Abstract :
Summary form only given, as follows. Four types of industrial vision systems are now being used in Japanese industry; these are vision systems based on the pattern matching method, pixel-counting method, feature-parameter extraction method, and slit-light method. These methods utilize, implicitly or explicitly, knowledge of objects and their environments, as well as knowledge of the processing schema. These methods are being enhanced with new technologies, which include an automatic selection algorithm of template patterns for effective pattern matching, a new position-finding algorithm based on structural matching, and a one-pass feature-measurement algorithm for real-time recognition. These recent topics are introduced. In addition, a design-pattern referring method for semiconductor wafer inspection is introduced, where a knowledge-directed approach is effectively applied. Some results on SEM image analysis for a future wafer inspection machine are also discussed
Keywords :
computer vision; computerised pattern recognition; computerised picture processing; inspection; real-time systems; scanning electron microscopy; Japanese industry; SEM image analysis; automatic selection algorithm; design-pattern referring method; feature-parameter extraction method; industrial vision systems; machine vision research; one-pass feature-measurement algorithm; pattern matching method; pixel-counting method; position-finding algorithm; real-time recognition; semiconductor wafer inspection; slit-light method; structural matching; template patterns; Design methodology; Feature extraction; Image analysis; Inspection; Laboratories; Machine vision; Parameter extraction; Pattern matching; Pattern recognition;