In order to solve the problem about infrared and visible image fusion target dis-match by imaging mechanism,the background information,illumination and other factors with the loss of target information,this paper proposes an improved fusion method based on dense SIFT. Firstly,the infrared and visible images are respectively processed with dense SIFT to obtain correspo nding initial fusion decision graph, and then through extracting interest points of infrared light image,the free poi nts are removed to obtain the final area.Finally,the infrared and visible image targets region and the background region are fused after mapping.The experimental results show that the method proposed in this paper is more accurate and can be used to improve the fusion performance,and increase t he information entropy by 0.6in the second and third groups of experiments.