Rethinking Multimodal Point Cloud Completion: A Completion-by-Correction Perspective

TL;DR

This paper proposes a new Completion-by-Correction paradigm for 3D point cloud completion, shifting from synthesis to guided refinement, resulting in more structurally consistent reconstructions by leveraging a topologically complete prior.

Contribution

It introduces a novel paradigm and PGNet framework that improve structural consistency in point cloud completion through feature-space correction and hierarchical refinement.

Findings

PGNet outperforms baselines with -23.5% Chamfer Distance

PGNet achieves +7.1% F-score

The new paradigm reduces structural artifacts in reconstructions

Abstract

Point cloud completion aims to reconstruct complete 3D shapes from partial observations, which is a challenging problem due to severe occlusions and missing geometry. Despite recent advances in multimodal techniques that leverage complementary RGB images to compensate for missing geometry, most methods still follow a Completion-by-Inpainting paradigm, synthesizing missing structures from fused latent features. We empirically show that this paradigm often results in structural inconsistencies and topological artifacts due to limited geometric and semantic constraints. To address this, we rethink the task and propose a more robust paradigm, termed Completion-by-Correction, which begins with a topologically complete shape prior generated by a pretrained image-to-3D model and performs feature-space correction to align it with the partial observation. This paradigm shifts completion from unconstrained synthesis to guided refinement, enabling structurally consistent and observation-aligned reconstruction. Building upon this paradigm, we introduce PGNet, a multi-stage framework that conducts dual-feature encoding to ground the generative prior, synthesizes a coarse yet structurally aligned scaffold, and progressively refines geometric details via hierarchical correction. Experiments on the ShapeNetViPC dataset demonstrate the superiority of PGNet over state-of-the-art baselines in terms of average Chamfer Distance (-23.5%) and F-score (+7.1%).