Hybrid Attention for Low-overlap Point Cloud Registration

Authors

• Huaijun Wang, Shaanxi Provincial Collaborative Innovation Center of Green Manufacturing for Modern Equipment, Xi'an University of Technology, Xi'an 710048, China; School of Computer Science and Engineering, Xi'an University of Technology, Xi'an 710048, China
• Zonghua Yu, School of Computer Science and Engineering, Xi'an University of Technology, Xi'an 710048, China
• Chenxin Xu, School of Computer Science and Engineering, Xi'an University of Technology, Xi'an 710048, China
• Junhuai Li, Shaanxi Provincial Collaborative Innovation Center of Green Manufacturing for Modern Equipment, Xi'an University of Technology, Xi'an 710048, China; School of Computer Science and Engineering, Xi'an University of Technology, Xi'an 710048, China
• Chenzhi Li, Shannxi Water Affairs Development Ecological Technology Research and Development Co., Ltd., Xi'an 710068, China
• Junjie Yin, School of Computer Science and Engineering, Xi'an University of Technology, Xi'an 710048, China

Abstract

Abstract: To address the significant registration errors caused by large initial pose deviations in low overlap ratio point cloud, this paper presents a two-stage optimization method that integrates selfattention and cross-attention mechanisms. Orientation feature representation is enhanced using equivariant self-attention, semantic alignment is achieved through invariant cross-attention, and rotational equivariance is modeled with an octahedral group for coarse registration. Various attention variants are employed to aggregate features along the anchor dimension, and a local-to-global module is utilized for fine registration. On the 3DLoMatch dataset, the proposed method achieves a registration recall of 77.0% and an inlier ratio of 44.3%. Even with the introduction of random rotations, it maintains a registration recall of 72.0%, and under varying point cloud densities, the recall remains above 75.0%. Experimental results demonstrate that the method effectively improves registration accuracy and rotational robustness in low-overlap scenarios through equivariant feature extraction and multi-attention fusion.

Recommended Citation

Wang, Huaijun; Yu, Zonghua; Xu, Chenxin; Li, Junhuai; Li, Chenzhi; and Yin, Junjie (2026) "Hybrid Attention for Low-overlap Point Cloud Registration," Journal of System Simulation: Vol. 38: Iss. 9, Article 5.
DOI: 10.16182/j.issn1004731x.joss.25-1261
Available at: https://dc-china-simulation.researchcommons.org/journal/vol38/iss9/5

First Page

2502

Last Page

2517

CLC

TP391

Recommended Citation

Wang Huaijun, Yu Zonghua, Xu Chenxin, et al. Hybrid Attention for Low-overlap Point Cloud Registration[J]. Journal of System Simulation, 2026, 38(9): 2502-2517.

Corresponding Author

Li Junhuai

DOI

10.16182/j.issn1004731x.joss.25-1261