Optimization And Implementation Of Search Algorithm Based On H.264

H.264/JVT is the latest international standard for video compressing code today. Compared to the standards existed before, such as MPEG-1/2/4 and H.261/H.263, the coding efficiency of H.264 has been improved prominently.Based on the complication of the existing algorithms for H.264, which becomes the obstacle of real time communication, in this paper, the disadvantages of the existed research algorithms are studied more, and a new method for image preprocess, the dynamic combination of macro block and an adaptive fast comprehensive search algorithm are put forward.In the image preprocess to optimize itself, the image complication is obtained by Fourier analysis of the video image frequency spectrum, then, the dynamic macro block combination modes and the adaptive search templates choice are given.An improved search algorithm is derived from the motion correlation of time-space domain, which is based on analysis of virtues and drawbacks of several traditional search algorithms. The prediction value can be obtained by using motion vector median value prediction, which then is compared to the threshold value T, if it is less than value T, the motion of video image is slow, SCSA is adopted; otherwise, the motion of video image is fast, LHDSP is adopted.In this paper, full search, diamond search and the IMP search algorithm are compared based on JM82 model. The results of simulation indicate that the number of searched points in IMP algorithm is 1.02% of that in FS algorithm and 69.79% of that in DS algorithm. Therefore, IMP algorithm improves the search speed greatly,...