CARTOSAT-1 Stereo Imagery: Potentialities about Orientation, DSM Extraction and Orthorectification

Abstract

CARTOSAT-1 satellite was launched in May 2005; the optical sensor is configured with two pushbroom cameras, namely Aft and Fore, tilted in along track direction by -5° and +26°, providing stereoscopic imagery in the same pass with a ground resolution of 2.5 m. Within the ISRO-ISPRS Cartosat Scientific Assessment Program (C-SAP), investigations were carried out on a stereopair belonging to the test field called TS-6, acquired on 8th June 2005 over the city of Rome, Italy (30 x 7.5 km2), in order to test the sensor potentialities for DSM generation. The stereo images were processed with PCI-OrthoEngine software, using a generic rigorous sensor model; the image orientation and the DSM geocoding were based on GCPs collected by GPS surveys with a mean 3D accuracy of 5-10 cm. For the evaluation of the DSM accuracy, it was compared with the DSM generated from a block of 6 aerial photos (scale 1:9’000) over an area of about 3.5 x 4 km2. An orientation with horizontal and vertical accuracy at sub-pixel level may be obtained with a “generic rigorous model”; as regards DSM extraction, an accuracy ranging from 1 to 2 pixel, depending on the feature of the land cover (urban zone, bare soil) is achievable with commercial image matchers.