Abstract: A new algorithm aimed at infrared and visible image registration based on cross-correlation coefficient and Canny edge region was presented:regarded the improved Cross-correlation coefficient as the similarity function, extracted the edge region and its nearby area, removed other regions with low correlation, used the particle swarm optimization(PSO) to seek. It shows that this algorithm can overcome the shortcomings which infrared and visible images are poor correlated and avoid the bad matching between multi-sensor images.Multi-sensor image registration could be achieved and small error is guaranteed.