Abstract: In depth-based multi-view distributed video coding (DMDVC), the quality of side
information is one of the key factors that affect the compression efficiency. In order to make full use
of the temporal side information, spatial side information and depth map to improve the quality of
side information of the DMDVC system, a fusion algorithm based on reliability evaluation is
proposed. The motion compensated temporal interpolation (MCTI) is used to generate temporal side
information, and then the spatial side information can be generated using the depth image based
rendering (DIBR) technique. The object’s fast movement in the scene would affect the quality of
temporal side information and the use of DIBR to generate spatial domain side information would
lead to holes. To solve these problems, this paper selectively corrects the fast moving regions in
temporal side information or the holes and edge regions in spatial side information. The experimental
results show that the algorithm can effectively fuse the temporal side information and the spatial side
information, which can achieve up to 5.6 dB improvement over [4].

Key words: multi-view distributed video coding, side information fusion, reliability evaluation;
temporal-spatial side information