An Operational System for Sensor Modeling and Dem Generation of Satellite Pushbroom Sensor Images

This paper introduces an operational and extensible system for sensor modeling, triangulation, DEM extraction and stereo mapping of space borne pushbroom sensor images. The methods for sensor modeling and triangulation are introduced with the consideration of the newer sensors such as EROS A1, ALOS PRISM, FORMOSAT2 and THEOS satellite systems. They are based on the collinearity equation and bundle block adjustment. The orbital ephemeris data are rigorously considered. They are equally applicable for SPOT5, QuickBird and GeoEye images. In case pushbroom sensor images are provided with rational functional model without the specifications of the sensor, an additional correction triangulation will be performed using polynomial model. Test results with real data are presented and a comparison between these two models is carried out. Problems associated with automatic extraction of a DEM from high resolution satellite images are investigated and an adaptive feature based image matching method is introduced. The results from several datasets show the good quality and efficiency of the proposed method. Together with the introduced algorithms and techniques, and rigorous software engineering requirements and standard a software system is built for mapping with high resolution satellite images with easy extensibility for new similar sensors. Several examples show the superior performance and high quality of the system and developed technology. * Corresponding author.