Comparison of Calibration Algorithms for a Low-Resolution, Wide Angle, 3D Camera

This Master thesis deals with the calibration of the optics of a novel time-of-flight range camera intended for use in automotive applications. The camera is different from standard cameras in that it can yield both regular intensity images and depth maps, but also in that its resolution is less than a tenth of a standard camera. The thesis treats the topics of feature detection in low resolution images, correction of radial distortion caused by wide angle lenses, and the robust and exact estimation of the camera’s geometry (internal calibration) as well as its position in relation to a scene (external calibration). A novel feature detector is proposed for a robust extraction of circular control points. The work compares different techniques for calibration and lens distortion correction. Various calibration algorithms learned from literature were implemented and compared to publicly available toolboxes. Assuming that the range measurements of the special camera used for this project are correct, then by calibrating it according to this work, the scene at a distance of one meter can be reconstructed up to only 1 cm inaccuracy. ii