Point Cloud Noise and Outlier Removal for Image-Based 3D Reconstruction
Katja Wolff (Disney Research Zurich, ETH Zurich)
Changil Kim (Disney Research Zurich)
Henning Zimmer (Disney Research Zurich)
Christopher Schroers (Disney Research Zurich)
Mario Botsch (Bielefeld University)
Olga Sorkine-Hornung (ETH Zurich)
Alexander Sorkine-Hornung (Disney Research Zurich)
3D International Conference on 3D Vision (3DV) 2016
October 25, 2016
Point sets generated by image-based 3D reconstruction techniques are often much noisier than those obtained using active techniques like laser scanning. Therefore, they pose greater challenges to the subsequent surface reconstruction (meshing) stage. We present a simple and effective method for removing noise and outliers from such point sets. Our algorithm uses the input images, and corresponding depth maps to remove pixels which are geometrically or photometrically inconsistent with the colored surface implied by the input. This allows standard surface reconstruction methods (such as Poisson surface reconstruction) to perform less smoothing and thus achieve higher quality surfaces with more features. Our algorithm is efficient, easy to implement, and robust to varying amounts of noise. We demonstrate the benefits of our algorithm in combination with a variety of state-of-the-art depth and surface reconstruction methods.
Download File "Point Cloud Noise and Outlier Removal for Image-Based 3D Reconstruction-Paper"
[pdf, 3.44 MB]