Analysis of image stitching error based on scale invariant feature transform and random sample consensus
Dou Shuang, Hua Cai, Yong Yang, Tailin Han, Menghan Xue · 2015
This article introduced the process of image stitching based on SIFT and RANSAC. It was pointed out that the reliability calculation of the RANSAC algorithm is based on an idealized assumption which does not hold. The false assumption will lead to the instability of image stitching results, especially in multi-map or panorama situations. To solve the problem, this paper presents a strategy to limit the range of randomly points selecting. Experiments show that the strategy can effectively reduce the effect caused by random errors, and obtain more stable image stitching result.