GS-Matching: Reconsidering Feature Matching task in Point Cloud Registration
Yaojie Zhang, Tianlun Huang, Weijun Wang, Wei Feng
TL;DR
This work tackles the challenge of feature matching in point cloud registration under partial overlap by arguing against the reliability of one-to-one assignment when used broadly. It introduces GS-Matching, a stable, Gale–Shapley–inspired heuristic that emphasizes non-repetitive inliers and efficiency, complemented by probabilistic analysis of feature scores. Through extensive experiments on indoor and outdoor datasets, GS-Matching consistently improves registration recall and reduces error metrics while acting as a plug-in module for existing PCR pipelines. These results suggest that stable, probability-informed matching policies can substantially enhance 3D registration in diverse environments and descriptor regimes.
Abstract
Traditional point cloud registration (PCR) methods for feature matching often employ the nearest neighbor policy. This leads to many-to-one matches and numerous potential inliers without any corresponding point. Recently, some approaches have framed the feature matching task as an assignment problem to achieve optimal one-to-one matches. We argue that the transition to the Assignment problem is not reliable for general correspondence-based PCR. In this paper, we propose a heuristics stable matching policy called GS-matching, inspired by the Gale-Shapley algorithm. Compared to the other matching policies, our method can perform efficiently and find more non-repetitive inliers under low overlapping conditions. Furthermore, we employ the probability theory to analyze the feature matching task, providing new insights into this research problem. Extensive experiments validate the effectiveness of our matching policy, achieving better registration recall on multiple datasets.