JOINT DECODING AND CORRELATION CHANNEL PARAMETER ESTIMATION IN WYNER-ZIV VIDEO CODING

Abstract — In Wyner-Ziv (WZ) video coding, or coding of video with decoder side information (SI), the correlation between the source and its SI at the decoder side is modeled as a virtual correlation channel, and decoding of the quantized source requires the estimation of the correlation channel parameter. In this paper, we present a joint decoding and correlation channel parameter estimation approach. In the approach, starting from an initial correlation channel parameter, channel decoding and correlation channel parameter estimation are carried out iteratively. In each iteration, channel decoding of a bit plane of the quantized source is carried out first. The probability that a channel output bit is different from the corresponding one in the SI is then collected. The correlation channel parameter is numerically estimated afterwards by solving the equation between the probability collected and another probability calculated with the SI, the already decoded bit planes of the quantized source, and the correlation channel parameter. The process is repeated till a bit plane of the quantized source is correctly decoded. Simulation results show that the proposed approach obtains very promising results.