An improved RANSAC algorithm based on the geometric constraints

Abstract

Eliminating false matching is an important part in image stitching technology. Traditional eliminating erroneous matching method in the field of image stitching is RANSAC algorithm, but this method need numerous iterations and complex computation, and it often could not completely eliminate the false matching. Focusing on these shortcomings in RANSAC, this paper presents an improved RANSAC algorithm which based on the geometric constraints. First, clustering and grouping the matching points, then it is based on the two geometric relationship that corresponding match points where a straight line should slope equal, distance equal to establish a pre-judgment geometric constraints model and pre-purified matching points. The experiments show that the algorithm compared to the traditional RANSAC algorithm, eliminating mis-matching, reducing the number of iterations, improving computational efficiency, thereby improving the efficiency of image matching algorithm.