English

Approximate Algorithms for Chamfer Distance Under Translation

Data Structures and Algorithms 2026-05-26 v1 Computational Complexity Computational Geometry

Abstract

Given two sets of points A and B, A=m|A| = m, B=n|B| = n, the Chamfer distance from AA to BB is defined as CD(A,B)=aAminbBd(a,b)\operatorname{CD}(A,B) = \sum_{a\in A} \min_{b\in B} d(a,b), where dd is a distance metric. Chamfer distance is a popular measure of dissimilarity between two sets of points that has seen increasing usage in computer vision and information retrieval as a substitute for the more computationally demanding Earth Mover's distance. We propose a new problem, Chamfer distance under translation, defined as CDuT(A,B):=mintRdCD(A+t,B)\operatorname{CDuT}(A,B) :=\min_{t\in \mathbb{R}^d} \operatorname{CD}(A+t,B), where A+tA+t denotes the translation of every point in AA by tt. Chamfer distance under translation is valuable in cases where translations capture aspects of the data unlikely to be relevant for dissimilarity, such as temporal, spatial, or other semantic information. For Chamfer distance under translation, we provide four algorithms: (1) an exact quadratic time algorithm in one dimension, (2) a near quadratic time (2+ε2+\varepsilon)-approximation algorithm in higher dimensions, (3) a (1+ε)(1+\varepsilon)-approximation algorithm with running time O(mn2ε(d+1))\mathcal{O}(mn^2\varepsilon^{-(d+1)}), and (4) a near-quadratic time (1+ε)(1+\varepsilon)-approximation algorithm for answering the decision version of CDuT\operatorname{CDuT} given a separation assumption on BB. We additionally explore the fine-grained complexity of CDuT\operatorname{CDuT}.

Keywords

Cite

@article{arxiv.2605.25280,
  title  = {Approximate Algorithms for Chamfer Distance Under Translation},
  author = {Gil Halevi and Daniel Zhang and Jason Zhang},
  journal= {arXiv preprint arXiv:2605.25280},
  year   = {2026}
}

Comments

Preprint. 18 pages