Optical quality inspection

KI-based optical quality inspection

Within the scope of quality assurance, manufacturing companies are increasingly required to use automated quality controls during production. Automated quality controls based on optical sensors and industrial image evaluation have been successfully used in industry for years. Today, however, classical image processing methods are increasingly reaching their limits when a high number of variants must be handled or the appearance of the workpieces and their defects are not always constant (e.g. food). In these cases, the use of classic image processing methods involves high programming efforts, especially when new object types have to be added to the inspection.

One possible solution for this is image processing methods based on artificial intelligence (AI), which, in contrast to classic image processing, do not require any time-consuming re-programming for new object types. The expert knowledge in the field of image processing is reduced to a minimum on the user side, which makes the technology easy to use, especially for medium-sized companies.

In various projects with industrial partners, Fraunhofer IOSB-INA has proven that the technology enables a very precise and reliable quality inspection of a wide range of object types (natural objects, e.g. potatoes, plastic injection molded parts for the automotive industry, and metallic turned and milled parts). In projects for industry, all necessary aspects are holistically considered during the introduction of the technology, which includes lighting technology, sensor technology, image evaluation as well as the mechanical, electrical and information technology integration into the respective production environment.