Morphing and Sampling Network for Dense Point Cloud Completion

• DOI: 10.1609/aaai.v34i07.6827
• Publication Date: 2020-06-29T18:34:43Z
• Authors: Minghua Liu; Lu Sheng; Sheng Yang; Jing Shao; Shi-Min Hu
• Keywords: Morphing

3D point cloud completion, the task of inferring complete geometric shape from a partial cloud, has been attracting attention in community. For acquiring high-fidelity dense clouds and avoiding uneven distribution, blurred details, or structural loss existing methods' results, we propose novel approach to two stages. Specifically, first stage, predicts but coarse-grained with collection parametric surface elements. Then, second it merges prediction input by sampling algorithm. Our method utilizes joint function guide distribution points. Extensive experiments verify effectiveness our demonstrate that outperforms methods both Earth Mover's Distance (EMD) Chamfer (CD).