Data compression of multiple-view movies via projective geometry of the viewpoints

In this paper, we propose a method for compressing the data of multiple-view movie considering redundancy of the viewpoints. This is achieved by dividing multiple-view movies of dynamic scenes captured by several cameras into foreground and background, and applying protective geometry to the viewpoints for each region. Specifically, a viewpoint, called the base viewpoint of compression, is selected from multiple viewpoints and another viewpoint, called the target viewpoint of compression, is predicted by applying projective geometry between the base viewpoint and the target viewpoint. We also propose a method to decide the optimal viewpoint for compressing multiple-view movies efficiently. The results of compressed multiple-view movies and obtaining an optimal viewpoint for compressing them are evaluated in experiments.