DFG-PCN: Point Cloud Completion with Degree-Flexible Point Graph
This work solves the problem of incomplete point cloud reconstruction for applications like 3D scanning and robotics, though it is incremental as it builds on existing graph-based methods with adaptive improvements.
The paper tackles point cloud completion by addressing inefficiencies from fixed local region partitioning, proposing DFG-PCN which adaptively assigns node degrees based on detail-aware metrics and integrates geometry-aware features, resulting in consistent outperformance over state-of-the-art methods on multiple benchmark datasets.
Point cloud completion is a vital task focused on reconstructing complete point clouds and addressing the incompleteness caused by occlusion and limited sensor resolution. Traditional methods relying on fixed local region partitioning, such as k-nearest neighbors, which fail to account for the highly uneven distribution of geometric complexity across different regions of a shape. This limitation leads to inefficient representation and suboptimal reconstruction, especially in areas with fine-grained details or structural discontinuities. This paper proposes a point cloud completion framework called Degree-Flexible Point Graph Completion Network (DFG-PCN). It adaptively assigns node degrees using a detail-aware metric that combines feature variation and curvature, focusing on structurally important regions. We further introduce a geometry-aware graph integration module that uses Manhattan distance for edge aggregation and detail-guided fusion of local and global features to enhance representation. Extensive experiments on multiple benchmark datasets demonstrate that our method consistently outperforms state-of-the-art approaches.