Self-calibration of terrestrial laser scanner using a M3C2-based planar patch algorithm

Abstract

Target- or intensity-feature-based calibration has been extensively explored for in-situ calibration of terrestrial laser scanners (TLSs). Those calibration approaches main drawbacks include the necessity of mounting physical targets in the scan area and the need for the presence of rich geometrical features allowing a feature-based distinction of different points in the point cloud. We propose an autonomous TLS calibration algorithm using planar patches that are ubiquitous in urban environments and can be found without manual preparation beforehand. The scanner calibration parameters are estimated and updated by minimizing the M3C2 normal distances between corresponding planar patches. Unlike target- or keypoint-based approaches, only medium-resolution scan data is needed. A comprehensive set of calibration model parameters are estimated using scans acquired from multiple positions, thereby providing a further advantage over current keypoint-based approaches that estimate only two-face sensitive parameters. To increase speed and reduce memory consumption, we propose to execute the calibration using only parts of scan data from a single station. Studies with two high-precision scanners, Leica RTC360 and Z+F Imager 5016, show results consistent with the laborious target-based approach with improved precision. Similar two-face sensitive calibration performance to the target- and keypoint-based approaches has been achieved, but with only subsets of the point cloud using the proposed algorithm.