Fast Image Stitching Based on ORB and MSAC Algorithm
Su Yan, Xin Wang, Qingfeng Qin, Xiaomei Liu · 2025
In order to solve the problem of slow image stitching, which cannot meet the requirements of real-time high-resolution image stitching, this paper proposes a fast image stitching method based on ORB (Oriented FAST and Rotated BRIEF) algorithm and MSAC (M-estimator Sample Consumption) algorithm. ORB algorithm matching speed is very fast, which can meet the real-time requirements of image stitching. First, this paper uses ORB algorithm to extract feature points; MSAC is an improvement of RANSAC algorithm, which can eliminate the pseudo matching point pairs in image stitching. Therefore, this paper optimizes the matching point pairs based on MSAC algorithm, removes the wrong matching point pairs, and then solves the image transformation matrix through the correct matching point pairs; Finally, bilinear interpolation fusion algorithm is used to eliminate visible seams and remove stitching traces. The experimental results show that the algorithm in this paper not only has obvious advantages in processing speed, but also has better quality of image stitching according to relevant evaluation indicators.