The Optimization And Implementation Of H.264 Encoding Algorithm On DSP

H.264 is the latest video compression coding standard and it has adopted many advanced coding techniques. Research on the H.264 standard is of great importance in pratical and is one of contents in many fields such as video compression coding, multimedia communication. However, the high encoding efficiency is based on the high complexity. Research on optimization algorithm of H.264 codec must be done to apply in video communications products.Firstly, this paper introduces the theory and implementation methods of H.264 encoder, DSP TMS320C64xx hardware platform and assembly instructions in detail. Then, some time-consuming modules of H.264 encoder are optimized at assembly level using SIMD (Single Instruction Multiple Data) technique and software pipeline technique. These modules include integer transform, quantization and CAVLC (Context-Adaptive Variable-Length Coding) entropy coding module. Furthermore, the method of early termination by detection of zero blocks is used to increase the speed rate of the encoder. At last, the H.264 encoder with proposed optimized methods in this dissertation is achieved on the DSP platform. The experiment results show that compared with the encoder without optimization, the H.264 encoder is obviously speeded up with the similar reconstructed image quality. Moreover, these optimized algorithms can be used in practice.