Studying For The Application Of Motion Estimation Of H.264 In Video Conference

The rapid development of information technology has greatly affected the working methods and the way of life of human society, many new applications are emerging, multi-media video conference system is the rise of a new multi-media communications in recent years. The video encoding technology is the core of the video conference system. H.264 is a new video standard that the IEO /IEC and ITU-T International Organization have developed jointly for standardization. As the latest standard in video codec field ,H.264 has such advantage as high compression rate, strong error resilience capacity, high advantage as high adaptability and so on. In order to gain good coding result, H.264 adopts so many new algorithms that the amount of calculation increase drastically. As a result, the application in PC environment is restricted. Experts found that motion estimation occupied nearly 70 percent of encoding time through the analysis of the each module in the entire process of encoding time-consuming using the performance analysis software . Therefore, in order to achieve the H.264-based real-time video conference system codec, its key technology is how to reduce consumption when the campaign is estimated to achieve high video data collected by the coding efficiency of high-quality transmission. Starting from this issue, papers focused on the motion estimation algorithm selection, optimization, and transplantation issues in video conference system based on H.264.First of all, block-matching motion estimation algorithm method is simple, and the estimated effects better, meet the potential of real-time computing systems ,so we systematically summary the current block matching the existing domestic and international norms and block-matching algorithms on the basis of the understanding the nature of the problem on block matching . This paper identified ARPS as the rapid motion estimation algorithm.Then, this paper reference to the H.264 code JM86 experimental platform for analysis its encoder. We reduce the redundant code that transplant do not need in order to enhance operational speed. At the same time, the function lencode_me.c of JM8.6 is replaced by the search algorithm ARPS. The results showed that we can save the encoding time 57 % to 61% after these optimization, image quality and bit rate changes rarely.Thirdly, algorithms are optimized on DM642 platform. By fully making use of the DM642 CPU structure and instruction, linear assembly program language is used to improve the key functions(function SAD and function quantify)which makes code size reduction, significantly lower clock cycle, to meet the real-time display functions.Finally, we built a hardware platform using digital cameras,development board VMD642-A,PC,hub and transplant the optimized code under the CCS3.1. At the same time, we also designed the video capture and compression software that including the four modules video capture,coding,code-network transmission and decoding shows . The paper achieves the requirements of the multi-channel collection,compress and display in Video Conference Based on the Motion Estimation of H.264.