Comparison of methods for geometric camera calibration

Kurzfassung

Methods for geometric calibration in close-range photogrammetry are established and well investigated. The most common one is based on larger test fields with well known pattern, which are observed from different directions during the calibration process. The parameters of a distortion model are calculated using bundle-block-adjustment. This methods works well for short focal lengths as used in close-range photogrammetry. It is essentially more problematic to use this method with large focal lengths, because they would require very large test fields and surrounding space. To overcome this problem, there is another common method for calibration used in remote sensing. It employs measurements using collimator and gimbal (pan-tilt-table). A third calibration method uses diffractive optical elements (DOE) to project holograms of well known patterns, which can be observed by a camera. It is of great interest, to compare these three calibration methods. In particular the accuracy of each method should be investigated. We calibrated one camera with the three methods mentioned above: Test field calibration as used in close-range photogrammetry, collimator-gimbal-calibration as used in remote sensing and a DOE based calibration. The test field data was evaluated using bundle-block-adjustment-algorithms implemented in Australis and is regarded as the reference. The two other calibration methods were evaluated using self implemented algorithms. All methods provide a set of distortion correction parameters as used by Australis, which is a reduced set of the parameters suggested by Duane C. Brown. The resulting parameters are very similar for all investigated methods. The three sets of distortion parameters are used as constant parameters in a bundle-block-adjustment-process. Only the exterior orientation is adjusted. The remaining RMS of the image coordinate residuals are taken as a measure of quality for each calibration method. This measure shows no significant differences across the different calibration methods. This means, that all methods can be used for accurate calibration. It can be concluded, that long focal length cameras can be geometrically calibrated with sufficient accuracy using collimator-gimbal-calibration or DOE based calibration and distortion models known from close-range photogrammetry.