Particle Swarm-based Video Motion Estimation Algorithms And Optimization

With the recent rapid development of multimedia technology, higher demand of real-time video image compression in many applications, rapid and efficient compression algorithm is the key to solve this problem. Motion estimation, the core video compression technology, is also a one of the most complex and time-consuming aspect of coding operation. In video coding, motion estimation accounts for about 50% to 80% of the whole encoding computation, so it is the most challenging research topic in video encoding. On the other hand, motion estimation which based on block-matching is simple and easy to implement, therefore, this estimation used by all current international video coding standards.Fast full-search matching algorithm is one of the most direct and simplest block-matching motion estimation methods. Because of its high computation complexity difficult to apply in real-time video encoding, many fast block-matching motion estimation methods are deduced in practical applications.In this dissertation, the distribution of motion vectors in the block-matching motion estimation are analyzed. Spatial and temporal correlation of the center-biased motion vector are proposed, more classical and well-known fast algorithm of the existing literature are classified.On the basis of in-depth analysis of motion estimation and particle swarm optimization (PSO), this dissertation conduct PSO algorithm in application of the area of Block-Matching Motion Estimation, and through the theoretical and experimental analysis of the involved parameters in PSO algorithm parameters, the final values are determined.Finally, three aspects of inadequacy of the standard PSO algorithm in block-matching motion estimation are analyzed in this dissertation. The characteristics of motion vector distribution combines variation characteristics of particles in the standard PSO algorithm to make the improvements of the inadequacy. The advanced PSO algorithm is presented so as to obtain faster speed and better search performance.