Abstract: Aiming at the problems of geometric misalignment and stitching seams in UAV image stitching, a joint application of the optimal stitching seam algorithm and image fusion is proposed to repair and seamlessly fuse image geometric misalignment and stitching seam regions. Firstly, the graph cut algorithm is used to search for the minimum difference line of interframe splicing to repair the geometric misalignment between frames. Then, the region of interest (ROI) of strip fusion is established based on the difference line, and the size of the fusion ROI range changes adaptively with the size of the tonal exposure difference between frames. Finally, the fusion ROI mask is introduced to improve the Laplace multi-resolution image fusion algorithm, and the tonal exposure difference between frames is smoothed,so as to achieve seamless fusion of UAV. And the seamless fusion of image stitching. The experimental results show that compared with the traditional Laplace fusion algorithm,the image fusion results in this paper improve the mutual information by an average of 12.79%,the peak signal-to-noise ratio by an average of 21.48%,and achieve a natural and seamless visual effect,which is of certain application value.

Key words: UAV, geometric misalignment, stitching seam, graph cut, multi-resolution fusion