FS-I2P: A Hierarchical Focus–Sweep Registration Network with Dynamically Allocated Depth

Zhixin Cheng, Yujia Chen, Xujing Tao, Bohao Liao, Xiaotian Yin, Baoqun Yin, Tianzhu Zhang
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:18782-18799, 2026.

Abstract

Image-to-point cloud registration is often challenged by viewpoint changes, cross-modal discrepancies, and repetitive textures, which induce scale ambiguity and consequently lead to erroneous correspondences. Recent detection-free methods alleviate this issue by leveraging multi-scale features and transformer-based interactions. However, they still suffer from attention drift across layers and intra-scale inconsistencies, hindering precise registration. Inspired by complex scene observation, we propose a “Focus–Sweep” paradigm and develop a Hierarchical Mamba Interaction Module within an SSM-based framework to enhance multi-level cross-modal feature association. In addition, we introduce a Dynamic Layer Allocation Strategy that adaptively determines the iteration depth to better exploit geometric constraints and improve matching robustness. Extensive experiments and ablations on two benchmarks, RGB-D Scenes V2 and 7-Scenes, demonstrate that our approach achieves state-of-the-art performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-cheng26b, title = {{FS}-{I}2{P}: A Hierarchical Focus–Sweep Registration Network with Dynamically Allocated Depth}, author = {Cheng, Zhixin and Chen, Yujia and Tao, Xujing and Liao, Bohao and Yin, Xiaotian and Yin, Baoqun and Zhang, Tianzhu}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {18782--18799}, year = {2026}, editor = {Zhang, Tong and Dudik, Miroslav and Jaggi, Martin and Agarwal, Alekh and Li, Sharon and Schuurmans, Dale and Zhu, Jerry and Berkenkamp, Felix and Dong, Hanze and Bietti, Alberto}, volume = {306}, series = {Proceedings of Machine Learning Research}, month = {06--11 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v306/main/assets/cheng26b/cheng26b.pdf}, url = {https://proceedings.mlr.press/v306/cheng26b.html}, abstract = {Image-to-point cloud registration is often challenged by viewpoint changes, cross-modal discrepancies, and repetitive textures, which induce scale ambiguity and consequently lead to erroneous correspondences. Recent detection-free methods alleviate this issue by leveraging multi-scale features and transformer-based interactions. However, they still suffer from attention drift across layers and intra-scale inconsistencies, hindering precise registration. Inspired by complex scene observation, we propose a “Focus–Sweep” paradigm and develop a Hierarchical Mamba Interaction Module within an SSM-based framework to enhance multi-level cross-modal feature association. In addition, we introduce a Dynamic Layer Allocation Strategy that adaptively determines the iteration depth to better exploit geometric constraints and improve matching robustness. Extensive experiments and ablations on two benchmarks, RGB-D Scenes V2 and 7-Scenes, demonstrate that our approach achieves state-of-the-art performance.} }
Endnote
%0 Conference Paper %T FS-I2P: A Hierarchical Focus–Sweep Registration Network with Dynamically Allocated Depth %A Zhixin Cheng %A Yujia Chen %A Xujing Tao %A Bohao Liao %A Xiaotian Yin %A Baoqun Yin %A Tianzhu Zhang %B Proceedings of the 43rd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2026 %E Tong Zhang %E Miroslav Dudik %E Martin Jaggi %E Alekh Agarwal %E Sharon Li %E Dale Schuurmans %E Jerry Zhu %E Felix Berkenkamp %E Hanze Dong %E Alberto Bietti %F pmlr-v306-cheng26b %I PMLR %P 18782--18799 %U https://proceedings.mlr.press/v306/cheng26b.html %V 306 %X Image-to-point cloud registration is often challenged by viewpoint changes, cross-modal discrepancies, and repetitive textures, which induce scale ambiguity and consequently lead to erroneous correspondences. Recent detection-free methods alleviate this issue by leveraging multi-scale features and transformer-based interactions. However, they still suffer from attention drift across layers and intra-scale inconsistencies, hindering precise registration. Inspired by complex scene observation, we propose a “Focus–Sweep” paradigm and develop a Hierarchical Mamba Interaction Module within an SSM-based framework to enhance multi-level cross-modal feature association. In addition, we introduce a Dynamic Layer Allocation Strategy that adaptively determines the iteration depth to better exploit geometric constraints and improve matching robustness. Extensive experiments and ablations on two benchmarks, RGB-D Scenes V2 and 7-Scenes, demonstrate that our approach achieves state-of-the-art performance.
APA
Cheng, Z., Chen, Y., Tao, X., Liao, B., Yin, X., Yin, B. & Zhang, T.. (2026). FS-I2P: A Hierarchical Focus–Sweep Registration Network with Dynamically Allocated Depth. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:18782-18799 Available from https://proceedings.mlr.press/v306/cheng26b.html.

Related Material