Skeleton-bridged Point Completion: From Global Inference to Local Adjustment

TL;DR

This paper introduces SK-PCN, a shape completion method that predicts a 3D skeleton for global structure and refines surface details through local adjustments, outperforming previous methods on diverse object categories.

Contribution

The paper proposes a skeleton-bridged network that decouples structure estimation from surface reconstruction, enabling better detail recovery and surface refinement in point cloud completion.

Findings

Outperforms existing methods on multiple object categories.

Supports point normal estimation for full surface mesh reconstruction.

Effectively combines global skeleton prediction with local surface adjustments.

Abstract

Point completion refers to complete the missing geometries of objects from partial point clouds. Existing works usually estimate the missing shape by decoding a latent feature encoded from the input points. However, real-world objects are usually with diverse topologies and surface details, which a latent feature may fail to represent to recover a clean and complete surface. To this end, we propose a skeleton-bridged point completion network (SK-PCN) for shape completion. Given a partial scan, our method first predicts its 3D skeleton to obtain the global structure, and completes the surface by learning displacements from skeletal points. We decouple the shape completion into structure estimation and surface reconstruction, which eases the learning difficulty and benefits our method to obtain on-surface details. Besides, considering the missing features during encoding input points, SK-PCN adopts a local adjustment strategy that merges the input point cloud to our predictions for surface refinement. Comparing with previous methods, our skeleton-bridged manner better supports point normal estimation to obtain the full surface mesh beyond point clouds. The qualitative and quantitative experiments on both point cloud and mesh completion show that our approach outperforms the existing methods on various object categories.