Study On Motion Estimation Algorithm Based On H.264 Standard

The motion estimation research on video compression, which is based on the new H.264 video compression standard, has been quite interested by researchers in video compression field. The effect of efficiency for motion estimation algorithm mainly lies in three areas, namely quality of reconstructed image, compression rate and speed of searching. Considering the accuracy of search, if the estimation for accurate motion is more accurate, then the quality of the predictive compensation image is better, the margin compensation and the bits for compensation is smaller, and the bit rate is lower. Considering the rate of search, the higher the rate is, the better application in the real-time environment is. Therefore, the research direction for motion estimation algorithm is to do improvements on both accuracy and rate for search.This essay studies on the technical characteristics of H.264 standard, and principles of motion estimation and motion Compensation. Firstly, an analysis for the algorithms of the classic three-step search, the new three-step search, the diamond search and the hexagon search is given. Then based on the improved algorithms till now, the author points out the research direction for the motion estimation algorithm.Based on the study of the UMHexagonS algorithm and aimed to the shortcoming of excessive search points and insufficiency, this paper proposes some improvement: because the helix search of 25 points do not considers the specific motion vector distribution, the search points is revised to nine points; the non-uniform multi-storey hexagon search do not take advantage of the image motion characteristic, which leads to excessive search points, so this paper proposes a new search cyclostyle; the unsymmetrical-cross search is modified to symmetrical-cross search. The new algorithm makes use of various of test sequences, makes sure that the PSBR and bit rate guaranteed, and lowers time for search and motion estimation. This paper also researches other improvement in the world and has better results compared with the two improvement ways in them.