Plane-based global registration for pavement 3D reconstruction using hybrid solid-state LiDAR point cloud

Abstract

This study aims to utilize the hybrid solid-state LiDAR (SSL) point cloud to achieve pavement 3D reconstruction and overcome the stratification problem of the pavement point cloud. A low-cost Mobile Laser Scanning (MLS) system based on the hybrid SSL is constructed beforehand. Then, a region of interest (ROI) extraction algorithm is designed to filter the useless point cloud. Finally, a plane-based global registration (PGR) approach is proposed to address the stratification problem of vertical direction. PGR includes two parts: coarse registration and refined registration. The coarse registration extracts the pavement plane to reduce the vertical error. The refined registration utilizes the Iterative Closest Point (ICP) algorithm to reduce the registration error further. Compared with the mainstream registration algorithms, PGR shows the highest registration accuracy in different road scenarios and maintains excellent real-time performance. Furthermore, the reconstruction accuracy is kept within 1 cm, sufficient to extract pavement information.