The invention discloses a BPIL algorithm-based image registration method, and aims at improving the image registration precision. The method comprises the following steps of: respectively constructing scale spaces of a reference image and a to-be-registered image; respectively carrying out extreme point detection in the scale spaces of the reference image and the to-be-registered image; generating feature vectors of key points according to extreme point detection results of the reference image and the to-be-registered image; generating key point feature descriptors according to the feature vectors of the reference image and the to-be-registered image; initially matching the reference image and the to-be-registered image according to the key point descriptors of the reference image and the to-be-registered image; purifying the initial matching results; screening the purified matching results by taking mutual information of the images as optimization objects by adoption of a PBIL algorithm; searching optimum transformation parameters according to the screened matching result; and fusing the reference image and the to-be-registered image to obtain a registered image. The method disclosed by the invention is high in image registration precision and good in registration result stability.