Optimization And Improvement Of Motion Estimation Based On The H.264 Standard

With the development of internet and multimedia technologies, sharp increase in the demand for visual communication, digital video technology has been increasingly widely used in the field of communications and broadcasting. Processing and transmission technology of video information and multimedia information has become the hot spot for current information in the internet and mobile networks. With video compression algorithms become more sophisticated, many video coding standards have proposed, and H.264/AVC is a new video coding standard brought forward by the ITU-T and ISO organizations, which brings together the latest research results in recent years. It can save up to 50% of stream than H.263 and MPEG-4 standards in the same conditions of image quality. But it is at the cost of the computation for the lower stream, and motion estimation is the largest part of the computation. Therefore, the research of the motion estimation is on high availability. In this dissertation JM model as a platform, full search and UMHexagonS algorithms are studied, which are optimized and improved in order to reduce the complexity of motion estimation, and the efficiency of coding is enhanced.Firstly, this dissertation introduces the basic theory and history of the video coding standards; then it analyzes the basic concepts, the basic framework and the key technologies of the H.264/AVC standard; finally it introduces the prediction techniques, describes the intra prediction mode and the motion estimation principle in detail, and further analyzes on the key technology of the motion estimation which includes choice of block matching criterion,prediction of initial search point, the search pattern and analysis of search strategy. Motion search algorithm can be divided into two categories according to its integrity: full search algorithm and the fast search algorithm. This dissertation is studied from two aspects, as follows.(1) It studies full search algorithm. In order to reduce the computation of full search algorithm in the motion estimation, a optimization model of a motion search algorithm is proposed in this dissertation. The model narrows the scope of the full search and accelerates the search process of the block-matching through optimizing the scope and method of the motion search. The experimental results show that the search speed of the mode has been significantly improved under the premise of optimizing the image quality, compared with the full search algorithm.(2) It analyzes several typical fast algorithms in great depth,and further studies the UMHexagonS algorithm. It proposed a optimization model of movement type,which combines the movement type and the direction of search pattern based on video features. The experimental results show that the improved algorithm significantly reduces the time of coding, and the efficiency of coding is enhanced under the premise of small loss of encoding quality, compared with the UMHexagonS algorithm.