Efficient image matching and panoramic image splicing technology based on improved SURF algorithm
Yu Yang · The Imaging Science Journal · 2025
This study proposal is an improved Accelerated Robust Feature (ARF) algorithm to enhance the efficiency and accuracy of image matching in panoramic image stitching, especially in complex scenes. The algorithm optimizes feature point detection and matching, improving image registration. A progressive transition technique is used during image fusion to achieve seamless stitching. The method achieved 98% accuracy on the HPatches dataset and 97% on SARptical, outperforming existing approaches. In Experiment 3, it scored 0.857 in content quality, 0.921 in structural consistency, 0.882 in texture fidelity, and 0.901 in spatial distribution stability. The results demonstrate that the proposed method offers high stitching quality and improved speed, making it effective for generating wide-field panoramic images.