Mpeg-4 Video Encoder Optimized

Along with the advancement of society and the rapid development of the information technology, people are required to have the resources of multimedia information available anywhere at any time. Unfortunately, video compress algorithm of the first generation can't meet the requirement. Therefore, its second generation, the MPEG-4 standard comes into existence, which is based on the objects and aims to provide a general platform for the application of multimedia. Now many companies have taken part in the research competition of MPEG-4 in the world. The most important three companies of them are Microsoft, Real and Apple. But it seems that the MPEG-4 is not so popular at home, and it is mainly applied in digital monitor and control system. So it has import significance to research the MPEG-4 standard and the method of real-time compression, especially in the domains like video-teleconference, visual telephone, IPTV and so on.Considering the developmental trend of the video compress and the demands of the people, this paper makes researches on the visual section of the MPEG-4 standard, present an amendment to the DS search algorithm, and does some optimization to the module of compress which is utilized frequently by the encoder so that the encoder can accomplish real-time compression on ADSP-Blackfin535 EVM board to the CIF picture, the size of which is 352 by 288. The main contents are as follows.Firstly, the movement estimation should be watched carefully, which is the core model of the video compress. Afterwards, the simulation of the frequently-used search algorithms will be undertaken, for instance, three step search algorithm, four step search algorithm, new three step search algorithm, diamond search algorithm etc. After the comparison between those search algorithms in terms of accurateness of search and complexity metrics of calculation, it is shown that the diamond search algorithm has the best performance. Then what will be done is just to optimize the search process of the diamond algorithm. In this case, reduplicate calculations can be avoided. Strategy of jump-out-ahead-of-time is applied in the process of SAD calculation. As a result, the complexity is reduced; in the meantime, the search accurateness is not affected.