PERFORMANCE OF LARGE-FORMAT DIGITAL CAMERAS
- Institute of Photogrammetry and GeoInformation, Leibniz University Hannover, Germany
Keywords: Photogrammetry, analysis, camera, geometric, radiometric, quality
Abstract. Based on test blocks and standard production blocks Z/I Imaging DMCII-140, 230 and 250 as well as UltraCam Eagle images have been analyzed by bundle block adjustment with self calibration. By analysis of image coordinate residuals it is possible to check remaining systematic image errors and to update the set of additional parameters to the required combination. The Hannover program BLUH is using a basic set of 12 additional parameters which is a combination between geometric parameters and Fourier parameters in polar coordinates. In addition it is necessary to use a set of special additional parameters for the combination of the 9 UltraCam sub-CCDs. CCDs are often not flat enough, causing a bending of the edges. For the correct handling of such effects, special additional parameters are required to eliminate or at least reduce remaining systematic effects at image corners.
For all DMCII-blocks with 5cm, 7cm, 9cm and 20cm GSD the root mean square size of the systematic image errors with 0.05pixels is very small. Only the basic set of 12 additional parameters is required, the special parameters for the image corners did not improve the accuracy determined with independent check points. Against former results with UltraCam images the monolithic stitching of the panchromatic sub-images to the green image improved the image geometry, nevertheless for reaching the highest accuracy the full set of 52 additional parameters is required, leading to systematic image errors in the root mean square of approximately 0.2pixels. This seems to be small, but it is causing a model deformation up to more as 1.0 GSD in the height, while in the case of the DMCII-images the model deformation did not exceed 0.2 GSD in Z. The major reason for the UltraCam Eagle image deformation seems to be caused by corner effects of the green reference image. Such an effect can be avoided with a better calibration.
The DMCII and the UltraCam Eagle images were improved by the firm ware for edge enhancement, influencing also the effective image resolution, determined by edge analysis. No real loss of the effective against the nominal resolution can be seen. Nevertheless the UltraCam Eagle images are a little noisy, which may be caused by the edge enhancement and by the imaging in January.