Tips For The Acceptance Of Classifying Image Processing Systems

Machine vision systems have become indispensable in industrial production. They work quickly and without contact in the system and complete their tasks in the production cycle. Quantitative ability considerations are already standard for measuring image processing systems. The measurement uncertainty is usually used as a parameter here. In contrast, there have been no corresponding established parameters for classifying image processing systems, the results of which are attributive variables. The new guideline VDI / VDE / VDMA 2632 sheet 3 closes this gap and considers parameters that describe the classification performance of an image processing system.

Avoid communication traps early on

When accepting a classifying image processing system, it depends on whether the system has the agreed performance and whether the assignment to the various error types is correct with the agreed security. Because of the diverse forms of classifying image processing systems, it is hardly possible to specify a system that is generally applicable to all tasks in order to assess performance. The new guideline therefore proposes procedures for evaluating the classification performance in the acceptance of image processing systems using typical examples from industrial testing technology. This should serve as a guideline for similar cases. The advantages and disadvantages of different acceptance concepts are discussed,such as acceptance with sample catalogs and acceptance with products from ongoing production. The guideline is aimed equally at users and providers of image processing systems and should help to avoid communication traps at an early stage.

Policy is available now

The VDI / VDE Society for Measurement and Automation Technology GMA and the Industrial Image Processing Department of the Association of German Machine and Plant Manufacturers VDMA have been cooperating since last year in the dissemination of guidelines for industrial image processing. By cooperating with the VDI, the VDMA wants to contribute to the fact that the VDI / VDE / VDMA 2632 series of guidelines is becoming known not only in the image processing community, but also among users. The guideline is intended to help avoid misunderstandings and to manage projects efficiently and successfully.

"The series of guidelines structures communication between providers and users - making image processing projects a success," says Anne Wendel, head of the VDMA industrial image processing department. "Employees of VDMA member companies are active on VDI and VDMA committees and I am pleased that everyone is pulling together on this important issue."

Editor of the guideline VDI / VDE / VDMA 2632 sheet 3 “Industrial image processing; Approval of Classifying Image Processing Systems”is the VDI / VDE Society for Measurement and Automation Technology. The guideline is now available as a draft in German / English at a price of 76.30 euros. There is an opportunity to help shape the policy through an opinion, which works through the electronic opposition portal. The opposition period ends on February 28, 2017.