Surface Mapping Using Image Triplets: Case Studies and Benefi t Assessment in Comparison to Stereo Image Processing

Mapping of the Earth’s surface from satellite images is a continuing application in remote sensing, which has been distinctly pushed with the increasing availability of very high-resolution image data. In most of the missions, stereo images can be acquired making 3D mapping from space very attractive, as high accuracy and level of detail become feasible from such data. In this paper, several case studies are presented, which have been applied to stereo pairs acquired by the SPOT sensors as well as the very highresolution Ikonos sensor. As an extension to standard stereo mapping, a 3D mapping approach which makes use of multiple image coverage is presented. For this work, image triplets including multi-sensor data sets have been used to extract 3D topography information and to generate digital surface models. The benefit of this extended approach is comparatively assessed with respect to results achieved from standard stereo mapping. Introduction Driven by the image acquisition capabilities of many of the past and present satellite missions, 3D mapping of the Earth’s surface using stereo image data is an ongoing subject of research and applications in remote sensing. The availability of digital stereo data was particularly pushed in the mid-eighties by the first SPOT mission and has meanwhile reached a high level, as stereo coverage can be acquired from space at very high-resolution in the meter and sub-meter range, as provided at present by the Ikonos and the Quickbird sensors. In contrast to the former SPOT stereo data collection from repetitive over-flights and at oblique viewing directions (cross-track stereo), optimized acquisition scenarios are meanwhile used where the data are acquired during a single over-flight at almost the same time through forward and backward viewing with one or two instruments (in-track stereo). At the Institute of Digital Image Processing, stereo mapping using satellite image pairs has a long tradition. First results were generated from analogue optical SPOT, as well as SAR image pairs through mapping on an analytical plotter (Raggam et al., 1989). Based on this work, the algorithmic baseline to extract 3D information from multi-sensor stereo data was established with respect to digital stereo data utilization (Raggam et al., 1990). Further on, the feasibility to combine image pairs acquired from different instruments was monitored in order to extract 3D information in case that Surface Mapping Using Image Triplets: Case Studies and Benefit Assessment in Comparison to Stereo Image Processing