Surfel-based 3D Registration with Equivariant SE(3) Features
This addresses alignment issues in 3D reconstruction for applications like remote sensing or digital heritage, offering an incremental improvement over existing methods.
The paper tackles the problem of 3D point cloud registration by proposing a surfel-based method that learns SE(3) equivariant features to handle noisy inputs and aggressive rotations, achieving superior and robust performance compared to state-of-the-art methods on indoor and outdoor datasets.
Point cloud registration is crucial for ensuring 3D alignment consistency of multiple local point clouds in 3D reconstruction for remote sensing or digital heritage. While various point cloud-based registration methods exist, both non-learning and learning-based, they ignore point orientations and point uncertainties, making the model susceptible to noisy input and aggressive rotations of the input point cloud like orthogonal transformation; thus, it necessitates extensive training point clouds with transformation augmentations. To address these issues, we propose a novel surfel-based pose learning regression approach. Our method can initialize surfels from Lidar point cloud using virtual perspective camera parameters, and learns explicit $\mathbf{SE(3)}$ equivariant features, including both position and rotation through $\mathbf{SE(3)}$ equivariant convolutional kernels to predict relative transformation between source and target scans. The model comprises an equivariant convolutional encoder, a cross-attention mechanism for similarity computation, a fully-connected decoder, and a non-linear Huber loss. Experimental results on indoor and outdoor datasets demonstrate our model superiority and robust performance on real point-cloud scans compared to state-of-the-art methods.