A New Adaptive Block-matching Algorithm Based On The Property Of Image

With the development and popularization of computer and network technology, video technology has been widely applied to various applications and plays a more and more important role nowadays. Motion estimation has become a hot subject because of its critical role in the video compression. Among the so many methods of motion estimation, block-matching is the best one of all. With the good effect of estimation, the high compression efficiency and satisfying the real time calculation easily of hardware system, it is adopted by many video coding standards such as MPEG-1/2/4 and ITU-T h.261/h.263.At first, this thesis gives basic theory of video coding and decoding, and then introduces the key points of core technology, which include rules of block distortion measuring, the choice of block's size and the advantages of block. After that, it introduces 13 classical block matching algorithms exhaustively, from which we analysis their origination, merits and limitation. On the basis of that, this thesis does a classification and summaries the tactics in the block matching, and predicts the forwarding tendency of the algorithm. In the next chapter, we put emphasis on the properties of video. By means of lots of experiments, we discover three characteristics such as the adjacent similarity, distributing based on the centre and distributing concentrating on one direction. Then we analysis their important role played in the development of the searching tactics. Based on the three features of the video and the summary of the searching, this thesis proposes one algorithm named MAPS (mixed adaptation and property of image searching). This algorithm not only has a new way of using the adjacent blocks, but also mixes various shape templates together. Compared with classical searching methods, it has a obvious improvement in the searching speed and searching precision. Especially it has a wide application of various video.