THIS PAPER DESCRIBES ALGORITHMS DEVELOPED FOR SEGMENTING MULTI-BAND GREYSCALE IMAGES INTO N-CLASS LABELED IMAGES IN REALTIME. THEY ARE BASED ON THE STATISTICAL PATTERN RECOGNITION PARADIGM AND LOCAL OPERATORS. ALGORITHMS AND THE ASSOCIATED HARDWARE ARCHITECTURES ARE BEING DEVELOPED FOR AN APPLICATION IN MICROELECTRONICS INSPECTION. THIS COMBINATION REPRESENTS A SIGNIFICANT ADVANCE IN MACHINE VISION TECHNOLOGY THAT ENABLES THE VISION SYSTEM DEVELOPER TO ADDRESS PROBLEMS THAT WOULD HAVE PREVIOUSLY BEEN INFEASIBLE.
