Difference of Normals as a Multi-Scale Operator in Unorganized Point Clouds

TL;DR

This paper introduces the Difference of Normals (DoN), a multi-scale operator for unorganized 3D point clouds, enabling efficient segmentation and pre-processing for object recognition in large outdoor LIDAR datasets.

Contribution

The paper presents a novel multi-scale operator, DoN, specifically designed for unorganized 3D point clouds, demonstrating its effectiveness in segmentation and pre-processing tasks.

Findings

DoN effectively segments large 3D point clouds into meaningful clusters.

The operator improves semi-automatic annotation and object recognition workflows.

Quantitative evaluation shows robust performance on outdoor LIDAR datasets.

Abstract

A novel multi-scale operator for unorganized 3D point clouds is introduced. The Difference of Normals (DoN) provides a computationally efficient, multi-scale approach to processing large unorganized 3D point clouds. The application of DoN in the multi-scale filtering of two different real-world outdoor urban LIDAR scene datasets is quantitatively and qualitatively demonstrated. In both datasets the DoN operator is shown to segment large 3D point clouds into scale-salient clusters, such as cars, people, and lamp posts towards applications in semi-automatic annotation, and as a pre-processing step in automatic object recognition. The application of the operator to segmentation is evaluated on a large public dataset of outdoor LIDAR scenes with ground truth annotations.