Realization Of Optimization On Motion Estimation Algorithm Based On H.264in The Platform Of DSP

The H.264video compression standard is developed by the Joint Video Team of ITU-U Video Coding Experts Group and ISO/ICE Moving Picture Experts Group. The high efficient H.264encoding high efficiency is built on high complexity encoding algorithm. Motion estimation is vital to video compression.In this paper, through the study of motion estimation algorithm, an optimized algorithm based on the different motion complexity of images is proposed. Combined with motion correlation, the images are divided into background area, quasi-background area and sports area. Then motion vector should be searched by the different motion estimation search algorithms in different area. The X264open source code is chosen as a reference model in this paper. Then we realize the video encoder by transplant the X264code to the Integrated Development Environment CCS of TMS320DM6437platform. Some methods such as the inline function optimization, the C language optimization, compiler optimization, linear assembly optimization are applied to the program. Finally, effective results are got through simulation in numerical environment and coding experiment on DSP platform.Experiments results show that, in the PSNR aspects, the optimization algorithm has good adaptability to a variety of motion complexity of image. The great improvements in coding speed can be got from the optimized algorithm. The encoder built on the DSP meets the requirement of the real-time video encoding.