Abstract:

This paper proposes the use of view synthesis prediction for reducing rate-overhead incurred by transmitting depth-maps in free viewpoint TV applications. In particular, the scenario in which depth-maps with varying degrees of quality is available at the decoder for free viewpoint video applications is considered. The depth-map for each view is encoded separately from the multiview video and used to generate intermediate views as well as view synthesis prediction for coding efficiency improvement. It is shown that the rate overhead incurred by coding high-quality depth maps can be offset by reducing the necessary bitrate for coding multiview (texture) video with the proposed technique. The effect of downsampling as well as the use of different QPs for the depth map are also discussed.