Automatic Seed Point Generation For Stereo Matching And Multi-image Registration

Three interest-operator-based methods are presented for automatic generation of seed points for SPOT satellite imagery, aerial photographs digitized on a kern DSR-11 analytical plotter, and any other imagery with stereo-overlap. Results for the three methods and a comparison of interest operators are also presented. In all cases and in virtually every example the results of the working versions of the seed-generation programs were promising. It is shown that interest-operator-based automatic seed-point generation for virtually any form of imagery is not only feasible, but can also be achievable with some degree of reliability.