Abstract—This article introduces an approach for the acquisition and coding of multiview video. Multiview video systems consist of several cameras simultaneously capturing a single scene. Therefore a significant level of inter-view redundancy is present among the videos that can be exploited into video compression. Inspired by the idea of motion estimation in MPEG4 video compression, we introduce the idea of rotation estimation and compensation that is used in conjunction with motion estimation and compensation in order to remove spacial as well as temporal redundancies from the compressed video. The main question to be answered is how to choose the best sequence of compression among the frames when both time and space domains are involved. In this article, we model the above problem as a minimum cost graph traversal problem where cameras are considered as graph nodes and the cost of an edge connecting two cameras is inversely proportional to the similarity between the videos recorded by those cameras. We will then find the solution of this problem as the optimal traversal sequence that result in a high compression ratio.

Index Terms—Multiview video, 3D video, orientational redundancy.

The authors are with the Distributed and Collaborative Virtual Environment Research Laboratory (DISCOVER Lab), University of Ottawa, Ottawa, Canada.

Cite:Chi Wa Leong, Behnoosh Hariri, Shervin Shirmohammadi, "Exploiting Orientational Redundancy in Multiview Video Compression," International Journal of Computer and Electrical Engineering vol. 7, no. 2, pp. 70-81, 2015.