Area-based similarity measures residue analysis for registration of remotely sensed aerial imagery for homeland-security wide area surveillance

ISR imagery for wide area surveillance, in general, is acquired by an airborne platform flying at an altitude of thousands of feet. At these altitudes, scaling and perspective are negligible to influence the registration process. Unfortunately, at these altitudes there may be a lack of discriminating dominant structures for the determination of the related corresponding discrete points, to facilitate the conventional feature extraction based registration. However, an area-based approach can sometimes be more viable at these relatively high altitudes, as it looks at the collective connected features/structures in a given reference and template. In particular, for a given application, various orientations, for determining registration parameters, may look very similar but may have small differences that may affect image registration and the related computer based techniques for threat detection. Therefore residue analysis, not based on landmarks or discrete structures, as is the case in classical registration methods, but on a set of connected structures or landmarks in a given area/template can provide a measure of effectiveness of a similarity measure for area-based techniques, which is the focus of this paper.