Real-Time Aerial Mapping by Image Features Extraction and Matching

We investigate and validate a way to perform a real-time aerial 2D mapping, performed by Micro Aerial Vehicle and calculated on-board. The process consists of image features extraction, matching, warping and merging them into expanding the map. The limitations of available hardware, selecting proper algorithms and optimisation are discussed as a part of the development of a robust workflow that may be used in various aerial missions.