Correspondence-Aware Chamfer Distance for Point Cloud Completion: Learning Distribution-Aligned Reconstruction
IEEE Access · 2026 · 14 · 102668-102678
DOI: 10.1109/access.2026.3711059 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Access, 2026. Verfasst von Hangjae Cho und 1 weiteren Autor:innen. Die Arbeit nennt 30 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0).
Abstract (Originalsprache)
Point cloud completion addresses the problem of reconstructing complete 3D geometric shapes from incomplete or partial observations. While deep learning-based methods have achieved success using Chamfer Distance (CD) as a loss function, inherent limitations remain. The nearest-neighbor correspondence mechanism of CD allows many-to-one mappings between point clouds, leading to distributional imbalance and undesired clustering artifacts. This distributional bias accumulates during training without being captured by CD metrics, resulting in misalignment with ground-truth distributions. To mitigate this issue, we 1) propose categorization to distinguish aligned from misaligned point correspondences; and 2) introduce a correspondence-aware loss function consisting of two complementary terms—an uncovered-region guidance loss and an interior-spreading loss—that provide explicit relocation guidance for misaligned points. Consequently, our approach achieves improved completion and better alignment with ground-truth distributions. The proposed loss reuses correspondence information already computed during CD, maintaining computational efficiency comparable to that of CD while avoiding the exponential overhead of EMD. Through experiments conducted across five point cloud completion architectures and three benchmark datasets, we demonstrate that the proposed loss function improves completion models quantitatively and qualitatively.
Zitieren
APA
Cho, H., & Jun, K. (2026). Correspondence-Aware Chamfer Distance for Point Cloud Completion: Learning Distribution-Aligned Reconstruction. IEEE Access, 14, 102668-102678. https://doi.org/10.1109/access.2026.3711059
Vancouver
Cho H, Jun K. Correspondence-Aware Chamfer Distance for Point Cloud Completion: Learning Distribution-Aligned Reconstruction. IEEE Access. 2026;14:102668-102678. doi:10.1109/access.2026.3711059
BibTeX
@article{Cho2026,
author = {Cho, Hangjae and Jun, Kyungkoo},
title = {Correspondence-Aware Chamfer Distance for Point Cloud Completion: Learning Distribution-Aligned Reconstruction},
journal = {IEEE Access},
year = {2026},
volume = {14},
pages = {102668--102678},
doi = {10.1109/access.2026.3711059}
}
EUVIMED