Researches On Algorithm Of H.264 And Its DSP Implementation

In the emerging communication system, research on fast coding algorithm is the key technology for improving video quality. It has been a hotspot project of application to video communications in recent years.H.264 is the latest video coding standard developed by the Joint Video Team (JVT) which is formed with ITU-T Video Coding Experts Group (VCEG) and ISO/IEC Motion Picture Experts Group (MPEG) in 2003. H.264 is a new developing video coding standard,as MPEG-4 part10,and it aims at proposing a high quality and low bit rate video compression standard by researching into the existing coding techno1ogies.H.264 has been the focus in the video compression area,it has a promising future in digital media market. So it is important for engineering application and theory to research deeply into H.264 and the implement of its fast H.264 encoder. In order to realize H.264 real-time codec,many optimization works have to be done. But these optimization gains should not be at the cost of obvious coding performance drop.Block matching motion estimation is relatively simple and can be easily realized,it is adopted by current international video coding standards. As the most direct and simplest motion estimation algorithm,full search has much higher computation complexity and it is difficult to be integrated into real time video encoding systems. Many fast blockmatching methods have been proposed to decrease computation load. So this thesis puts forward the algorithms of cancelling time redundant between neighboring frame in rhomboid research and the model of judging the thrown frame by analyzing the error, based on the visual sense, the study of relevant frame, and the comparing analysis through motion estimation. The paper also discussed different strategies and compensatory methods in discarding frame, and tested the algorithm in the H.264 standard. An adaptive optimal coding algorithm was proposed to reduce the computing redundancy in H.264 recommendation algorithm. A relation between packet loss rate and times of simulating decoding in encoder was setup in proposed algorithm for the first time. Using this relation,encoders could adaptively determine the times of simulating decoding according to packet loss rate and the number of simulating channel states. So the computing complexity and coding time in encoder was decreased effectively. Simulation has shown that the proposed scheme can effectively reduce the computing complexity and save up to nearly 60 Percent of coding time by decreasing the times of simulating decoding in encoder. Under the same coding time as H.264 algorithm,the average standard deviation of decoding video information can be reduced about 90 percent by using proposed algorithm. The target of the project is to develop a DSP based video compression system. The key hardware of the system is Texas Instruments Inc.'s high performance TMS320C6416 DSP. At last, the system software and hardware optimization flow is supplied, and the system evaluation results are illustrated.