Helsinki University of Technology
Department of Surveying
The subject of this thesis is photogrammetric mapping from Very High Resolution (VHR) satellite imagery. The study presents a concept of geometric properties of VHR satellite imagery and describes an approach to mapping from IKONOS stereo imagery in a digital stereo workstation. Accuracy potential of point measurements in IKONOS stereo imagery is explored and some potential applications for exploitation of the imagery are estimated.
The first commercial one-meter resolution satellite IKONOS as launched in the end of 1999. More than ten VHR satellites will be orbiting before 2006. This is a challenge for the photogrammetric user group. Geometric reconstruction, processing methods and operability of VHR satellite imagery from linear arrays in pushbroom mode differs from present satellite and aerial imaging systems. Also there are no comprehensive studies made yet related to geometric accuracy and suitability of IKONOS stereo images for topographic mapping.
IKONOS stereomodel is delivered to customer as a ready prepared product. The stereomodel product contains preprocessed epipolar images and rational function coefficients by which user can carry out the object-to-image transformation. The producer of IKONOS stereo product calculates the coefficients by using only ephemerids without ground control. Therefore, stereomodel differs from "ground truth". An absolute orientation method which uses 7-parameter similarity transformation was developed to match the stereomodel into user's ground coordinate system by exploiting the GCPs. In this study, geometric accuracy test for point measurement in stereomodel was made by using very accurate large scale maps as a reference. The accuracies of point measurement without absolute orientation of stereomodel were 3 m and 7 m in plane and height coordinates, respectively. Accuracies after absolute orientation were 1,0 m and 0,6 m in plane and height coordinates, respectively.
Keywords: VHR satellite images, digital stereo workstation, photogrammetric mapping
Language: Finnish, English abstract