In the approach proposed in this thesis, computational power will be shifted to an offline training phase to achieve high-speed during live operation. In contrast to visual SLAM frameworks, this project considers scenarios in which the environment is known beforehand or can be even actively modified. The environment is non-dynamic and planar, therefore, the UAV will make use of texture on the bottom or ceiling of the environment. This opens the door for improving the proposed algorithm by changing the map. On the basis of desired characteristics of a given map, an evaluation technique was developed that determines the suitability of an environment for the proposed approach. This technique allows for spotting distant regions with similar image features, which could lead to deteriorated performance. The evaluation can be performed using a given map image or recorded images during flight. In the former case, synthetic images will be generated from the map image that simulate images taken during flight.