A LOW COST INDUSTRIAL REALTIME IMAGE PROCESSING SYSTEM FOR MEASUREMENT AND PART RECOGNITION

This paper describes a low cost visual sensor system which can be easily integrated into the production and assembly sequence to solve various tasks in part-handling and quality assurance. The hardware conception of the PC-based image processing board, which interfaces to line scan cameras, is shown. The system is equiped with a digital signal processor (DSP) to meet the "real-time" requirements of industrial applications (e. g. 100% inline automated visual inspection of parts). New algorithms have been developed to analyze the line data stream in realtime, without storing a complete frame, as is usually done in image processing. The performance of the system is demonstrated on a part recognition/position determination task.