The Research Of Rate Control Algorithm And Encoder Optimization Based On H.264

As a new generation video coding standard, H.264has thecharacteristics of high compression ratio, good network affinity, strongadaptability and sufficient quality. H.264algorithm has high encodingefficiency but also high complexity, this feature represents a bottleneck forsystem implementation based on normal platform. With the launch of the dualcore processor by TI, DAVINCI platform is a good solution to the problem ofsystem implementation of H.264coding algorithm. Now it is the idealplatform for video coding system.The thesis emphasis on the research of H.264rate control algorithm andits encoder implementation and optimization based on DM6467platform. Themain work of this paper is as below.In the beginning part, the paper review of international video codingstandard, the principle of video encoding and new features of H.264. Then thepaper focuses on the related theory of rate control algorithm.Then in view of drawbacks of existing rate control algorithms, an improverate control algorithm for H.264is proposed in the following part. On the onehand, four novel weighted models using both temporal and spatial informationare proposed to predict the MAD of different motion intensity. On the otherhand, the changes of MAD are adopted to represent the image complexity, andused as a coefficient of the target bit allocation. Compared with algorithms ofJVT-H017and references, the bit rate of proposed algorithm is closer to thetarget bit, and the average of peak signal to noise ratio (PSNR) is improved.In the last part, the paper introduces the structure and developmentenvironment of DAVINCI platform, mainly for DM6467introduction, andconstruct and test the SEED-DVS6467T development environment. Thenpresent the transplant and optimization of x264encoding algorithm. Finallythe system encoding performance was tested through video sequence. The test results show that, the frame rate has been greatly improved after optimization.