A Study Of Image Registration Of Multi-source Image Based On Artificial Immune System

With the rapid development of remote sensing technology and the current diversity of new sensors, it usually can get plenty of multi-source remote sensing data in the same area. However, image registration is the precondition and foundation in the fuse of multi-source data. The theory and algorithm of image registration have been developed for a long time, but image registration still has many unsolved issues, especially in registering multi-source images, and it needs to do more further research.This dissertation develops a technique for the registration of multi-source images based on artificial immune system (A1S). This method consists of two steps. In first step, it extracts the large edges using the method that based on Wavelet and Canny edge detector, and takes Hausdorff distance as similarity measure and uses artificial immune network algorithm to speed up the searching of the initial transformation parameters. Then in the second step, an area-based method utilized to refine the initial transformation and enhance the registration accuracy. The proposed method can register multi-source images with much alignment because that partial Hausdorff has the advantages that is relative insensitivity to small perturbation and natural allowance for portions of one shape to be compared with another, and that artificial immune network algorithm have the good global search capabilities. The dissertation focuses on the main process of multi-source remote sensing image registration, and its main work is summarized as follows.(1)We propose a new algorithm that combines artificial immune network algorithm and Tabu search algorithm to solve the problem of multi-model functions optimization. The global convergence of the algorithm in theory is proved and its performance is analyzed by simulation. The algorithm provides a good way for image registration to search the best transformation parameters.(2)A new edge detector is proposed. Combined Wavelet and Canny detector, it can preserve the large-scale edges and ignore the sharp textures. It is suit for multi-source image registration.(3) The sketch of the multi-source image registration algorithm based on A1S is proposed, and how to realize the algorithm is introduced. The proposed algorithm can be used to register the multi-sensor or multi-temporal remote...