Imaging engineering
Line-Scan Imaging
A line-scan camera builds one image of an endless surface. Its accuracy depends on motion synchronization as much as on the sensor.
Direct answer
When should you use a line-scan camera?
Use line scan when the product is continuous or effectively endless — web, film, paper, extrusion, coil — or when a cylindrical part is rotated under the sensor, and when the required resolution across a wide surface would demand an impractical number of area-scan cameras.
How resolution is set
- Cross-web: web width ÷ sensor pixels. A 1,000 mm web on an 8,192-pixel sensor gives roughly 0.12 mm per pixel.
- Down-web: set by encoder pulses per line. Square pixels require matching the two, or geometric measurements distort.
- Line rate required ≈ web speed ÷ down-web resolution. Confirm the camera supports it with an achievable exposure.
Fig. — Line-scan resolution, encoder synchronisation and image construction
Cross-web resolution
web width ÷ sensor pixels
Down-web resolution
encoder increment per line trigger
Required line rate
v ÷ down-web resolution
Illumination is the hard part
Line scan integrates light only during a very short line period, so it needs high-intensity, highly uniform illumination along the whole scan line. Non-uniformity shows up as fixed vertical banding that flat-field correction can mask but not truly fix, and it will bias defect thresholds across the web.
- Use purpose-built line lights aligned to the scan line, not adapted bar lights.
- Perform flat-field/shading correction, and re-verify it on a schedule.
- Budget for illumination aging; measure, do not assume.
Where line scan is the wrong choice
Discrete parts arriving on an indexed conveyor, unstable or non-encoded motion, and short parts with generous resolution requirements are all better served by area scan. Line scan adds synchronization complexity that only pays back when the surface is genuinely continuous.
Related engineering material