A Real-Time Machine Vision System for Grading Quartz Mineral

Akkoyun F.

Journal of Testing and Evaluation, vol.50, no.5, 2022 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 50 Issue: 5
  • Publication Date: 2022
  • Doi Number: 10.1520/jte20210758
  • Journal Name: Journal of Testing and Evaluation
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Applied Science & Technology Source, Computer & Applied Sciences, INSPEC
  • Keywords: industrial stone grading, real-time detection, image processing, conveyor line, COMPUTER VISION, IMAGE, INSPECTION, PREDICTION, DESIGN
  • Karadeniz Technical University Affiliated: No


© 2022 by ASTM InternationalClassification is an indispensable process in industrial mass production applications when competitive marketing and increasing the product value by reducing time and saving costs are in concern. In this regard, the machine vision system (MVS) is a prominent technology, especially for automated industrial production flow lines. Recent studies consisted of increasing the accuracy of such systems using advanced technology and complex solutions. Nevertheless, for automated industrial production flow lines, considering only the accuracy rate of an MVS by ignoring the cost and processing speed is not a sufficient parameter to evaluate the success rate concerning the marketing capability. In this study, a relatively low-cost and automated MVS production line for grading white stones in real-time is demonstrated. A conveyor line and a rotary mechanism are integrated into the system for performing a stone grading task. A conditioned cabinet is used for inspecting the flow line continuously. Forty different stone samples in four groups are evaluated in the experimental stage to observe the flow speed and processing accuracy. Different flow speeds of the conveyor line are investigated. The results are demonstrated that the low-cost MVS is successfully operated for grading white stones at relatively high speed with a 92 % accuracy.