Automatic Registration of Mobile LiDAR Data Using High-Reflectivity Traffic Signs

Abstract

Mobile light detection and ranging (LiDAR) is currently one of the most used instruments in road inspections to collect geometric and radiometric data. Different strips of the same area are common and bring problems in data registration related to the inaccuracies of global navigation satellite systems. This paper shows a methodology for automatic and robust data registration. It is based on segmentation of highly reflective data (i.e., road signs) and avoids the use of control points manually marked, typically used in the traditional techniques. The data set must be acquired under favorable climatological conditions. Root-mean-square errors in registration show values around 2 cm, while if the high-reflectivity segmentation is not used values higher than 8 cm are obtained.