Parallel Process of Hyper-Space-Based Multiview Video Compression

Multiview video coding (MVC) is a key technology in free-viewpoint television. MVC based on traditional existing codec system has been studied widely, but all of them need powerful computational capacity in processing. Parallel process of MVC can facilitate the efficient implementation of encoder and decoder and has been required as a function by MPEG. In this paper, a parallelization methodology for MVC based on hyper-space theory is presented and tested on the local area multi-computer - message passing interface (LAM-MPI) parallel platform and modified H.264 codec. Experimental results show that the proposed method can speed up processing of multiview video compression and obtain high rate-distortion results.