Glossary · Machine vision
Optical character recognition (OCR)
Also known as: OCR, Text recognition
German: Optische Zeichenerkennung (OCR)
In machine vision and document processing, optical character recognition (OCR) is the automatic conversion of printed, marked or handwritten characters in an image into machine-readable text. In production it reads batch codes, dates and serial numbers on products and labels.
- Machine vision
- AI
In one sentence
Optical character recognition (OCR) turns characters in images into machine-readable text, for example batch codes and dates on products.
Example
OCR reads the best-before date printed on each carton and compares it with the date set for the production order.
How it applies
- Engineering: Industrial OCR must cope with inkjet dots, laser marks, curved surfaces and poor contrast. Modern systems use trained neural networks; fonts and character sets are defined for the application.
- Operation: OCR is often combined with a check against expected content; misreads between similar characters (0 and O, 8 and B) are handled by restricting the character set per position.
- Documentation: The documentation team should document the expected text format, character set, validation results and the reject handling. In documentation workflows, OCR also converts scanned legacy manuals into text, where errors must be reviewed before reuse.
OCR vs. OCV
OCR reads unknown text. Optical character verification (OCV) checks that known text is present and legible. Verification is more robust when the expected content is known.