The strip adjustment of mobile LiDAR point clouds using iterative closest point (ICP) algorithm

Mobile light detection and ranging (LiDAR) scanning systems consist of global navigation satellite system (GNSS) and inertial navigation system (INS) equipment for positioning in addition to LiDAR sensors and cameras. All of these units work collaboratively to generate an accurate point cloud positioned in a three-dimensional (3D) geodetic coordinate system. However, various factors affect the measurement quality with mobile LiDAR systems (MLS), and these factors include inertial drift problems, GNSS positioning errors, errors stemming from insufficient calibration, etc. The measurements with multiple scanners in mobile systems require a calibration process to deal with the inconsistency problem in overlapping strips. The calibration is done between the laser, camera, INS, and GNSS sensors. The cameras and laser scanners become ready to use after the calibration stage. The strip adjustment is a crucial process for eliminating the errors, which come from the GNSS/INS measurements and remain despite the calibration process, in obtaining an accurate point cloud of the scanned structure. This study aims testing and clarifying the performance of the iterative closest point (ICP) algorithm in precise registration of LiDAR strips and hence proposes a data-driven approach for strip adjustment in mobile LiDAR scanning applications. In the result of the numerical case study, fine registration of the point clouds obtained from forward and backward scanning using ICP yields a 3.2 cm root mean square error and 1.5 cm mean difference, which corresponds to an almost 84% improvement in mean differences of the point clouds in comparison to the initial phase before the ICP registration. Thus, the results clarify that the ICP algorithm succeeds in the strip adjustment process by reducing the average difference between the mutual points of multi-strip point clouds in mobile LiDAR applications.