Down-Sampling of Large LiDAR Dataset in the Context of Off-Road Objects Extraction

Abstract

Nowadays, LiDAR (Light Detection and Ranging) is used in many fields, such as transportation. Thanks to the recent technological improvements, the current generation of LiDAR mapping instruments available on the market allows to acquire up to millions of three-dimensional (3D) points per second. On the one hand, such improvements allowed the development of LiDAR-based systems with increased productivity, enabling the quick acquisition of detailed 3D descriptions of the objects of interest. However, on the other hand, the extraction of the information of interest from such huge amount of acquired data can be quite challenging and time demanding. Motivated by such observation, this paper proposes the use of the Optimum Dataset method in order to ease and speed up the information extraction phase by significantly reducing the size of the acquired dataset while preserving (retain) the information of interest. This paper focuses on the data reduction of LiDAR datasets acquired on roads, with the goal of extraction the off-road objects. Mostly motivated by the need of mapping roads and quickly determining car position along a road, the development of efficient methods for the extraction of such kind of information is becoming a hot topic in the research community.

Keywords

• off-road objects
• OptD method
• reduction
• LiDAR