Image Pattern Recognition in Industrial Inspection

摘要

Most manufacturing processes require a visual inspection of some aspect of the process. In the most straight forward applications, a vision system is used to inspect for part completeness, or to check for flaws in the manufacturing product. In more sophisticated use of vision systems the inspection task may be secondary to tasks such as part location, identification, or determining part orientation. In general these tasks are important when the vision system is used in conjunction with a robot manipulator. In robot systems, vision is needed to allow the robot to acquire, manipulate, and inspect parts without the need for elaborate fixturing, or complex part delivery systems. The labor cost of manual inspection, and the high cost of special purpose part delivery systems for robots have led many manufacturers to investigate vision systems for use in manufacturing. Digital image processing and pattern recognition are providing the basis for a growing number of attempts to achieve an automated vision system. This paper begins with a brief historical perspective on image processing and pattern recognition. Next a series of state of the art examples of visual inspection system, and then robot vision systems is presented. The paper contains a list of other areas in manufacturing for the application of vision systems, and concludes with an assessment of the future of vision systems in manufacturing.