In recent years, semi-global matching (SGM) approaches have proven to be among the most popular and successful algorithms in the fields of stereo vision and photogrammetry (Klette et al. 2011, Michael et al. 2013). To extract the height information from the aerial imagery, we used this matching approach, which utilizes intensity differences, mutual information (as the cost function) and an approximation of the global energy function that is being optimized path-wise (16 paths in this study) from all directions over the image. The cost function is significantly influenced by the use of penalty values, which were chosen based on performance tests and represent varying magnitudes of disparity changes. These variables have a strong impact on the matching performance and the robustness that is related to this processing step. The term 'semi-global arises from the combination of both global and local methods in a way that the complexity of the process is lowered and the quality of the matching is drastically improved. While the computation time for these global methods is often considerably higher, the overall performance increases compared to local matching algorithms. Further, pixel-wise calculated matching cost, contrary to the calculation along image paths, poses negative effects of insufficient correspondences related to low texture and ambiguity (Hirschmüller 2007).