Our method first interpolates the low-res point cloud according to a given upsampling rate. And then refine the positions of the interpolated points with an iterative optimization process, guided by a trained model estimating the difference between the current point cloud and the high-res target.

Most existing point cloud upsampling methods have roughly three steps: feature extraction, feature expansion and 3D coordinate prediction. However, they usually suffer from two critical issues: (1) fixed upsampling rate after one-time training, since the feature expansion unit is customized for each upsampling rate; (2) outliers or shrinkage artifact caused by the difficulty of precisely predicting 3D coordinates or residuals of upsampled points. To adress them, we propose a new framework for accurate point cloud upsampling that supports arbitrary upsampling rates. Our method first interpolates the low-res point cloud according to a given upsampling rate. And then refine the positions of the interpolated points with an iterative optimization process, guided by a trained model estimating the difference between the current point cloud and the high-res target. Extensive quantitative and qualitative results on benchmarks and downstream tasks demonstrate that our method achieves the state-of-the-art accuracy and efficiency.

Our P2PNet contains two submodules: a feature extractor and point-to-point distance regressor. For feature extractor, we stack an initial MLP and three dense blocks with intra-block dense connection, where each dense block has three convolution groups to capture local features and one transition down layer to reduce channel. In distance regressor, we estimate the point-to-point distance for each query point conditioned on the extracted local and global features.

Y. He, D. Tang, Y. Zhang, X. Xue, Y. Fu Grad-PU: Arbitrary-Scale Point Cloud Upsampling via Gradient Descent with Learned Distance Functions CVPR 2023 [arXiv]     [GitHub]

Qualitative comparisons with SOTA.