Fast automatic registration for satellite imagery by using heuristic planning

A fast algorithm for the automatic registration of remotely sensed images obtained from satellites has been developed. The automatic section is based on an orbital model and a library of ground control points. The real problem of an excessive searching requirement for similarity measure is tackled by using heuristic planning. This method reduces the search space to a less detailed compact size, which is later used as a plan for finding the exact coordinates and which confines the search window to a small region. The improved procedure has achieved subpixel accuracy with a substantially reduced registration time.