Contour Registration Based on Multi-Expression Programming and the Improved ICP
Yanan Wang, Bo Yang, Xiuyang Zhao · 2009
A novel contour registration method based on multiple-expression programming (MEP)and the improved iterative closest point (ICP) is presented. Firstly, based on the fast global minimization of the active contour model, the image contours were extracted. The feature points were obtained by means of the centroid balance sampling method. Secondly, traditional ICP method which is liable to trap in local minimum value was improved by the genetic algorithm. The feature points were matched by the improved ICP. Finally, the registration model was established automatically using MEP.The experiment results show that the improved ICP can obtain the optimum feature point-pair set and the registration method presented in this paper is efficient and accurate.