ARCHITECTURE FOR OBJECT IDENTIFICATION - INCORPORATING AN OPTICAL CORRELATOR AND DIGITAL PROCESSING FOR DISPLAY AND RECORDING OF OPTICAL-DATA

An architecture is described incorporating the advantages of both digital and optical image processing, This hybrid system for object identification builds on the parallel and instantaneous processing of the optical correlator technology and digital techniques to display and record optical images and filters. A hierarchical computational logic leads to recognition and object identification through three levels of increasing computational complexity. The shape analysis leads to classification by the geometrical parameters, including scale, shape, and rotation for each blob. Filters are stored and indexed by their geometrical descriptors. The optical image processor recognizes objects by performing a correlation operation on computer-stored filters, generated with a priori knowledge.