Adaptive quantization for transform-domain Wyner-Ziv residual coding

Abstract

As prediction processes are not in Wyner-Ziv encoder but in decoder, the Wyner-Ziv coding cannot perform better than conventional video encoder. In order to improve its compression efficiency, Wyner-Ziv residual coding has been proposed which implements a prediction process with low complexity in the encoder. Although it has a good performance in pixel domain, it does not have any improvements in transform domain, since pre-defined quantization matrices are not compatible with the Wyner-Ziv residual coding. In this paper, we propose a new quantization method for transform domain Wyner-Ziv residual coding. Experimental result shows up to 36% gain in BDBR and 3.09dB gain in BDPSNR.