Learning to Denoise and Remove Outliers from Dense Point Clouds

Marie-Julie Rakotosaona1, Vittorio La Barbera2, Paul Guerrero2, Niloy J. Mitra2,3, Maks Ovsjanikov1

  • 1LIX, Ecole Polytechnique, CNRS
  • 2University College London
  • 3Adobe Research


Point clouds obtained with 3D scanners or by image based reconstruction techniques are often corrupted with significant amount of noise and outliers. Traditional methods for point cloud denoising largely rely on local surface fitting (e.g. jets or MLS surfaces), local or non‐local averaging or on statistical assumptions about the underlying noise model. In contrast, we develop a simple data‐driven method for removing outliers and reducing noise in unordered point clouds. We base our approach on a deep learning architecture adapted from PCPNet, which was recently proposed for estimating local 3D shape properties in point clouds. Our method first classifies and discards outlier samples, and then estimates correction vectors that project noisy points onto the original clean surfaces. The approach is efficient and robust to varying amounts of noise and outliers, while being able to handle large densely sampled point clouds. In our extensive evaluation, both on synthetic and real data, we show an increased robustness to strong noise levels compared to various state‐of‐the‐art methods, enabling accurate surface reconstruction from extremely noisy real data obtained by range scans. Finally, the simplicity and universality of our approach makes it very easy to integrate in any existing geometry processing pipeline.


We present PointDenoisingBenchmark dataset. Our dataset features 28 different shapes, which we split into 18 training shapes and 10 test shapes.

  • PointDenoisingBenchmark for outliers removal: contains noisy point clouds with different levels of gaussian noise and the corresponding clean ground truths.
  • PointDenoisingBenchmark for denoising: contains noisy point clouds with different levels of noise and density of outliers and the corresponding clean ground truths.


We present PointCleanNet, a two-stage network that takes a raw point cloud and first removes outliers (top) and then denoises the remaining pointset (bottom). Our method, unlike many traditional approaches, is parameter-free and automatically discovers and preserves high-curvature features without requiring additional information about the underlying surface type or device characteristics.

We show the results of our pipeline on noisy real data (first 3 from left) and noisy synthetic data (last 3).


If you use our work or our data, please cite our paper:

  title={PointCleanNet: Learning to Denoise and Remove Outliers from Dense Point Clouds},
  author={Rakotosaona, Marie-Julie and La Barbera, Vittorio and Guerrero, Paul and Mitra, Niloy J and Ovsjanikov, Maks},
  journal={Computer Graphics Forum},


Parts of this work were supported by the KAUST OSR Award No. CRG-2017-3426, a gift from the NVIDIA Corporation, the ERC Starting Grants EXPROTEA (StG-2017-758800) and SmartGeometry (StG-2013-335373), a Google Faculty Award, and gifts from Adobe.