DSM Generation from High Resolution Satellite Imagery Using Additional Information Contained in Existing DSM

Abstract

Information about the shape of the Earth's surface are required for several tasks like the creation of orthoimages or flood control. The generation of digital elevation models (DEM) with stereo images from space, or with interferometric synthetic aperture radar (InSAR) have been proven to be reliable and cost effective for larger regions, especially in remote areas and developing countries. Instead of a DEM, a digital surface model (DSM) is generated with remote sensing methods. Optical stereo image pairs are matched to get a large number of automatically located conjugate points. These points are mapped by forward intersection into the object space. The DSM is retrieved from these points by triangulation and interpolation. This method is mainly used to generate DSM from scratch. Thereby, the matching of optical images results often in areas without matching points, e.g. due to low contrast. This paper follows the approach to include the information of existing InSAR DSM into the point data set before or during the DSM generation from optical stereo images. Several methods are discussed and a SPOT image pair is analysed. Comparisons are shown for two inter¬po¬lations with and without additional input. Advantages of using the single data sets and of the combined data sets are discussed.