Confidence features extraction for Wyner-Ziv video decoding

Distributed video coding (DVC) is well suited for low complexity encoding and error robust transmission. The feedback channel is the main handicap of state-of-the-art DVC systems. We previously proposed error locating coding (ELC) in conjunction with image inpainting as solution. This paper is focused on ELC with confidence information. It is shown that the proposed confidence feature improves the ELC performance and adapts well to the data rate.