MAGE: View-guided Point Cloud Completion with Efficient Modality Alignment and Adaptive Geometry Enhancement

2026-06-30

AI Summary

This work addresses the challenges of cross-modal geometric inconsistency, weak modality alignment, and insufficient self-geometric enhancement in single-view image-guided point cloud completion by proposing a unified geometry-aware framework. The method introduces a geometry-aware modality alignment mechanism, an adaptive geometry-aware self-attention module, and a geometry-aware anchor refinement strategy. These components are integrated with a shared self-attention Transformer, cross-modal reconstruction supervision, and spatially varying feature fusion to achieve effective modality alignment and enhanced local geometric detail. Extensive experiments on multiple synthetic and real-world datasets demonstrate that the proposed approach significantly outperforms existing methods, yielding notably higher accuracy and completeness in point cloud completion results.

Abstract

View-based point cloud completion aims to recover a complete 3D shape from a partial point cloud, guided by a single-view image. However, existing approaches often suffer from limited performance due to weak modality alignment and limited self-geometry enhancement. To overcome these challenges, we propose a unified geometry-aware framework that integrates efficient modality alignment and adaptive geometry enhancement, mainly to address cross-modal geometric inconsistency of view-guided point cloud completion. Specifically, we propose a geometry-aware modality alignment by integrating a shared self-attention Transformer and cross-modality reconstruction supervision, which aims to bring features of the image and point cloud close to each other in a shared latent space describing the 3D object. To enhance the perception of global shape and local geometric details, we propose an adaptive geometry-aware self-attention module, which simultaneously considers local geometry-aware attention computation and the spatially-variant feature fusion. In addition, we apply a geometry-perceptive anchor refinement module to reorganize the anchor points (representing a local region of the shape) under appropriate supervision, further boosting the completion performance of our method. Extensive experiments on both synthetic and real-world datasets demonstrate that our method achieves superior performance over existing approaches. Our code will be available at https://github.com/weizequan/MAGE.

Problem

• point cloud completion
• view-guided
• modality alignment
• geometry enhancement
• cross-modal inconsistency

Innovation

• modality alignment
• geometry-aware attention
• point cloud completion
• view-guided reconstruction
• adaptive feature fusion