Abstract:Efficiently finding optimal correspondences between point clouds is crucial for solving both rigid and non-rigid point cloud registration problems. Existing methods often rely on geometric or semantic feature embedding to establish correspondences and estimate transformations or flow fields. Recently, state-of-the-art methods have employed RAFT-like iterative updates to refine the solution. However, these methods have certain limitations. Firstly, their iterative refinement design lacks transparency, and their iterative updates follow a fixed path during the refinement process, which can lead to suboptimal results. Secondly, these methods overlook the importance of refining or optimizing correspondences (or matching matrices) as a precursor to solving transformations or flow fields. They typically compute candidate correspondences based on distances in the point feature space. However, they only project the candidate matching matrix into some matrix space once with Sinkhorn or dual softmax operations to obtain final correspondences. This one-shot projected matching matrix may be far from the globally optimal one, and these approaches do not consider the distribution of the target matching matrix. In this paper, we propose a novel approach that exploits the Denoising Diffusion Model to predict a searching gradient for the optimal matching matrix within the Doubly Stochastic Matrix Space. During the reverse denoising process, our method iteratively searches for better solutions along this denoising gradient, which points towards the maximum likelihood direction of the target matching matrix. Our method offers flexibility by allowing the search to start from any initial matching matrix provided by the online backbone or white noise. Experimental evaluations on the 3DMatch/3DLoMatch and 4DMatch/4DLoMatch datasets demonstrate the effectiveness of our newly designed framework.

PosDiffNet: Positional Neural Diffusion for Point Cloud Registration in a Large Field of View with Perturbations

PointDifformer: Robust Point Cloud Registration With Neural Diffusion and Transformer

Diff-Reg v1: Diffusion Matching Model for Registration Problem

DOPNet: Achieving Accurate and Efficient Point Cloud Registration Based on Deep Learning and Multi-Level Features

PCRDiffusion: Diffusion Probabilistic Models for Point Cloud Registration

FreeReg: Image-to-Point Cloud Registration Leveraging Pretrained Diffusion Models and Monocular Depth Estimators

Diff-Reg: Diffusion Model in Doubly Stochastic Matrix Space for Registration Problem

PointpartNet: 3D point-cloud registration via deep part-based feature extraction

PARE-Net: Position-Aware Rotation-Equivariant Networks for Robust Point Cloud Registration

Diff-PCR: Diffusion-Based Correspondence Searching in Doubly Stochastic Matrix Space for Point Cloud Registration

Consistent Two-Flow Network for Tele-Registration of Point Clouds

PCRNet: Point Cloud Registration Network using PointNet Encoding

PointTr: Low-Overlap Point Cloud Registration with Transformer

Point Cloud Registration using Representative Overlapping Points

SE(3) Diffusion Model-based Point Cloud Registration for Robust 6D Object Pose Estimation

Deep Interactive Full Transformer Framework for Point Cloud Registration.

Distinctiveness oriented Positional Equilibrium for Point Cloud Registration

An Efficient and Stable Registration Framework for Large Point Clouds at Two Different Moments

Differentiable Registration of Images and LiDAR Point Clouds with VoxelPoint-to-Pixel Matching

OMNet: Learning Overlapping Mask for Partial-to-Partial Point Cloud Registration