Automatic aerial image matching
Jan 2009 - Dec 2010 (2 years)
The project is dedicated to automated matching of high-resolution aerial images. This problem arises in such areas as cartography, geodesy, Earth monitoring and others. Image matching is the first stage within the aerial image processing cycle and must be carried out robustly in order to produce qualitative orthophotomaps and precise digital surface and terrain models (DSM, DTM). The proposed algorithm employs the feature-based image matching approach. It takes a block of images as an input and returns a number of feature points, each being present in two or more images. Thus, the output of the algorithm can be used for computing image transformations and for image stitching.