DualGenerator: Information Interaction-Based Generative Network for Point Cloud Completion

Point cloud completion estimates complete shapes from incomplete point clouds to obtain higher-quality point cloud data. Most existing methods only consider global object features, ignoring spatial and semantic information of adjacent points. They cannot distinguish structural information well between different object parts, and the robustness of models is poor. To tackle these challenges, we propose an information interaction-based generative network for point cloud completion (DualGenerator). It contains an upper adversarial generation path and a lower variational generation path, which interact with each other and share weights. DualGenerator introduces a local refinement module in lower path, which captures general structures from partial inputs, and then refines shape details of the point cloud. It promotes completion in the unknown region and makes a distinction between different parts more obvious.The upper path effectively provides support for completion by establishing a comprehensive distribution of complete point clouds, while the design of DGStyleGan enhances the robustness of network. Qualitative and quantitative evaluations demonstrate that our method is superior on MVP and Completion3D datasets. The performance will not degrade significantly after adding noise interference or sparse sampling.