Fast view synthesis using platelet-based depth representation

This paper presents a novel approach to speeding-up the view synthesis process. A general case of view arrangement in a 3D video system is considered. The proposed approach offers a certain acceleration practically without any quality loss, and further acceleration at the cost of lower quality of synthesis. Amount of the acceleration can be gradually exchanged with the degradation of synthesized image quality via adaptive block size depth data partitioning. The proposal exploits depth modeling - the depth data is divided into blocks of various sizes and modeled as planes. In such case only 4 corners of each block need to be transformed during view synthesis, instead of transforming every pixel in the block. This way depth data is adaptively simplified. The experimental results on wide range of multiview test sequences are provided, for both original and compressed depth data.