基于改进蚁群算法和互信息的多光谱图像的配准-红外技术.pdf

第32 卷 第1 期 红 外 技 术 Vol.32 No.1 2010 年1 月 Infrared Technology Jan. 2010 基于改进蚁群算法和互信息的多光谱图像的配准 杨绪超，王敬东，李 鹏 （南京航空航天大学 自动化学院，江苏 南京 210016 ） 摘要：基于互信息的配准算法有精度高，自动化程度高，不需要对图像进行预处理等优点，已经被大 量应用于多光谱图像配准中。但是在计算互信息时，出现了很多局部极值，这就为目标函数的寻优过 程带来了很大的困难。提出了一种基于蚁群算法的配准参数寻优方法，改进蚂蚁在每层节点间的转移 准则以及全局信息素的更新策略。实验证明，此方法不需要人工干预，对参数的初始值没有依赖性， 配准成功率高。 关键字：图像配准；蚁群算法；互信息 中图分类号：TP391 文献标识码：A 文章编号：1001-8891(2010)01-0041-05 Multi-spectral Images Registration Based on improved Ant Colony Algorithm and Mutual Information YANG Xu-chao，WANG Jing-dong，LI Peng (College of Automation Engineering, Nanjing University of Aeronautics Astronautics, Nanjing 210016, China) Abstract ：Mutual Information-based image registration has been widely exploited due to its accuracy, high-level automation and without the need for preprocessing of the image, especially in multi-spectral images registration. However, the mutual information function is a function contains many local maxima, which has a large influence on optimization. A robust method based on Ant Colony Algorithm is proposed for the optimization of the registration parameters. The transfer criterion of the ants between each layer and global pheromone updating process are described. Some experiments results show that this method can register multi-spectral images accurately and automatically without manual intervention and the initialization. Key words ：image registration ，ant colony algorithm，mutual information 素相似性的方法[2,3] [4,5]