A fast and seamless high resolution splicing method based on feature invariant description is proposed. The affine transform is used to correct the distortion of multiple high resolution images. The feature points of the image are extracted by Harris corner detection operator

and the SIFT descriptor is used to process the feature points

which makes it rotation-invariant. Based on the principle of shortest path and minimum error in graph theory

the reference image is selected. The geometric error median of the matching feature and the average approximation of the matching feature point are taken as the weights of the edge for image registration. Finally

the weighted average fusion method is used for seamless smoothing of multiple images. The experimental results show that the proposed method is efficient and can effectively eliminate the stitching trace and maintain high resolution.