Context based motion and disparity vector prediction algorithm for H.264 based stereoscopic video coding

Abstract

H.264 video coding standard supports 3D video through dedicated Stereo Video Information (SVI) Supplemental Enhancement Information (SEI) message. This SEI message allows main and auxiliary frames be coded into alternative pictures in which disparity information can also be utilized. However, due to the presence of both disparity and motion vectors at the neighborhood of a given macroblock, the H.264 predicted motion vector motion vector concept becomes less efficient. This paper presents a novel motion and disparity vector prediction algorithm that overcomes this drawback. Results show that the proposed algorithm achieves up to 1dB gain compared to the conventional motion vector prediction algorithm.