November 2, 2017 - In the acceptance procedure for a classifying machine-vision system it is a matter of whether the system has the agreed performance and whether assignment to the different error types is in line with the certainty level agreed. The wide variety of types of classifying machine-vision systems makes it almost impossible in assessing performance to provide a systematic approach universally valid for all tasks.
Standard VDI/VDE/VDMA 2632 Part 3 therefore puts forward some procedures for assessing classificatory performance during the acceptance of machine-vision systems, doing so with the aid of typical examples from industrial inspection technology which can serve as guidance in similar cases. The various acceptance concepts, such as acceptance on the basis of sample catalogs and acceptance by reference to products from current production are also discussed, together with their respective advantages and disadvantages. The standard is intended for users and suppliers alike, and should help at an early stage in avoiding communication pitfalls.
The publisher of standard VDI/VDE/VDMA 2632 Part 3 'Machine vision/industrial image processing - Acceptance test of classifying machine vision systems' is the VDI/VDE Society Measurement and Automatic Control. The standard was published as a 'white copy' in October 2017 and replaces the draft version of September 2016.
