Optimization Of Inter-Frame Prediction Algorithm Of H.264 Based On DSP

With the fast development of multimedia technology, Multimedia services such as video conference system and HDTV gradually become the focus of attention. The information explosion has made the transmission rate and channel bandwidth become bottlenecks in the development of multimedia technology. 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 can achieve high-quality, low bit rate coding effect and reach a high coding efficiency. However, the H.264 high compression ratio is improved at the cost of complexity. In the practical application it must be optimized.In this paper it analyzed the key parts of H.264 such as Prediction Algorithm, integral transform, quantization, entropy coding and loop filter especially to inter-frame prediction which cost the most time in H.264. After detailed research in T264 which is the test software of H.264 this paper analyses the realization process of inter-frame prediction algorithm. A video compress system is developed and the solution of signal integrity problem which may be happen in the system is given. At last T264 is optimized on the hardware platform including optimization of code level and memory optimization. The result of optimization can achieve real-time coding for QCIF image.