Point Cloud Denoising and Outlier Detection with Local Geometric Structure by Dynamic Graph CNN
Authors:
Kosuke Nakayama,
Hiroto Fukuta,
Hiroshi Watanabe
Abstract:
The digitalization of society is rapidly developing toward the realization of the digital twin and metaverse. In particular, point clouds are attracting attention as a media format for 3D space. Point cloud data is contaminated with noise and outliers due to measurement errors. Therefore, denoising and outlier detection are necessary for point cloud processing. Among them, PointCleanNet is an effe…
▽ More
The digitalization of society is rapidly developing toward the realization of the digital twin and metaverse. In particular, point clouds are attracting attention as a media format for 3D space. Point cloud data is contaminated with noise and outliers due to measurement errors. Therefore, denoising and outlier detection are necessary for point cloud processing. Among them, PointCleanNet is an effective method for point cloud denoising and outlier detection. However, it does not consider the local geometric structure of the patch. We solve this problem by applying two types of graph convolutional layer designed based on the Dynamic Graph CNN. Experimental results show that the proposed methods outperform the conventional method in AUPR, which indicates outlier detection accuracy, and Chamfer Distance, which indicates denoising accuracy.
△ Less
Submitted 21 October, 2023; v1 submitted 11 October, 2023;
originally announced October 2023.