Generation of Orthoimage from High-Resolution DEM and High-Resolution Image


Department of Geomatics Engineering,University of Tehran


Generating an orthoimage from high-resolution satellite images is an important undertaking for
various remote sensing and photogrammetric applications. In this paper, a method is proposed
that uses Arti cial Neural Networks (ANN) to generate orthoimage Ikonos Geo images. For
orthoimage generation, a Digital Elevation Model (DEM) with a cell size of 4 m and RMS error
of 0.24 m is constructed with neural networks, based on a Quad Tree (QT) structure. In order
to determine object-to-image relationships, rational function models, polynomials and neural
networks with back propagation learning algorithms were used. Ground Control Points (GCPs)
and check points were taken from topographic maps of 1:2000, with a contour interval of 2.5 m,
to evaluate the accuracy of DEM and object-to-image transformations. The method described
in this paper is tested with an Ikonos Geo image from a region of Bilesavar, Iran.