Industrial OCR and plain-text recognition
Batch numbers, serial numbers, best-before dates and other printed or directly marked text have to be read and verified automatically in many production processes.
Industrial OCR systems recognize characters straight from a camera image and make the result available to the controller, a database or a traceability system.
Discuss your applicationTypical OCR applications
- reading batch numbers
- capturing serial numbers
- checking best-before dates
- reading production dates
- capturing article and type designations
- reading laser markings
- checking label printing
- comparing character strings with database values
- using OCR together with DataMatrix or barcode reading
OCR or OCV?
The two terms are often mixed up.
- OCR - optical character recognition
- The system reads a character string and outputs the recognized text. Example: an arbitrary serial number is to be captured.
- OCV - optical character verification
- An expected text is already known and the system verifies that it has been applied correctly. Example: checking whether “LOT 260827” has actually been printed on a product.
Difficult typefaces
In industry, text is rarely as clean as it is on a screen. Problems arise from:
- varying contrast
- weak printing
- laser engraving
- reflective metal surfaces
- curved packaging
- dot matrix type
- different typefaces
- contaminated surfaces
- partially damaged characters
AI and deep-learning based OCR can be an advantage for such tasks. Even so: a good image beats a complicated algorithm. The right lighting and optics should therefore be considered before the software is chosen.
Combining OCR and traceability
The combination of DataMatrix code and plain text, or of barcode and batch number, is particularly useful.
Information can then be read by machine while the visual markings are verified at the same time.
This information helps us specify the system
Example values - real images of good and bad parts are the most informative thing you can send.
- Character height
- 2.5 mm
- Number of characters
- 12 characters, one line
- Marking method
- laser on stainless steel
- Field of view
- 60 × 20 mm
- Cycle time
- 400 ms
Suitable product groups
Want to know whether your marking can be read automatically?
Send us a few typical images of good and bad parts, or sample parts themselves.
Have your OCR application reviewedFrequently asked questions
What is the difference between OCR and OCV?
OCR reads an unknown character string and outputs it. OCV checks whether an already known, expected text has been applied correctly.
Does OCR work on shiny metal?
Yes, but the lighting is decisive here. Reflections can be reduced considerably by polarization, dome or dark field lighting.
Can OCR run together with code reading?
Yes. Many systems can evaluate both plain text and DataMatrix or barcodes in the same image and compare the results.
Related applications
- Traceability with fixed-mount code readersRead 1D, 2D and DPM codes automatically in the process and identify workpieces across production and packaging steps.
- Industrial quality controlCombine several inspection features - from completeness and dimensional accuracy to surface, code and plain text.
- Barcode reading with handheld readersMobile capture of 1D, 2D and DPM codes in production, goods receipt, order picking and dispatch.