Experimental evaluation of a camera rig extrinsic calibration method based on retro-reflective markers detection. (July 2019)

• Record Type: Journal Article
• Authors: Chiodini, Sebastiano; Pertile, Marco; Giubilato, Riccardo; Salvioli, Federico; Barrera, Marco; Franceschetti, Paola; Debei, Stefano
• Is Part Of: Measurement. Volume 140(2019)
• Page Start: 47
• Page End: 55
• Publication Date: 2019-07
• Subjects: Calibration -- Motion capture -- Camera resectioning -- Image segmentation; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2019.03.036
• Languages: English
• ISSNs: 0263-2241

Abstract

Highlights: Extrinsic camera calibration method relative to motion capture system is presented. Three different spherical markers segmentation methods are analyzed. Two approaches to fuse the retro-reflective marker measurements are carried out. Calibration uncertainty propagation is evaluated via Monte Carlo simulation. Camera extrinsics are retrieved with an uncertainty of the order of the millimeter. Abstract: Nowadays, Motion Capture (MC) systems are used more and more to evaluate the accuracy of Visual Odometry and visual Simultaneous Localization and Mapping (SLAM) algorithms. However, the misalignment between the camera optical frame and the camera body frame, as tracked by a MC system, leads to a drift between the reconstructed trajectory and the reference one. In this work, we present a calibration method which estimates the relative orientation and position between this two reference frames. The proposed method is highly efficient because it uses a calibration target composed by a set of retro-reflective markers which are tracked by the MC system itself. Three segmentation methods and two different optimization approaches have been tested. The uncertainty propagation analysis, performed by means of a Monte Carlo simulation, shows that it is possible to calibrate the extrinsic parameters of a stereo-camera with a position accuracy of one millimeter and an orientation accuracy better than 1°.