Automated surface and defect inspection

Scratches on metal, voids in plastic, burrs on a contour or contamination on a surface are classic machine vision tasks.

They are also the applications where good system design matters most, because a defect can only be detected if it is sufficiently visible in the captured image.

Discuss your application

Typical defect types

scratches
cracks
dents
pores
burrs
chips
contamination
stains
coating defects
material deviations
print defects
damaged edges
voids
uneven surface structures

With surfaces, the lighting usually decides

A more expensive camera helps little if the defect you are looking for produces virtually no contrast in the image.

Depending on the surface, a different lighting geometry may be needed:

Dark field: Can bring out scratches and edges.
Coaxial light: Can help on flat reflective surfaces.
Dome lighting: Can even out reflections on curved or glossy surfaces.
Backlight: Particularly suited to contours and edges.

Wavelength, polarization and angle of incidence come on top of that.

Area scan or line scan camera?

For individual workpieces, an area scan camera is often suitable.

For continuous materials or long surfaces, line scan cameras can be an advantage, for example with:

film
paper
textiles
sheet metal
web material
continuously conveyed products

Classic machine vision or AI?

Not every defect inspection needs artificial intelligence.

A clearly defined scratch with reproducible contrast can often be detected very well with rule-based methods.

AI and deep learning become interesting when good parts have a natural variation and defects are hard to describe using fixed thresholds, contours or geometric rules.

This information helps us specify the system

Example values - for surface inspection, a test with real parts is usually the fastest route.

Field of view: 120 × 80 mm
Smallest defect: 0.2 mm
Working distance: approx. 300 mm
Cycle time: 500 ms
Material: polished aluminum

Suitable product groups

A test with real parts gives the fastest answer

Send us sample parts, images or a description of the smallest relevant defect.

Request a feasibility check

Frequently asked questions

Why is lighting so important for surface inspection?

A defect can only be detected if it produces contrast in the image. Depending on the lighting geometry, scratches on shiny metal are either clearly visible or practically invisible - camera resolution changes little about that.

When does a line scan camera make sense?

Above all for continuously moving web material such as film, paper, textiles or sheet metal, and for very long surfaces that cannot be captured in a single image.

Do I always need AI for defect inspection?

No. Clearly defined defects with reproducible contrast can usually be solved with rules. AI pays off when good parts vary widely and defects cannot be described by fixed rules.