关键词: 信息处理技术, 图像配准, 互信息, 测度函数, 文化粒子群优化算法

Abstract: Some problems, such as registration inaccuracy and slow convergence, exist in image registration when the mutual information of images is used as similarity function. A measure function based on maximized mutual information is constructed according to the image gray and spatial structure information. The cultural particle swarm optimization, which is suitable for automatic image registration, is combined to present a novel image registration algorithm. Two kinds of search spaces, i.e., population space and belief space, are set in the algorithm. The population space is evolved with adaptive PSO strategy, and the belief space is evolved with update function. The particle swarms of population space not only track individual extremum and global extremum to update themselves, but also interact with best individuals in belief space to speed up the convergence speed, which overcomes the problems, such as the large amount of computation in image registration, slow search speed and so on. Experiments show that the proposed algorithm has stronger robustness and better registration accuracy compared with the existing PSO registration algorithm.

Key words: information processing technology, image registration, mutual information, measure function, culture particle swarm optimization algorithm