Abstract: Image registration is a key step in image mosaic, image fusion and object localization. Given the problem that the aerial remote sensing images may have the uncertainty in image gray level and may be lack of feature information, a registration algorithm of aerial remote sensing images based on lateral inhibition competition was proposed, Firstly, the bright feature points and dark feature points of the reference image and the under registration image were detected with the basic idea of lateral inhibition competition. Then, the similar nature feature points of two images were combined with feature matching strategy and the matching-points sets was shaped. At last, the image transform model was calculated and the images were registered. A new thinking was provided for image registration. Comparing with the traditional feature registration algorithm, the experimental results show that the proposed algorithm has better registration, the aerial remote sensing images can be registered effective. Some conclusions are obtained through the analysis and explanation of the experimental data, which lay a solid foundation for further research.