Rate-distortion optimised quantisation for direct current coefficients of Wyner–Ziv frame in unidirectional distributed video coding

Abstract

This study presents a rate-distortion (RD) optimised quantisation algorithm for direct current (DC) coefficients of the Wyner–Ziv (WZ) frames in unidirectional distributed video coding. As the side information (SI) could be viewed as a ‘corrupted’ version of the original WZ frame, the correlation noise (WZ-SI) can be interpreted as a virtual channel with an error pattern. In this study, the distortion caused by virtual channel errors, besides quantisation errors, is analysed emphatically for unidirectional WZ video coding. In their presented quantisation algorithm, a novel RD model is developed to find out the optimal quantisation step size, in which the quantisation distortion and virtual channel error distortion together with the encoding rate are formulated using the quantisation step size without feedback channels. Then the RD optimal step size for DC coefficients can be solved through minimising the RD cost. In simulations, the comparison of RD performance between their proposed algorithm and the quantisation method used in distributed coding for video services (DISCOVER) is presented. Experimental results show that the proposed RD optimised quantisation for DC coefficients can improve the average peak signal-to-noise ratio of WZ frames by 0.4–1.2 dB for quarter common intermediate format (QCIF) sequences and 0.2–0.4 dB for common intermediate format (CIF), at the expense of increasing the encoding complexity slightly.