Feature Points Matching Based on Hysteretic Chaotic Neural Network

A new feature point matching method was proposed to improve the target recognition result. The feature points were chosen in the area that was evidently different with its peripheral region. The number of the feature points could be controlled by choosing appropriate threshold value. Three matching criterion function were defined according to the number of feature points in the template image and target image. The matching criterion functions contained not only the relative position matching information of feature points but also gray matching information. Hysteretic chaotic neural network were adopted to optimize the criterion functions. Simulation results show that the method can match feature points between the template image and target image validly.