Methods for quality control of large Cartosat-1 stereo blocks

Kurzfassung

In the field of satellite remote sensing, the geometric accuracy of satellite imagery is strongly influenced by the quality of the satellite orientation. For example, rational polynomial coefficients (RPCs) are provided by the Indian agency as a universal sensor model for each scene of the Cartosat-1 stereo pairs. However, these RPCs derived from orbit and altitude information have the accuracy far worse than the pixel size of 2.5m (normally in the order of a few hundred meters). Hence, a new estimation of orientation parameters is necessary. This process requires well-distributed ground control points (GCPs) with sub-pixel accuracy. After adding accurate GCPs, improved orientation parameters can be obtained. By using GCPs, some institutes and universities such as ISRO, ETH Zurich, Erdas and University Hannover have made some important achievements for improving the satellite orientation. However, in many application fields, such as continent wide reconstruction or crisis support applications, acquiring the required GCPs is a time consuming task or might even be impossible, if a fast response is required. In this diploma thesis, a new method of block adjustment, which helps to correct the orientation parameters, will be introduced and evaluated. The data resource is a block of 428 Cartosat-1 stereo pairs (region in north Italy). In this thesis, a special attention will be put on the adjustment, which bases on the primary reference material which is digital surface model (DSM). This approach can be completely automated and does not require manual GCP collection. Another focal point is the independent quality control for the adjustment. Accurate check points derived from aerial images are chosen to realize the accuracy analysis. Finally, the results in this thesis are also compared with results of other research institute such as ISRO and ETH Zurich.