Future directions in industrial machine vision: A case study of semiconductor manufacturing applications

Abstract

Machine vision systems are being increasingly used for sophisticated applications such as classification and process control. Vision systems have access to richer information (colour/texture/depth), and employ more powerful hardware than their predecessors. Though there is significant potential for the increased deployment of vision systems, a number of important problems have to be addressed to sustain growth in the area of industrial machine vision. This paper identifies some of these problems and the future research directions which show promise in solving these problems. Some of the directions examined include non-conventional imaging modes such as scanning electron microscopy and atomic force microscopy, visual defect classification, integration of colour, texture and depth information, and configurability of vision systems for changing field requirements.