EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

MaLCA: Point Cloud Registration with Mamba-Enhanced Features and Local Correspondence Augmentation

Yuchen Huo, Longyun Zhang, Huijuan Guo, Jingyi Gong, Liqun Kuang, Xie Han, Fengguang Xiong

Algorithms · 11.05.2026 · 19 (5) · 380

DOI: 10.3390/a19050380 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Algorithms, online erschienen am 11.05.2026. Verfasst von Yuchen Huo und 6 weiteren Autor:innen. Die Arbeit nennt 39 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Natural Science Foundation of Shanxi Province, National Natural Science Foundation of China, Foundation of Shanxi Key Laboratory of Machine Vision and Virtual Reality.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

High-quality correspondences are critical to the accuracy and robustness of point cloud registration. Existing Transformer-based methods are fundamentally constrained by the quadratic computational complexity of self-attention, resulting in limited scalability. Moreover, conventional outlier removal paradigms operate by pruning initial correspondences, and thus fail catastrophically in low-overlap scenarios where initial inliers are inherently scarce. To address these challenges, we propose MaLCA, a point cloud registration method based on Mamba-enhanced features and local correspondence augmentation. We first adopt KPFCN as the backbone to extract multi-scale geometric features from raw point clouds. A Mamba selective state space model then replaces self-attention for global context modeling with linear complexity, while cross-attention is retained to facilitate inter-point-cloud feature interaction. Rather than following the conventional subtraction-based outlier removal paradigm, we introduce a prior-guided local rematching strategy combined with a fused neighbor matching mechanism that iteratively constructs dense, high-quality correspondences from sparse initial inliers, fundamentally overcoming the bottleneck of inlier scarcity in challenging scenes. Extensive experiments on the 3DMatch/3DLoMatch and 4DMatch/4DLoMatch benchmarks demonstrate that MaLCA achieves competitive registration performance across both rigid and deformable scenarios, with particular advantages in low-overlap cases.

Förderung

  • Natural Science Foundation of Shanxi Province
  • National Natural Science Foundation of China
  • Foundation of Shanxi Key Laboratory of Machine Vision and Virtual Reality
  • Shanxi Province Science and Technology Major Special Plan “Reveal the List” project

Zitieren

APA

Huo, Y., Zhang, L., Guo, H., Gong, J., Kuang, L., Han, X., & Xiong, F. (2026). MaLCA: Point Cloud Registration with Mamba-Enhanced Features and Local Correspondence Augmentation. Algorithms, 19(5), 380. https://doi.org/10.3390/a19050380

Vancouver

Huo Y, Zhang L, Guo H, Gong J, Kuang L, Han X, et al. MaLCA: Point Cloud Registration with Mamba-Enhanced Features and Local Correspondence Augmentation. Algorithms. 2026;19(5):380. doi:10.3390/a19050380

BibTeX
@article{Huo2026,
  author = {Huo, Yuchen and Zhang, Longyun and Guo, Huijuan and Gong, Jingyi and Kuang, Liqun and Han, Xie and Xiong, Fengguang},
  title = {MaLCA: Point Cloud Registration with Mamba-Enhanced Features and Local Correspondence Augmentation},
  journal = {Algorithms},
  year = {2026},
  volume = {19},
  number = {5},
  pages = {380},
  doi = {10.3390/a19050380}
}
Kostenlos starten →