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

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.

Many real-world 3D applications need to align two point clouds, which are sets of points representing the shape of an object or scene. This is easy when the object only moves as a whole, but much harder when parts of the object bend, stretch, or move independently. Existing methods usually handle these two cases separately, which makes them less flexible across different types of data. In this work, we propose RGGT, a unified method that can handle both rigid and deformable point cloud alignment in one framework. The key idea is to use knowledge learned from large 3D generative models to better understand both local shape details and overall object structure. Experiments on several standard datasets show that RGGT achieves strong performance across both rigid and non-rigid alignment tasks.