Fast Inter Motion Estamation Algorithm Research For H.264

H.264 is the newest video standard which is been developed by ITU-T VCEG and ISO MPEG, its high efficiency in coding has attracted extensive human attention. But it is difficult to be widely used now because of its high operation. The main work of this paper is to analyze and research the H.264 coding process particularly, and with the heart of improving the coding time—the part of its inter-frames motion estimation, optimized and experimented the part of its motion estimation, especially in H.264 mode selection.Motion estimation is the most important part in video encoder; its algorithms can directly influence the operation quantity and speed of coding. H.264 standard includes 7 block modes for the first time, and improve the motion compensate precision to one fourth, though it could improve lots of coding efficiency, whereas, it increase the operation complication of motion estimation much more at the same time. The previously research on motion estimation is focus on fast search algorithms, but research little on fast mode selection. In this paper, during analyzed characteristic of video, and sum up these videos' common characteristic, such as color and motion characteristic, propose a fast inter mode selection method based on color and motion information base with these. This method simplified mode selection process, and reduced the operation quantity, operation complication, and operation time of ME's (motion estimation) greatly.Secondly, because the H.264 coding standard has high friendly with network, make it be advantageous to the transition of internet bit stream. So the H.264 coding algorithms should have great coding effect to the video bit streams that have different quality levels. Then, another mode selecting algorithm with adaptive video bit stream quantization was proposed based on the fast mode selecting algorithm proposed in this paper. With the new algorithm, both the well quality video bit stream and the low quality video bit stream could also be coded fast and effectively.According to the experiment, the algorithm proposed in this paper could improve the motion estimation speed effectively, reduce the complexity of motion estimation and keep high PSNR and low output bit rates to the video bit streams with different quantization levels.