Research On Motion And Disparity Estimation In Multiview Video Coding

About 70 percent of the information which people get from the natural is captured by eyes. So the transportation of videos and images is much more important than before. Videos are often captured by cameras that as an extension of people's eyes. There are complicated object motion and several color change in video images. So when transport them, too much data should be transport, which needs lots of bandwidth. That makes the efficiency of transportation worse. Especially in multiview video, with increasing of the camera number, the information increases timely. In fact, almost 80 percent of time wasted in video coding is used in motion and disparity estimation. In other side, as the most popular and efficient coding method, H.264/AVC is used in this paper, based on which we did deeply research in motion and disparity estimation in multiview video coding.This paper paid main attention in analyzing JMVC (Joint Multiview Video Coding Model) which is the standard coding model designed for multiview video, and then proposing a new search method in the process of disparity estimation which makes the coding time decreased by about 40 percent compared with the original search method in JMVC. As an addition to the forward work, the relationship between motion and disparity estimation vector is exploited based on the setups of cameras which is parallel to make the motion estimation much more efficiently and prove that the algorithm is feasibility.