A novel reconstruction approach for pixel-domain distributed video coding

Abstract

Distributed Video Coding (DVC) is a new video coding which shifts the complexity from the encoder to the decoder. In DVC, the side information is only known at the decoder to exploit the source statistics. Therefore it has a lower-complexity encoder but higher-complexity decoder. The reconstruction function plays an important role in DVC which can improve the quality of video sequences. In the previous study, pixel values of video frames were considered as the continuous random variables, which is impractical. In this paper, we propose a novel reconstruction algorithm in which pixel values are thought as discrete random variables. We also present a method to model the correlation noise at the encoder to obtain a more accurate estimation of the correlation noise model (CNM). The experiment results show that the proposed reconstruction algorithm is comparable to the optimal reconstruction algorithm and can result in better rate distortion performance than the straightforward reconstruction algorithm.