Efficient Multiview Video Coding Based on MPEG-4

Abstract

In this paper, we propose an MPEG-4 based multiview video encoder. The main view of the sequences is encoded using the MPEG-4 encoder and the auxiliary views are encoded by joint motion and disparity compensation. The output of the encoder contains multiple bitstreams and the main bitstream can be decoded by a standard MPEG-4 decoder. Extensive experimental results show that our proposed multiview coding can achieve significant performance gain over the conventional multiview video encoder, which is implemented by applying the concept of the MPEG-2 Multi-View Profile (MVP) on the MPEG-4 platform. The improvements come from a more efficient reference structure and the joint estimation of disparity and motion fields in our proposed multiview coding system. In addition, in the case of five-view encoding, we also compare four different prediction structures in order to find the best structure under certain scenarios. The proposed encoder is very promising for the applications of video-conferencing and 3D telepresence.