A New Method for Medical Image Registration Based on Wavelet Multi-scale Transform

Medical image registration is the first step for image fusion and other imaging process. Current registration algorithm can be divided into feature-based and statistic-based approaches. In this paper, feature points of medical image are detected by using wavelet multi-scale product firstly, then registration of feature points based on mutual information is introduced, and translation parameters are calculated by using particle swarm optimization(PSO) algorithm. Experiments show that our method can reduce the computation of image registration while maintaining good accuracy.