An image registration and fusion algorithm based on gray feature was proposed for sequence infrared array scanning images. Firstly,the corner points were detected by Harris algorithm in the selected region,and an adaptive threshold algorithm was proposed to extract interesting points with obvious features and limit the number of the corner points; secondly,an appropriate window was selected,and then the images were matched by the normalized cross correlation (NCC) algorithm in neighbourhood gray range of interesting points and the transform relation between them was obtained; finally,an image fusion algorithm based on bilinear interpolation algorithm and gradated in-and-out amalgamation algorithm was proposed. The results show that the interesting points after the corner detection with adaptive threshold have obvious features and uniform distribution,and at the same time,the numbers of the corner points are well limited. The images using the algorithm have accurate matching ability,and the final fusion image has smooth transition without evident seams and gray jump phenomenon.