Embossed OCR Code Reading
Read low-contrast direct part marked characters on cast or machined parts
Graphical programming environment for deep learning-based industrial image analysis
Part traceability along the entire manufacturing value chain has become increasingly important for safety, compliance, and counterfeit detection. Many cast or machined metal parts in machinery, automobiles, and other products are traceable by raised characters that are either cast or embossed on the part’s surface. Such direct part marked (DPM) characters are rugged and can survive for the length of the product’s useful life.
DPM characters are made of the same material as the part, which means that the contrast between character and substrate is low. The characters are raised from the surface, so shadows and reflections can change significantly with small variations in lighting angle and intensity. A slight difference in composition between one lot and another can alter the appearance of the resulting part and its DPM characters.
Because of this high variability and low contrast, standard machine vision has a great deal of difficulty reading embossed or cast characters consistently, particularly when lighting or material composition changes.
AI-based technology is an ideal solution to reliably read embossed DPM OCR codes. The AI-based OCR tool trains on a small sample set of images of the part and its embossed characters. Thanks to a pretrained font library, the tool is easily set up and deployed to read the direct part marked OCR codes consistently and accurately for complete traceability.
If a change in supplier results in a different color or texture of part, quick retraining with a new set of images ensures that the OCR tool continues to identify the raised lettering on the part.