Dmc Geometric Performance Analysis

Since the apparition of the first large format digital aerial cameras there are great expectations on their performance. The dream to get aerial images virtually free of geometric errors and with a greater radiometric quality is nearly going to be fulfilled. Nevertheless, some reports on systematic image residuals (Honkavaara et al., 2006a and 2006b, Alamús et al., 2005 and 2006, Cramer, 2007), unexpected height errors in aerotriangulation and the need for additional self-calibration parameters (Alamús et al., 2005, Cramer, 2007) have been reported since 2005. In this paper a preliminary analysis on the theoretical accuracies of the ZEISS/INTERGRAPH (Z/I) Digital Metric Camera (DMC) and analogue camera in aerotriangulation is carried out. This analysis considers a mathematical model where the image has conical geometry and it is free of systematic errors. It is studied the influence of the base-to-height ratio, the image pointing precision (manual and automatic), as well as, tie point density and distribution (classical Von Gruber distribution or high, dense and uniform distribution) and GPS observations for projection centre. The expected accuracy in the case of the aerotriangulation of analogue images using “current” aerotriangulation set up (the a priori accuracy for image pointing, ground control measurement and GPS and tie point distribution) is computed. Then, a minimum aerotriangulation set up for the DMC camera is derived in order to warranty the same, or even better, theoretical accuracy level that is obtained in aerotriangulation with analogue images. ICC experiences have proven that the expected theoretical accuracy in aerotriangulation is sometimes hard to obtain without considering an appropriate self-calibration parameter set in the bundle block adjustment. As it is well known that additional selfcalibration parameters absorb error propagation effects, which can be caused by non modelled systematic error sources. Some authors (Alamús et al., 2006 and E. Honkavaara, et al. 2006b) have detected systematic residuals in the order of one tenth of a pixel rms (root mean squared) in DMC image space. Investigations on the systematic error characterization, distribution in image space and stability over time and flying height are carried out. Finally conclusions are drawn from the investigations.