Ground truth data provision for ExoMars PanCam 3D Vision
The stereo images of the PanCam (panoramic camera) instrument of the ExoMars 2018 Mission will be used to characterize the Rover's environment in 3D, select potential scientifically interesting targets for further investigation and help define the Rover's trajectory for its journey on the surface of the Red Planet. To test and verify the 3D vision processing chain currently under development by JR, as well as characterize its accuracy and robustness under various illumination and topographic conditions, it is essential to use objective data sets captured from well-known geometry under controlled conditions. The thesis will explore possibilities to generate high resolution reference data (e.g. digital elevation models, triangular meshes) from known sources (laser scans, photogrammetric processing chains) , obtain high-resolution synthetically rendered images with given orientation (position / viewing angles / intrinsic calibration parameters) therefrom, exemplarily test an existing 3D reconstruction processing chain therewith and compare the result with the original 3D data to verify the strategy. The results will be used further in the ExoMars PanCam 3D Vision development to optimize the processing chain in order to minimize the deviations between original 3D data set and the 3D data set produced from the rendered images.
C/C++, Python, knowledge in computer graphics, 3D vision and/or photogrammetry.