Research On Image Registration Based On Open Tail Gene Expressing Programming

GEP is a fusion of genetic algorithm(GA) and genetic programming(GP) advantages.GEP inherited the fixed line coding feature from GA and the tree structure from GP, which makes it possess the advantages from both algorithms. With these features, GEP can solve complex problems with simple coding rules and make it faster than traditional evolutionary computation by 2~4 orders of magnitude. GEP has been widely used in many areas such as function discovery, data mining.Image registration is a basic problem of image analysis and processing. It has important applications in aerial image, image's three dimensional reconstructions, computer vision,pattern recognition, medical image processing, image analysis and many other areas.The main research results of this essay are listed as follows:(1)The paper introduced gene expression programming(GEP), image registration and their background and theoretical knowledge and studied the structure of GEP and its application in many areas especially in image processing.(2)Based on standard GEP, the paper proposed open tail gene expression programming(OTGEP) algorithm. Through introducing the optimal individual in the process of GEP algorithm, the paper realized the open tail GEP algorithm. Symbolic regression experiments showed that open tail gene expression programming performed better than standard GEP algorithm on accuracy.(3)With GEP's function discovery ability, this paper transferred image registration problem to a function discovery problem. Compared to the standard GEP, OTGEP has better performance in accuracy; OTGEP was applied to image registration. Experiments showed that OTGEP has a better image registration accuracy than standard GEP.