RGGT: A Generative-Prior-Guided Transformer for Unified Rigid and Non-Rigid Point Cloud Registration

Chengyu Zheng, Songlin Yang, Jin Huang, Honghua Chen, Weiming Wang, Haoran Xie, Fu Lee Wang, Mingqiang Wei
Proceedings of the 43rd International Conference on Machine Learning, PMLR 306:163830-163849, 2026.

Abstract

Point cloud registration can be categorized into rigid and non-rigid settings depending on the motion characteristics of the underlying objects. Rigid alignment assumes a single global transformation under which corresponding points remain geometrically consistent across scales, whereas non-rigid alignment involves spatially varying deformations, where geometric similarity holds only locally and semantic correspondence dominates at larger scales. This multi-scale discrepancy creates an optimization gap that has made unified registration particularly challenging. To this end, we propose RGGT, a Generative-Prior-Guided Transformer that unifies rigid and non-rigid registration within a shared optimization space. Through coordinated design at the representation, architecture, and supervision levels, RGGT jointly captures local geometric details and global structural semantics: generative priors enrich point features with unified geometric-semantic cues; a Global-Self-Cross Attention module models long-range structure, local interaction, and bidirectional cross-shape reasoning; and a dual correspondence-reconstruction objective provides consistent supervision for both deformation types. Extensive experiments on rigid (ModelNet40, 3DMatch, KITTI) and non-rigid (4DMatch) benchmarks demonstrate that RGGT achieves state-of-the-art accuracy across both rigid and non-rigid settings within a single unified framework. Code is available at https://github.com/zhengcy-lambo/RGGT.

Cite this Paper

BibTeX
@InProceedings{pmlr-v306-zheng26ab, title = {{RGGT}: A Generative-Prior-Guided Transformer for Unified Rigid and Non-Rigid Point Cloud Registration}, author = {Zheng, Chengyu and Yang, Songlin and Huang, Jin and Chen, Honghua and Wang, Weiming and Xie, Haoran and Wang, Fu Lee and Wei, Mingqiang}, booktitle = {Proceedings of the 43rd International Conference on Machine Learning}, pages = {163830--163849}, 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/zheng26ab/zheng26ab.pdf}, url = {https://proceedings.mlr.press/v306/zheng26ab.html}, abstract = {Point cloud registration can be categorized into rigid and non-rigid settings depending on the motion characteristics of the underlying objects. Rigid alignment assumes a single global transformation under which corresponding points remain geometrically consistent across scales, whereas non-rigid alignment involves spatially varying deformations, where geometric similarity holds only locally and semantic correspondence dominates at larger scales. This multi-scale discrepancy creates an optimization gap that has made unified registration particularly challenging. To this end, we propose RGGT, a Generative-Prior-Guided Transformer that unifies rigid and non-rigid registration within a shared optimization space. Through coordinated design at the representation, architecture, and supervision levels, RGGT jointly captures local geometric details and global structural semantics: generative priors enrich point features with unified geometric-semantic cues; a Global-Self-Cross Attention module models long-range structure, local interaction, and bidirectional cross-shape reasoning; and a dual correspondence-reconstruction objective provides consistent supervision for both deformation types. Extensive experiments on rigid (ModelNet40, 3DMatch, KITTI) and non-rigid (4DMatch) benchmarks demonstrate that RGGT achieves state-of-the-art accuracy across both rigid and non-rigid settings within a single unified framework. Code is available at https://github.com/zhengcy-lambo/RGGT.} }
Endnote
%0 Conference Paper %T RGGT: A Generative-Prior-Guided Transformer for Unified Rigid and Non-Rigid Point Cloud Registration %A Chengyu Zheng %A Songlin Yang %A Jin Huang %A Honghua Chen %A Weiming Wang %A Haoran Xie %A Fu Lee Wang %A Mingqiang Wei %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-zheng26ab %I PMLR %P 163830--163849 %U https://proceedings.mlr.press/v306/zheng26ab.html %V 306 %X Point cloud registration can be categorized into rigid and non-rigid settings depending on the motion characteristics of the underlying objects. Rigid alignment assumes a single global transformation under which corresponding points remain geometrically consistent across scales, whereas non-rigid alignment involves spatially varying deformations, where geometric similarity holds only locally and semantic correspondence dominates at larger scales. This multi-scale discrepancy creates an optimization gap that has made unified registration particularly challenging. To this end, we propose RGGT, a Generative-Prior-Guided Transformer that unifies rigid and non-rigid registration within a shared optimization space. Through coordinated design at the representation, architecture, and supervision levels, RGGT jointly captures local geometric details and global structural semantics: generative priors enrich point features with unified geometric-semantic cues; a Global-Self-Cross Attention module models long-range structure, local interaction, and bidirectional cross-shape reasoning; and a dual correspondence-reconstruction objective provides consistent supervision for both deformation types. Extensive experiments on rigid (ModelNet40, 3DMatch, KITTI) and non-rigid (4DMatch) benchmarks demonstrate that RGGT achieves state-of-the-art accuracy across both rigid and non-rigid settings within a single unified framework. Code is available at https://github.com/zhengcy-lambo/RGGT.
APA
Zheng, C., Yang, S., Huang, J., Chen, H., Wang, W., Xie, H., Wang, F.L. & Wei, M.. (2026). RGGT: A Generative-Prior-Guided Transformer for Unified Rigid and Non-Rigid Point Cloud Registration. Proceedings of the 43rd International Conference on Machine Learning, in Proceedings of Machine Learning Research 306:163830-163849 Available from https://proceedings.mlr.press/v306/zheng26ab.html.

Related Material