Decomposed Global Optimization for Robust Point Matching with Low-Dimensional Branching
Abstract
Numerous applications require algorithms that can align partially overlapping point sets while maintaining invariance to geometric transformations (e.g., similarity, affine, rigid). This paper introduces a novel global optimization method for this task by minimizing the objective function of the Robust Point Matching (RPM) algorithm. We first reveal that the original RPM objective is a cubic polynomial. Through a concise variable substitution, we transform this objective into a quadratic function. By leveraging the convex envelope of bilinear monomials, we derive a tight lower bound for this quadratic function. This lower bound problem conveniently and efficiently decomposes into two parts: a standard linear assignment problem (solvable in polynomial time) and a low-dimensional convex quadratic program. Furthermore, we devise a specialized Branch-and-Bound (BnB) algorithm that branches exclusively on the transformation parameters, which significantly accelerates convergence by confining the search space. Experiments on 2D and 3D synthetic and real-world data demonstrate that our method, compared to state-of-the-art approaches, exhibits superior robustness to non-rigid deformations, positional noise, and outliers, particularly in scenarios where outliers are distinct from inliers.
Cite
@article{arxiv.2405.08589,
title = {Decomposed Global Optimization for Robust Point Matching with Low-Dimensional Branching},
author = {Wei Lian and Zhesen Cui and Fei Ma and Hang Pan and Wangmeng Zuo and Jianmei Zhang},
journal= {arXiv preprint arXiv:2405.08589},
year = {2025}
}