Hybrid pattern recognition by features extracted from object patterns and Fraunhofer diffraction patterns: development of a more useful method

A shift-, rotation-, and scale-invariant pattern recognition method is proposed. In this method, an object image is observed by a movable sensing head, and then simple features, which are invariant under shift, rotation, and change of scale of the object image, are extracted by digital image processing. At the same time, the Fourier spectrum of the edge-enhanced object image is obtained by coherent optical image processing, and the shift-, rotation-, and scale-invariant features are derived from the spectrum. Classification of the object is carried out by a small-scale digital processor using both features obtained by digital and analog image processing.