Computer Science · Computer Vision and Pattern Recognition
Reliable automatic camera-laser calibration
Abdallah Kassir, Thierry Peynot
See on the mapAbstract
Camera-laser calibration is necessary for many robotics and computer vision applications. However, existing calibration toolboxes still require laborious effort from the operator in order to achieve reliable and accurate results. This paper proposes algorithms that augment two existing trustful calibration methods with an automatic extraction of the calibration object from the sensor data. The result is a complete procedure that allows for automatic camera-laser calibration. The first stage of the procedure is automatic camera calibration which is useful in its own right for many applications. The chessboard extraction algorithm it provides is shown to outperform openly available techniques. The second stage completes the procedure by providing automatic camera-laser calibration. The procedure has been verified by extensive experimental tests with the proposed algorithms providing a major reduction in time required from an operator in comparison to manual methods.