Digital elevation modelling by radargrammetry in real-beam ground mapping mode

Abstract: In this thesis the problem of estimating terrain elevation using two-dimensional radar data from the multirole ghter aircraft JAS 39 Gripen is considered. Radar data contains information about range from the aircraft to the reflecting terrain, as well as horizontal angle. In general, radar data has high resolution in range and low resolution in angle, giving rise to interesting problems. A new radar with higher resolution is in development for the next-generation Gripen. This thesis aims at answering whether estimation of terrain height can be done using available radar data, in order to evaluate the plausibility of doing this with data from the new radar. The approach is to find matching terrain features in subsequent images of the ground, and use this information to calculate terrain elevation. Two approaches are implemented and studied, both on simulated radar data and on real datasets. One approach uses Harris corner detection and the other uses Speeded-Up Robust Features (SURF). Conclusions drawn are that the algorithms do not work for the available radar data, but that they possibly could work when higher resolution data from the new radar is available.