Machine vision uses cameras, lighting and software to inspect parts, read codes, measure features and guide robots. Smart cameras put everything in one unit for simpler tasks; PC-based systems handle complex or multi-camera inspections. Lighting is often the deciding factor: consistent, well-chosen lighting makes inspections reliable.
Vision turns cameras into flexible sensors.
Common applications
| Task | Example |
|---|---|
| Presence or absence | Is the cap on? |
| Measurement | Hole diameter |
| Defect detection | Scratches, cracks |
| Code reading | Barcodes, 2D codes |
| Robot guidance | Locate parts for picking |
Smart camera or PC-based?
- Smart camera: single inspection, simpler setup.
- PC-based: multiple cameras, complex algorithms, AI tools.
Getting it right
- Define pass and fail clearly.
- Choose lighting first, then the camera and lens.
- Collect sample good and bad parts.
- Validate on production variation.
See industrial sensors and safety light curtains vs scanners. Browse sensors and vision.
Common questions
How do I choose camera resolution?
Work out the smallest feature you need to detect and make sure it covers several pixels in the field of view.
What lighting should I use?
It depends on the part: backlights for silhouettes, diffuse light for shiny parts, low-angle light for surface features.
Is AI-based vision worth it?
Deep learning tools help with variable defects and textures. Rule-based tools are simpler for measurements and presence checks.