Research Of H.264 Codec Optimization And Implementation Technology

H.264 is the newest video coding standard, which are developed by ITU and ISO jointly. By adopting a number of new coding techniques, H.264 standard has achieved a significant improvement in compression performance and network adaptability compared to the existing video coding standards. As a result, this standard obtains much attention from academic and industrial fields. But due to the great computational complexity and the low execution speed, H.264 standard has a limited application area. Therefore, how to reduce its computational complexity and accelerate the execution speed while maintaining the outstanding compression performance is a crucial problem in the H.264 application.Through studying each part arithmetic of H.264 standard and analyzing H.264 encoder, we can conclude that motion estimation is one of the core techniques of video coding and by using motion estimation and motion compensation techniques, large amount of temporal redundancy that exists between frames of video sequences can be reduced and then high compression performance can be obtained. In whole process of video coding, motion estimation is one of the most time-consuming parts, so how to increase the efficiency of motion estimation and look for a faster and effective motion estimation algorithm has became a hot research hotspot at present.Based on studying the actual fast algorithms of motion estimation, the fast motion estimation algorithms are classified by the approaches of how to accelerate the search process. Then a deeply research on fast motion estimation algorithm based on search pattern and a summary of characteristic are made. And it is found that almost of these algorithms are based on serial search ideas.Through summarizing the statistical characteristic of video sequence and analyzing the shortages of serial search ideas, the idea combined cross-shaped pattern with parallel search is proposed. Furthermore, based on this idea three new fast algorithms are designed. After analyzing the three algorithms, the best one with general capability is found. Both the experiment results and the theoretical analysis have proved that the best algorithm based on cross-shaped pattern and parallel search ideas is much quicker than existing algorithms. At the same time it can maintain the similar quality of video sequence.