Multiview video compression with 3D-DCT

Multiview video is predicted to be the video of next generation, which has the advantages of high accuracy and interactivity. This paper considers the problem of compression of multiview video with 3D-DCT. In this work, we firstly develop an algorithm to extend the scheme to handle multiview video with arbitrary camera arrangement by organizing multiple views into a sequence similar with video sequence, in which neighboring frames show biggest coherence. In 3D-DCT approaches, both inter-view and intra-view spatial coherence are fully exploited so as to achieve superior compression efficiency and visual quality of reconstructed images. And since no disparity estimation is required in 3D-DCT, the number of en/decoding operations per pixel is greatly reduced.