An optimal solution for RANSAC using constraint of saliency region

RANSAC and RANSAC-like algorithm are most employed for the robust computation of relations from a number of potential matches in the field of computer vision, such as stereo matching, image retrieval, mosaic and elsewhere. There have been a number of recent efforts that aim to increase the efficiency of the standard RANSAC algorithm. This paper presents a novel optimal solution of the RANSAC algorithm that is much more efficient. The contributions of this work are two-fold: firstly, the nearest neighbor mutual constraint rule is used to get matching feature points; secondly, the relate information of saliency region is proposed as a constraint condition to refine the corresponding points. The results on real-world images demonstrate that the proposed method is able to achieve a significant performance gain compared to the standard RANSAC and PROSAC.