The Study Of Motion Estimate And Regional Separation Coding In H.264

With the fast development of networks and multimedia technologies, video, image processing has been widely used in many fields, but the digital video data is always too large, which leads it is difficulty to store and transport. In order to overcome these shortcomings, lots of video compression standards have been raised. Nowadays the best video compression standard is the H.264, which was raised by JVT (the Joint Video Team) in 2003. Compared to previous compression standards, the technologies of intra prediction, Multi-Reference frame, integer DCT transform and deblocking filter have been applied in H.264 standard. However, these technologies have greatly increased the complexity of H.264 encoding, which affects the application of H.264 and makes it more difficult to apply in real-time.The paper introduces the characteristics of the H.264 standard firstly. For the high coding complexity problem of the H.264 encoder, two optimizing algorithms have been proposed, the regional separation of coding scheme and a new self-adaptive method of motion estimation search range.For video conference, a regional separation of coding scheme is proposed. This scheme firstly using the existing face detection technology to separate the video image into two parts, the face and background, then encodes the two parts in different ways with the H.264 standard, which is to improve the efficiency of the H.264 encoder.Based on the motion situation and image complexity of the video sequence, a new self-adaptive method of motion estimation search range is proposed. By fully considering the time and space relativity of the video sequence in the proposed algorithm, the motion situation of micro-block layer is predicted efficiently, and the searching range of the picture is adjusted self-adaptively base on different motion situation and picture complexity. On this basis, a directed diamond template is used to optimize the UMHexagonS motion estimation algorithm, which has obtained a higher coding efficiency.The experimental results show that for the video image containing face information, the sub-regional coding scheme has a good coding performance. And the new self-adaptive method of motion estimation search range can accelerate the calculating speed of motion estimation; meanwhile it can still guarantee the video quality.