A two-description video coding method based on Wyner Ziv theory

摘要

#$%^&*AU2012100852A620120802.pdf#####ABSTRACT: A Wyner-Ziv based two-description video coding method including the following steps: 1) Low-complexity encoding, i.e., odd/even frames splitting; zero-motion based H.264 coding for the first part of a description; residual SW-SPIHT coding for the second part of the description; redundancy optimizition; 2) High-complexity decoding, i.e., when two descriptions are received, central decoding is implemented; suppose a description is lost, e.g., only description 1 is received, its first part W', is obtained by zero-motion based H.264 decoding, then the reference frame We2 and side information Y2 are generated by interpolating, and the residual Dj=' -Wre 2 ; the restored residual D', is obtained by SW-SPIHT decoding, and W" 2 =D',+Wr2 , in the end, the entire restored video sequence V'i is obtained by incorporating W'1 and W"2 . This invention maintains a low-complexity encoding and robust transmission for video, and achieves better compression performance, especially when a description is lost, which can satisfy the requirement of video transmission over networks for low-power devices. (Drawing: Figure 4)2/4 Wyner-Ziv Wyner-Ziv -ZivInverse Decoded Frame Encoder Decoder Transform Wyner-Ziv Frame Y Transform Motion Estimation Decoded Frame Figure 3 Encoding Decoding 1 H.264 0-mv part 1 part 1 H.264 0-mv W side V1 even --0 fraes encoding description 1 decoding decoding 1 frarte2 part Ipart 2 W D , decoding & interpolating re inte latin & subtracting wI w op2 iigI2 D 0 9re" f u nd | W 1 Y2, D y 2 splittiSW-SPIHT SW-SPIHT central 0 encoding decoding decoding video 2 y1 T ri-Fe,1, De decoding & | interpolating interpolatingz -& subtracting odd part 2 O.part 2 w ' D2' frames H.264 0-mv description 2 H.264 0-m sid V'2 encodin part 1 part 1 decoding ':2 decodin 2 Figure 4