-
Process Systems
-
Machine Vision
Frequently Asked Questions
- OCR reads what the characters are.
- OCV verifies the characters match expected content and meet print-quality expectations.
OCV is often used when quality systems require confidence that codes are not only readable, but correct and well formed.
Yes, provided the system is designed around:
- The mark type (etched vs printed vs embossed)
- Proper lighting (often angled/dark-field for etch marks)
- Sufficient resolution and focus
- Code localization and verification rules
Poor contrast or surface glare is usually solved with lighting and optics changes.
A typical workflow includes:
- Locate the symbol region
- Attempt decode
- Validate content (format, expected prefixes, check digits)
- Optionally grade symbol quality (especially for compliance needs)
2D codes are typically more robust for small footprints and damaged symbols, but still rely on print/mark quality.
Key factors include:
- Contrast and edge sharpness
- Quiet zone spacing
- Curvature or distortion on bottles/tubes
- Motion blur at speed
- Glare from reflective packaging or films
Better lighting and stable capture conditions usually fix “intermittent read” issues.
Common approaches include:
- Diffuse lighting to reduce specular reflections
- Polarizing filters
- Coaxial lighting for flat reflective surfaces
- Camera angle adjustments so glare moves off the code region
Glare control is often more important than higher camera resolution.
Successful OCR on low contrast typically uses:
- Coaxial or diffuse lighting for even illumination
- Polarization to reduce glare
- Backlighting for certain translucent substrates
- Increased resolution/focus stability
For highly inconsistent printing, AI-driven OCR can boost robustness.
Yes—common steps are:
- Read text with OCR
- Validate date format and expected pattern rules
- Cross-check against production batch parameters
- Log results (pass/fail + string read + image)
This supports traceability and helps prevent mislabeling incidents.
Best practices include:
- Constrain print location and font where possible
- Use pattern-based OCR regions (anchored to stable features)
- Train OCR profiles for known fonts/variations
- Use AI OCR for high variability cases
- Implement confidence thresholds and exception handling
A well-defined changeover/recipe strategy is often the difference-maker.
Yes—typical reject logic includes:
- Missing code region
- Decode failure or low confidence
- Code content mismatch vs recipe rules
- Code quality below threshold
Most systems also store failure images and the attempted read string for troubleshooting.