The Research Based On Fast Motion Estimation Of H.264

With rapid development of network communications and wireless communication systems, the current video coding technology can no longer meet the demand for multimedia services, how to make efficient, fast video coding algorithm has become a hot research spot. As a high-performance video encoding technology, H.264has high-quality smooth image with a high compression ratio. It can relief the contradictions of video communication and lack network bandwidth. In the video encoding process, the computation of motion estimation is70%of the total coding operation. Therefore, the merits of the motion estimation affects the efficiency of entire coding. It has become a bottleneck that will hinder the development of video coding.In the video sequence, there is a strong temporal correlation between the different frames of same scene. The motion estimation is used to eliminate the temporal redundancy, which can improve the coding efficiency. However, there are some defects such as time-consuming, computational complexity and so on. The current research of motion estimation focuses on the change in the search template, the threshold of early termination, the optimization of matching criteria and so on. The macroblock feature is difficult to be extracted in encoding process, which results little studies from this aspect.The optimization algorithm of fast motion estimation is put forward based on the brightness and frequency domain characteristics of macro block, including the model segmentation algorithm and the prediction algorithm of starting motion vectors. The basic idea of the mode segmentation is that firstly discrete wavelet transform to video frames with the unit of1616macro block. The after low-frequency coefficients is regarded as the research object. Define the sum of coefficients. If the value is greater or equal to the threshold, the corresponding areas will be divided into texture areas, or divided into smooth areas. The current block is compared with the corresponding block in the reference frames. If the result is equal, the current block will be marked as background block, or sport block, and the corresponding segmentation mode is selected at the same time. The basic idea of prediction starting motion vectors algorithm is that based on the marked sport blocks, the starting motion vector of current block is the median value of motion vectors to adjacent sport blocks in the initial motion vector prediction.If the motion estimation time, PSNR and bit rate are regarded as the evaluation standards, the experimental results show that the optimization algorithm can improve the coding time, which can keep the video characteristics and rate distortion compared with the original algorithm. Especially to the sequence with rich movement details, the effect is more outstanding.