A feature-based image matching technique is presented for matching two images containing natural terrain and/or man-made structures.
Features of the line segments extracted from images taken-possibly-by different sensors, in the different environments, etc, are used as the primitives of the match process.
The R.S.T. transformation parameters are derived by the technique of clustering in the parameter space, and the clustering by pairing possibly pairs of the primitives, is constrained by the combinations of mathematical, synthetic and heuristic techniques.
Results of the simulation are also provided as an example applied to the landmark recognition for a navigational guidance, with the actual down-looking aerial image and the synthetic images.
