Machine vision uses cameras, lighting, and computer processors in an industrial environment to automatically extract information from digital images of products and analyzes them against defined criteria. If a product is outside of tolerances, it is deemed a reject and then removed from the production line.
Compared to academic computer vision systems, industrial machine vision integration demands greater robustness and must be designed with fault tolerance in mind. Industrial vision systems must contend with environments with vibration, temperature variability from 10 C - 40 C (50 F - 100 F), dust, and occasionally EM noise.
Machine vision systems are used to measure, find defects, verify components, read/verify print quality, guide a robot, and much more. By catching defects early in the production, process waste is reduced. Bad parts can be identified and eliminated before they are built into larger assemblies, reducing the need to rework the product.
Jams and production stoppage can be minimized by having a vision identify and reject misaligned parts before they are fed into a machine.
Systems can also be designed to inspect diﬀerent parts or conﬁgurations with the same setup. Using an HMI display, an operator can select a product from a list, and its inspection parameters will be pushed automatically, thus significantly reducing process setup time. Before starting a new batch, challenge parts can be presented to the system to validate performance.
Traceability requires extending a machine vision system with data collection capabilities. Parts are tracked by grabbing a unique identifier via reading 1D and 2D barcodes, performing OCR to read the text on a label, or reading an RFID tag. Test data can then be associated with this unique identifier and later stored in a database. Such data could be vision inspection data, results from a leak tester, thermal tester, or electrical tester. In the event of a failure (either during ﬁnal testing or out in the ﬁeld), test results can easily be recalled from a database and used as part of root cause analysis. In the event of a recall, track and trace initiatives can narrow down the scope and financial impact.