2008-08-08 | 2008-08-08 | Wyner-ziv video encoding and decoding method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN 200810063342 CN101360236B (en) | 2008-08-08 | 2008-08-08 | Wyner-ziv video encoding and decoding method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101360236A CN101360236A (en) | 2009-02-04 |
| CN101360236B true CN101360236B (en) | 2010-08-11 |
Family
ID=40332559
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN 200810063342 Expired - Fee Related CN101360236B (en) | 2008-08-08 | 2008-08-08 | Wyner-ziv video encoding and decoding method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101360236B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101835042B (en) * | 2010-03-19 | 2013-01-23 | 西安电子科技大学 | Wyner-Ziv video coding system controlled on the basis of non feedback speed rate and method |
| CN101854548B (en) * | 2010-05-25 | 2011-09-07 | 南京邮电大学 | Wireless multimedia sensor network-oriented video compression method |
| CN101977323B (en) * | 2010-11-16 | 2012-04-25 | 上海交通大学 | Method for reconstructing distributed video coding based on constraints on temporal-spatial correlation of video |
| CN102088612B (en) * | 2011-03-11 | 2012-07-25 | 西安电子科技大学 | Non-feedback Wyner-Ziv video decoder with robustness |
| CN102186077B (en) * | 2011-05-04 | 2012-12-26 | 西安电子科技大学 | Wyner-Ziv-video-coding-based Wyner-Ziv frame code rate control system and method |
-
2008
- 2008-08-08 CN CN 200810063342 patent/CN101360236B/en not_active Expired - Fee Related
Also Published As
| Publication number | Publication date |
|---|---|
| CN101360236A (en) | 2009-02-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| KR101972407B1 (en) | Apparatus and method for image coding and decoding | |
| RU2355126C2 (en) | Coding/decoding method based on layer set and video coder/decoder using smoothing prediction | |
| US10013746B2 (en) | High dynamic range video tone mapping | |
| US7469011B2 (en) | Escape mode code resizing for fields and slices | |
| WO2019242491A1 (en) | Video encoding and decoding method and device, computer device, and storage medium | |
| CN113923455B (en) | Bidirectional inter-frame prediction method and device | |
| WO1999057906A1 (en) | Method and apparatus for increasing video frame rate | |
| US7936824B2 (en) | Method for coding and decoding moving picture | |
| JP2012521734A (en) | Method and apparatus for selecting a filter | |
| JP2005123732A (en) | Apparatus and method for deblocking filter processing | |
| GB2509707A (en) | Encoding or decoding a video sequence using SAO parameters | |
| WO2008145039A1 (en) | Methods, systems and devices for generating upsample filter and downsample filter and for performing encoding | |
| WO2011110088A1 (en) | Method and divece for video predictive encoding | |
| CN1695381A (en) | Sharpness enhancement in post-processing of digital video signals using coding information and local spatial features | |
| KR20100024484A (en) | Joint coding of multiple transform blocks with reduced number of coefficients | |
| CN101360236B (en) | Wyner-ziv video encoding and decoding method | |
| CN110677654A (en) | Quantization parameter cascade method of high-efficiency video coding standard low-delay coding structure | |
| CN1878304A (en) | Method and device for coding and decoding a sequence of images | |
| Ma et al. | Rate-perception optimized preprocessing for video coding | |
| CN108353192B (en) | Hierarchical deblocking filtering in video processing systems and methods | |
| US20050259879A1 (en) | Video decoder with deblocker within decoding loop | |
| CN111757126A (en) | VR-oriented real-time image compression method, system and storage medium | |
| CN114827616A (en) | Compressed video quality enhancement method based on space-time information balance | |
| Gao et al. | Occupancy-Assisted Attribute Artifact Reduction for Video-Based Point Cloud Compression | |
| Guleryuz et al. | Sandwiched Compression: Repurposing Standard Codecs with Neural Network Wrappers |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| ASS | Succession or assignment of patent right |
Owner name: SHANGHAI SILICON INTELLECTUAL PROPERTY EXCHANGE CE Free format text: FORMER OWNER: NINGBO UNIVERSITY Effective date: 20120105 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 315211 NINGBO, ZHEJIANG PROVINCE TO: 200030 XUHUI, SHANGHAI |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20120105 Address after: 200030 Shanghai City No. 333 Yishan Road Huixin International Building 1 building 1704 Patentee after: Shanghai Silicon Intellectual Property Exchange Address before: 315211 Zhejiang Province, Ningbo Jiangbei District Fenghua Road No. 818 Patentee before: Ningbo University |
|
| ASS | Succession or assignment of patent right |
Owner name: SHANGHAI SIPAI KESI TECHNOLOGY CO., LTD. Free format text: FORMER OWNER: SHANGHAI SILICON INTELLECTUAL PROPERTY EXCHANGE CENTER CO., LTD. Effective date: 20120217 |
|
| C41 | Transfer of patent application or patent right or utility model | ||
| COR | Change of bibliographic data |
Free format text: CORRECT: ADDRESS; FROM: 200030 XUHUI, SHANGHAI TO: 201203 PUDONG NEW AREA, SHANGHAI |
|
| TR01 | Transfer of patent right |
Effective date of registration: 20120217 Address after: 201203 Shanghai Chunxiao Road No. 350 South Building Room 207 Patentee after: Shanghai spparks Technology Co. Ltd. Address before: 200030 Shanghai City No. 333 Yishan Road Huixin International Building 1 building 1704 Patentee before: Shanghai Silicon Intellectual Property Exchange |
|
| C56 | Change in the name or address of the patentee | ||
| CP02 | Change in the address of a patent holder |
Address after: 200030 Shanghai City No. 333 Yishan Road Huixin International Building 1 building 1706 Patentee after: Shanghai spparks Technology Co. Ltd. Address before: 201203 Shanghai Chunxiao Road No. 350 South Building Room 207 Patentee before: Shanghai spparks Technology Co. Ltd. |
|
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20100811 Termination date: 20190808 |
|
| CF01 | Termination of patent right due to non-payment of annual fee |