A fast image stitching algorithm based on SURF
Qiaochu Fu, Haiying Wang · 2017
This paper proposes a fast image stitching algorithm with a new way to determine feature searching area. Consider that flat area, such as ground, of an image has less information, and features always mainly gather on other objects. So in this paper we set the region of interest (ROI) by finding a set of potential feature points. We use an algorithm model to identify candidate feature points, and set up a ROI searching area. After setting the ROI we use speeded up robust feature (SURF) descriptor to achieve image registration. Finally, we compare the proposed algorithm with other stitching algorithms. The result of the experiment indicates that the proposed algorithm improves the computational speed while maintaining the accuracy of splicing.