The Research Of Selective Algorithm Based On H.264/AVC Motion Estimation

This article introduces the H.264/AVC video compression technology, details of its principle, and focus on its core technology of motion estimation, it discusses a search algorithm called EPZS based on block matching, then improves this algorithm, reduces the motion estimation time and encoding time without destroying the video quality. EPZS is a predictive algorithm, it can be divided into the partition about selecting the best vector and the fine search. The former is a process that let the value of the motion estimation block be compared with the threshold and the second threshold, with the situation that satisfy some condition,it can quit this process, else it should enter the fine search. The fine search means,it takes the most advantage of the motion vector as center, then searches the best matching with some searching pattern.Because the video type that be processed by EPZS is not selective, this paper focus on this shortcoming, according to some video sequences that has flat textured background, adding a pattern that decides and selects the all-zero4×4block, eliminating this kind of block in the process of running EPZS. At last with the JM10.2test model and the situation of almost not changing the value of bit rate and Y-SNR, it deduces the motion estimation time and encoding time. The effectiveness of the improving can be proved.