Mirror Edge Inspection Using AI-Based Vision System

By February 15, 2022March 21st, 2022Industry Use Cases
mirror quality check

CLIENT BACKGROUND

Our client is an ISO 9001-2008 company certified by TUV Nord, engaged in offering innovative automation solutions in the engineering field for the last 8 years. They have experienced amazing business growth exceeding more than 70% year on year, through its innovative & cost-effective salutations in various products and focus on complete customer satisfaction with competitive cost & excellent service.

PROBLEM

  • Inaccuracy in defect identification on the edge surface of the glasses (after glass cutting)
  • The danger of getting a cut in manual inspection
  • Speed of the production is high and the minimum defect size is 0.5mm causing many glasses to be uninspected

PROBLEM IMPLICATIONS

  • Irregularities in the edge surface of the mirror do not let it sit on the frame properly
  • The quality standards do not meet and affect the customer experience badly

CLIENT REQUIREMENTS

  • To automate the inspection of the mirrors and automatically reject the defective ones with the help of machine vision technology
  • To achieve higher accuracy for defect identification in order to reduce or eliminate the false acceptance rate
  • To identify the defects like chip-offs, scratches, etc

CURRENT PROCESS

The process is being done manually which is a very time-consuming process requiring a lot of labor and there is a significant scope of error as well.

Manual check mirror/glass quality

The mirror glass is guided to the inspection station on a conveyor, where there will be a camera placed above the mirror glass with proper illumination to avoid reflection in the captured images.

  • AI-based segmentation technique is used to identify the surface of the glass for defects like chips, scratches, etc.
Good Mirror Quality
Glass edge defects - rough edges

CONCLUSION

With the help of an AI-based machine vision system, the following potential benefits could be observed –

  • Can achieve an accuracy of close to 99 percent to identify defects with missing screws.

Leave a Reply

Schedule A Demo
close slider