Learned Radius Estimation for UDF-Based Point Cloud Reconstruction
For 3D reconstruction from point clouds, this work addresses the bottleneck of support radius selection in local-patch UDF methods, offering an incremental improvement.
Local-patch UDF methods for point cloud reconstruction suffer from fixed or heuristically chosen support radii. The authors propose a learned per-query radius selector that improves fine-scale reconstruction accuracy.
Surface reconstruction from point clouds is important for consumer-grade 3D capture, including AR/VR and indoor scanning. Local-patch Unsigned Distance Field (UDF) methods are lightweight and generalizable, but their accuracy depends on the support radius, traditionally fixed or selected by a one-dimensional curvature heuristic that cannot capture heterogeneous local geometry. We propose a learned per-query radius selector that predicts a continuous support radius and plugs into a frozen LoSF-UDF backbone. The selector is trained using off-grid target radii obtained by parabolic interpolation of cached UDF error curves. Experiments show improved fine-scale reconstruction accuracy.