GEOMETRIC QUALITY ANALYSIS OF TERRESTRIAL LASER SCANNING DATA FOR INDUSTRIAL USAGE

Abstract

Terrestrial laser scanning is a potential emerging technology increasingly used in several applications, including reverse engineering, digital reconstruction, deformation monitoring, forensic crime scene preservation, and construction (AEC) applications. The data tolerance accepted in these applications ranges from tens of millimeters (e.g., historical monument digitization) to tens of micrometers (e.g., industrial high-precision manufacturing and assembling). Instrument mechanism, atmospheric conditions, object surface characteristics, and scan geometry are the four main factors that affect the laser point clouds produced by the Time of Flight (TOF) 3D laser scanner. Consequently, research groups worldwide have put a significant effort into modeling the sources of TOF-TLS errors and design-specific performance evaluation methodologies. This paper investigated the influence of scanning geometry parameterized by incidence angle and the range on the quality of TOF-TLS data in industrial sites. The quality of an indoor sample dataset of an industrial case study was studied and assessed. The results showed that the incidence angle and range parameters substantially impacted the quality of the TOF-TLS data. The suggested methodology can accurately correct the laser data to eliminate the incidence angle and range effects. A revised and optimized point cloud dataset was reconstructed by utilizing these features in conjunction with the approximated quality of the individual points. Furthermore, when assessing the quality of individual point clouds, the accuracy validation obtained through the RMSE value was 3 mm based on ground-truth reference points. On the other hand, the standard deviation values computed through the Multi-Scale Model-to-model cloud (M3C2) analysis were revealed to reach 1mm, which shows better performance results than the Cloud-to-Cloud (C2C) and Cloud-to-Model (C2M) comparison analysis. However, the proposed method may result in the elimination of several significant laser points. These points of high incidence angle values are not eliminated in every instance. The effect of scanning geometry, represented by the angle of incidence with the normalized intensity of the scanning points, should be studied intensively in future studies.