Wyner-Ziv residual coding of video

Abstract

In Wyner-Ziv coding of video, a frame is encoded independent of other frames, but decoded using adjacent frames as side information. In this work we extend our system by Wyner-Ziv encoding the residual of a frame with respect to a known reference at the encoder. At the decoder, better side information is generated through motion estimation and is used to reconstruct the frame. We show through experimental results that by allowing the encoder this additional complexity of frame store and frame subtraction, we achieve better rate-distortion performance compared to previous Wyner-Ziv coding schemes.