Stereoscopic video coding based on global displacement compensated prediction

In this paper, we propose a novel algorithm for coding the stereoscopic video sequence based on the global displacement information between the left- and right-view images. MPEG-4 temporal scalable video coding scheme is employed at the baseline, in which the left- and right-view images are compressed at the base and enhancement layer, respectively. To further improve the coding efficiency, some intuitive properties of stereoscopic videos are utilized. One of the major properties lies in the global displacement between the left- and right-view images. In this paper, we propose coding the right-view image with global motion compensated prediction from the left-view image. A further main property of stereoscopic video arises from the fact that, objects simultaneously existing in the left- and right-view images usually result in the very close motion vectors. Based on the global motion information, the motion vectors of the right-view sequence can be easily predicted from that of the left-view sequence. Rate-distortion optimization is also used to select the coding mode. Experimental results show that, compared to the MPEG-4 temporal scalability algorithm for low bitrate stereoscopic video coding, the proposed algorithm can save the bitrate up to 20%.