Support Vector Machine based fusion for multi-view Distributed Video Coding

In multi-view Distributed Video Coding, Side Information can be computed either from previously decoded frames in the same view, or from previously decoded frames in adjacent views. In this paper, we address the problem of effectively fusing these two predictions. For this purpose, we propose two Support Vector Machine based fusion algorithms. We also identify a number of features to be used in classification. Experimental results show the efficiency of the proposed approach and its robustness over test sequences with greatly varying characteristics.