Point cloud generation for critical transportation infrastructure through Bézier curve

Abstract

While Light Detection and Ranging (LiDAR)-based sensors exhibit considerable potential for transportation infrastructure management, not all infrastructure elements can be comprehensively captured by point clouds, leading to the formation of undesirable “holes” due to both temporary and permanent occlusions. It is imperative to devise mechanisms for identifying and predicting the missing data within these “holes” to ensure the continuous acquisition of critical inventory information. This paper describes a method for generating point clouds based on Bézier curves, which effectively fills the voids within the infrastructure. This method comprises three integral processes, including angle-based boundary detection, identification of principal elevation change, and Bézier curve-based hole filling. The method demonstrates promising results on different roadside surfaces and at different ranges of scales of “holes”. Case studies on the sidewalk, and sound barrier inventories show that the proposed method can significantly improve the quality of the point cloud data for subsequent measurements.