The scale-space theory is used to study the algorithm of infrared and visible images registration.The method of using multi-scale feature points and edges of infrared and visible images as features in the registration is presented.The method of using characteristic scale to determine the sub-images size for the similarity matching is put forward and a LTS-Hausdorff distance is used to evaluate the similarity of the sub-images.Scale space theory and multi-scale features extraction algorithm can describe the image more comprehensive.After getting the match pairs,random sample and consensus algorithm is used to delete the false match pairs to gain the correct parameters of space transformation,then the parameters is used to registrate and fuse the infrared and visible images.The experiment results show that the method of image registration based on multi-scale can match infrared images and visible images effectively.