Image interpolation with dense disparity estimation in multiview distributed video coding

Abstract

This paper deals with the side information (SI) generation problem in multiview distributed video coding (MDVC). For inter-view interpolation, we propose a novel dense disparity estimation (DE) approach combined with a popular distributed video coding (DVC) technique. As disparity vectors are computed at the decoder side, and no coding rate is needed to transmit them, dense estimation techniques are made possible, leading to improved results without requiring high coding resources. Experimental results show that the proposed interpolation technique can achieve up to 2.0 dB improvement in SI reconstruction performance, when compared to state-of-the-art DVC techniques.