Photogrammetric surface reconstruction from oblique and nadir looking aerial images in the extreme environment of the Himalayan area "Sabche Cirque"

Kurzfassung

Remote sensing in areas with extreme altitude is particularly challenging in respect to conduct field campaigns as well as to obtain a desired high quality output. In high mountain areas specifically, steep slopes result in reduced ground pixel resolution and degraded quality in the Digital Terrain Model (DTM). Exceptionally high brightness differences can in part no longer be imaged by the radiometry of ordinary sensors. Very detailed and accurate 3D maps have many application areas. They provide a basic tool for the analysis of natural hazards and the monitoring of glacier surfaces in high mountain areas. In this regard DLR Institute of Optical Sensor Systems has developed a new camera, the Modular Aerial Camera System MACS-Himalaya. The camera was designed specifically for the needs of high mountain areas using High Dynamic Range (HDR) for high radiometric resolution and oblique camera geometries adjusted to the needs of steep terrain. It can be seen that with HDR mode the images are clearly visible in high contrast difference areas. In this thesis the image data from the Sabche Cirque in the Annapurna range in Nepal obtained with MACS Himalaya camera was processed using new photogrammetric interpretation methods with the software Pix4D mapper which is based upon the structure-from-motion (SfM) algorithm. With the development of computer systems , software and research in computer vision, photogrammetry has now extended from traditional to SfM which is a highly automated process. The software detects the matching points between the images to generate point clouds and different GIS products like DTM, orthomaps etc. The fundamental of photogrammetry and description of software pix4Dmapper has been discussed here. The result from the software after initial processing gives the quality and reliability of the bundle adjustment and aerial triangulation. The Automatic Tie Points (ATP) formed in the image are matched to the same image features of di erent overlapping images to form sparse point cloud. This sparse point cloud is further extended to form a dense point cloud and different GIS products . This result has been analyzed as the quality of other products are dependent upon the quality of sparse point cloud. There are different GIS products among them orthomosaic and DTM/DSM are of our main interest. The impact of high brightness differences, steep slope, homogeneous surface on the image has been analyzed from the orthomosaic. The result shows a visually accurate orthomap. The use of these map on GIS and Geology field has been discussed. The quality of DTM model is the prerequisite for the reliability of all the products. The reliability of DTM is evaluated by comparing it with available DTM and reference coordinates. Our DTM was compared with existing available DTM from SRTM and ALOS mission, which shows that there is no large deformation in our DTM. However there are errors in our DTM at the edge of project area due to small number of overlapping images. The low elevation difference in low slope areas are also shown. The coordinate from ICESat mission was used to compare the difference in height . The analysis shows the absolute accuracy at the edges of our DTM to be about 3m.