| As a new generation of video coding standard, H.264/AVC absorbs the latest research results and algorithms in video coding domain, which can significantly improve the video coding efficiency. H.264/AVC encoder is very complex because of the many advanced but computationally complicated schemes it employs in achieving a high-coding performance, so it need a long time to encode a frame.Multi-view video is an interactive multimedia application. It shows the directions of the next generation applications in multimedia, such as networking, interactive and realistic. Since the H.264/AVC can provide higher compression efficiency in deal with the large amount of data, we select the H.264/AVC framework to study the multi-view video coding methods.The main work of this paper includes three main parts. Firstly, we analyze the structure and process of the monocular video encoding framework based on JM platform. Secondly, we study multi-view video coding methods using the H.264/AVC framework. Thirdly, we accomplish the coding algorithm of binocular video on JM platform. We complete three algorithms of encoding multi-view video especially binocular video on the basis of the analyzed structure. The first algorithm is only to store each frame of binocular video alternately and to encode sequentially. The second algorithm is based on the characteristics of binocular video with considering the time relevance between the coding frame and the reference frame. When we encode the first video, the reference frame is just selected inside. The time of motion estimation is reduced by more than 50% compared with the first algorithm. The third algorithm uses the fundamental matrix constraint between two views. When we encode the second video, if the corresponding frame is encoded in the first video, we select the corresponding frame as the reference frame and adjust the predicted motion vector MV_P. The time of motion estimation for the second video is reduced by 15% compared with the second algorithm. |
PDF Full Text Request