H.264 Fast Intra Prediction Mode Selection Algorithm

H.264 standard is established by JTV ((?) Video Team) consisting of experts from ISO/IEC MPEG (Moving Picture Experts Group) and ITU-T VCEG(Video Coding Experts Group) and is formally put out at 2003. As the newest generation of video coding standard, H.264 standard introduces a large number of advanced video coding technology and new ideas. That makes it have much higher coding efficient, better video quality, network friendship and fault tolerant character. And that also brings it great computation complexity, which limits its application area, such as in the real time communication application or in some embedded systems. Aiming at the status, many lesea(?)ches have been done and being done to decrease compute complexity and together improve running efficiency while keeping high image quality.First we introduce the basic theories and standards of video coding, and the background of home and abroad. Next we simply introduced the structure of encoder and decoder, key technology and man performance of H.264 standard, especially described on the main technology in video coding layer of H.264. Then we analyze the encoder's performance and computational complexity of H.264 in detail, and research deeply on the part of prediction mode selection which has a highly computational complexity.Intra prediction of H.264 utilizes the spatial relativity to compress. H.264 adopts Full Search algorithm based on RDO(Rate Distortion Optimization) to select the best predictive mode. On the basis of large number of experiments and in-depth study, the correlation of the macroblock and the characteristics of the differences between the adjacent pixels of the macroblock are analyzed. An improved algorithm based on edge direction histogram is present. Then we implement both improved DES(Dominant Edge Strength) algorithm and the pre-decision algorithm on H.264 test model JM8.6. Experimental results show the combinative algorithm save about 64% of the encoding time when ensuring the capability of the rate distortion and the bits rate.