Research Onadaptive Parameters Based On MAAR Modelfor Frame Rate Upconversion

Current mainstream coding standards, such as H.264/AVC, AVS compression performance more than doubled than any previous methods. When the network bandwidth is limited, these video coding technology is still difficult to achieve satisfactory results, and prone to errors during transmission. Therefore, how to further improve video coding efficiency has drawn more and more attention when the network bandwidth is the bottleneck. The performance of frame rate up-conversion and interpolation algorithms affects the efficiency of video compression coding methodIn the field of frame rate up-conversion, MAAR is a very effective and ideal adaptive method. In MAAR model, each pixel in the interpolated frame is obtained by linear combination of pixel in a sample space. The sample space consists of adjacent pixels in spatial and temporal. In the MAAR model, the size of the interpolation window needs to be predefined. The research shows that different sizes of interpolation windows result in different effects. Since pixels in the same window have the same weight. Thus, adjusting the size of interpolated window according to features of video sequences can make the process of MAAR algorithm finer.Based on the above consideration, observing the strong correlation between distortion of the bi-directional motion estimation block and improper block, this paper proposed an adaptive approach based on MAAR model. First, we calculated the residual energy of each macro block and divide the macro block with high energy. Second, we divided the sub-block according to the confidence. Then the unreliable blocks are merged. We build the motion vector and refine it. Finally the refined motion vector is used as initial value of the iteration algorithm to get better performance and efficiency.Because the adaptive algorithm to avoid the need to achieve the specified parameters of the original constraints, the algorithm also have universal application in dealing with different resolution sequences. As a conclusion, the proposed adaptive method has important innovation significance as well as great application value.