A Standing Tree Image Mosaic Algorithm Based on SIFT

This paper concerns the problem of standing tree image mosaic based on SIFT. In standing tree measurement based on image processing, the first step is to take photo of the measured tree. Due to some special tall trees, or limited by geographical conditions, it is impossible to shoot full image of the tree one time, and need to shoot sequence tree images with overlapped areas and stitch them. In this paper, a tree image mosaic method based on scale invariant feature (SIFT) was put forward. Firstly, the SIFT feature vector of each image was calculated and the feature points were matched using SIFT features. Then the perspective transformation was done to the image which was to be joined by selected four pixels. Lastly, the sequence images were stitched based on weighted factor. The experiment of true tree image sequence stitching achieves the better smooth mosaic effect and show the method is effective.