AN INCOHERENT HOLOGRAPHIC CORRELATOR (FOR REALTIME CLASSIFICATION OF STOCHASTIC IMAGES)

AN INCOHERENT OPTICAL CORRELATOR, USING A COLOR TV MONITOR AND HOLOGRAPHIC SPECTRAL-SPATIAL FILTERS, WAS RECENTLY USED TO DEMONSTRATE REAL TIME COLOR PATTERN RECOGNITION. THE SPECTRAL-SPATIAL FILTERS WERE DESIGNED USING A LEAST-SQUARES LINEAR MAPPING TECHNIQUE TO COMPENSATE FOR THE SMEARING EFFECTS CAUSED BY THE FINITE SPECTRAL BANDWIDTH OF THE LIGHT OUTPUT OF THE CRT. THE EXPERIMENT DEMONSTRATED THAT THESE SPECTRAL-SPATIAL FILTERS PROVIDE BETTER RECOGNITION RELIABILITY THAN SPATIAL FILTERS ALONE. THE OVERALL PURPOSE OF THE PROPOSED PROGRAM IS TO PROVIDE A SIMPLE, ECONOMICAL, INCOHERENT PROCESSOR SYSTEM FOR REAL TIME IMAGE IDENTIFICATION AND CLASSIFICATION. THIS TYPE OF SYSTEM WILL BE APPLICABLE FOR USE IN RECOGNIZING COLOR IMAGES, PSEUDO-COLOR INFRARED IMAGES AND COLOR CODED IMAGES. IT WILL BE DESIGNED IN A COMPACT FORM SUITABLE FOR SUBSEQUENT MANUFACTURE AND WILL UTILIZE COMMERCIALLY AVAILABLE COMPONENTS. THE MAJOR PHASE I TASKS WILL BE: THE DEVELOPMENT OF AN IMAGE DATA BASE; THE DESIGN OF THE SPECTRAL-SPATIAL FILTER FOR PROCESSING WITH INCOHERENT LIGHT; THE ANALYSIS OF THE INFORMATION PROCESSING CAPABILITIES OF THE FILTER; AND THE DESIGN AND PERFORMANCE ANALYSIS OF THE COMPLETE IDENTIFICATION SYSTEM. DURING PHASE II, THE SPECTRAL-SPATIAL FILTER WILL BE BUILT USING COMPUTER GENERATED ELECTRON-BEAM HOLOGRAMS, THE SYSTEM WILL BE TESTED AND DEMONSTRATED AND ITS PERFORMANCE ANALYZED AND COMPARED TO PREDICTED PERFORMANCE.