An adaptive uniform distribution surf for image stitching

Abstract

Speeded up robust features (SURF) algorithm is widely used because of its excellent computational efficiency and robustness in the field of image matching and image stitching. In this paper, we propose an adaptive uniform distribution SURF algorithm for image stitching. The proposed SURF algorithm will adaptively select the appropriate uniform distribution radius which is depending on image complexity and utilize this radius to wipe off amounts of unnecessary interest points. The remaining interest points will be distributed well in the image and much less than the number of interest points detected by the original SURF, which will result in reducing the computation complexity of subsequent SURF procedure significantly. Experimental results show that our proposed method is able to shorten the processing time distinctly without loss of matching precision.