Abstract
Distributed Multiview Video Coding (DMVC), also known as Multiview Distributed Video Coding (MDVC), is a new coding paradigm which applies the distributed video coding principles to multiview video coding. The performance of this new coding scheme strongly relies on the quality of the side information (SI), an estimation made by the decoder. In order to explore the temporal and interview correlations, the final SI is generated using an appropriate fusion technique between temporal SI and interview SI. In this paper, we propose a novel block-based fusion refinement algorithm for transform domain DMVC. First, a block-based fusion is applied based on different characteristics of the two SIs. Then, the fused SI is successively refined during decoding process using a reference frame which is reconstructed with already decoded coefficients. The experimental results show the proposed approach significantly improves the final SI as compared to exiting fusion approach, and is much close to ideal fusion.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
