Knowledge · Cameras

What is a line scan camera?

Inspecting a moving strip of film, paper or metal requires enough detail across its width and along its length. A line scan camera records successive rows of pixels as the material passes, building a two-dimensional image from that movement.

Learn how line frequency, exposure and measured movement determine the image, and which conditions to test before selecting hardware.

Industrial steel and continuous material inspection environment
Line scan can inspect continuous material such as metal strip, film, paper and textiles.

What a line scan camera is

A line scan camera records a narrow line of the object with each exposure. As the object or camera moves, successive lines form a two-dimensional image. Sensors may contain one row or several rows, for example for colour capture or Time Delay Integration (TDI), which combines signal from successive rows in step with movement.

This geometry suits long surfaces because image length follows the number of acquired lines. Area scan can also inspect moving material with short exposures and overlapping images. The choice depends on inspection width, feature size, movement and the effort needed to assemble and process the images.

In catalogues you may see line camera, linescan, web inspection camera or linear sensor camera. Available families are listed under Cameras. These names do not specify pixel count, line frequency or supported synchronization.

In one sentence

Each exposure records a strip across the object. Controlled movement supplies the second image dimension.

How a line scan camera works

Think of four linked decisions: motion → line exposure → optics & light → interface & reconstruction.

Industrial camera connected in a vision setup
Encoder, lighting and interface timing are as important as the sensor itself.

1. Motion supplies the second axis

The material moves past a fixed camera, or the camera moves across the object. Each line records a narrow part of the surface. Changes in the distance travelled between exposures stretch or compress the reconstructed image. Synchronise acquisition to measured movement when fixed-frequency acquisition cannot meet the required geometric tolerance.

2. Each line is a short exposure

Line frequency states how many lines are acquired per second. Available exposure depends on the camera mode and line interval. Test whether the light supplies enough signal at the required exposure; raising gain also amplifies noise. Slowing acquisition without slowing the material increases the distance between lines.

3. Optics map web width onto the line

Resolution across the material depends on how many active pixels cover the inspection width. The field of view and active pixel count determine object-space pixel size; lens quality and focus determine whether that detail is resolved. See the lenses guide when specifying the inspection width and working distance.

4. The interface streams lines continuously

The receiving system orders lines by acquisition sequence and groups them into images or processes a continuous stream. It must preserve line order and detect missing data. Interface and processing selection follow the calculated payload and synchronization requirements.

StageWhat happensWhat this means
Encoder / syncSpeed signal drives line acquisitionCorrect aspect ratio along travel
Line exposureOne row integrates lightLight budget vs line rate
ReadoutLine becomes digital samplesMax line rate & bit depth
Transfer & buildLines form a 2D image or streamBandwidth, grabber, PC load

Line scan vs area scan, smart, and 3D

The distinction is how the second image dimension is acquired. Use the dedicated comparison guide for the complete selection decision.

FamilyCapturesBest whenWatch-outs
Line scanOne line at a time, built into a 2D imageContinuous webs, high-speed strip, wide FOV at fine resolutionNeeds precise motion sync & strong line lighting
Area scanFull 2D frame per triggerDiscrete parts in a FOV; most QC stationsFast continuous webs may blur or miss coverage
Smart cameraImage + onboard processingStandard ID, OCR, presence on discrete partsRarely the first choice for high-speed web inspection
3D cameraShape, height, volume, poseHeight defects, volume, bin pickingDifferent problem than surface print on a flat web

Line scan vs area scan: area scan captures a rectangular field; line scan builds the travel direction from successive lines. Motion alone does not select the family. See the area scan guidefor the alternative image geometry.

Line scan vs smart camera: sensor geometry and processing location are separate choices. Check sustained data handling and machine interfaces using smart camera vs PC-based vision.

Line scan vs 3D: line scan answers “what does the surface look like along the travel?” 3D answers “what is the height or shape?” Some lines combine a 2D line scan for print/coating with a separate 3D profiler for thickness or warp.

Where line scan cameras are used

Industrial production environment for Computer Vision
Typical contexts: converting, print, metals, packaging webs and high-speed logistics belts.

Typical applications inspect a surface as it passes the imaging line:

  • Web and film inspection, Coating streaks, gels, holes, contamination and gauge marks on plastic film, foil and paper.
  • Print and converting, Register, missing print, color bars, splice detection and barcode/OCR on continuous packaging material.
  • Metal strip and steel, Surface scratches, roll marks, edge defects and scale on continuous strip at high line speeds.
  • Textiles and nonwovens, Holes, thick spots, contamination and weave defects across wide webs.
  • High-speed conveyors, Sorting or surface checks where parts never pause and a single area-scan FOV cannot cover width at the required resolution.
  • Flat panel and glass (selected lines), Long, continuous scanning of large sheets where stitching many area frames is less practical.

For each task, establish inspection width, smallest feature and material displacement per line. A continuous surface does not automatically rule out area scan.

Specifications that determine image quality

Read the following specifications as limits on image formation. The worked example connects the two sampling directions.

Pixels across the line (resolution)

Across-material sampling equals field width divided by active pixel count. Establish the provisional pixels per defect with the intended algorithm, then test optics and contrast. Sensor pixel count alone is not proof of usable detail.

Line rate

Along-travel spacing equals material speed divided by line frequency. Check the specified rate at the actual active width, pixel format and interface. Missing lines create a separate defect in the reconstructed image.

Encoder sync and trigger modes

An encoder measures movement and can trigger each line after a defined travel distance. Check the supported signal type and maximum input frequency. Fixed-frequency acquisition can be suitable when speed variation stays within the permitted image distortion. Measure that variation during acceleration, normal running and stopping.

Mono vs color vs multi-spectral

Monochrome imaging suits defects distinguished by brightness or shape. Choose colour when colour differences are part of the inspection criterion. Check whether separate sensor rows introduce displacement between colour channels at the expected movement and product height.

Interface and bandwidth

Calculate the image payload from active pixels per line × transmitted bits per pixel × lines per second. Add capacity for protocol overhead and confirm sustained transfer through the interface, acquisition hardware and PC. If images are stored, test storage throughput and retention separately.

Lighting dependency

Uneven lighting, reflections and insufficient exposure can hide defects. Test illumination across the full inspection width with representative surfaces. Use the lighting page when you design the station.

Before you buy, eight questions

  1. What is the web or conveyor width and the smallest defect that must be seen?
  2. What is maximum material speed (m/min or m/s)?
  3. How many pixels across the defect do you need for stable detection?
  4. Do you have (or can you add) a reliable encoder on the drive?
  5. Is the decision mono intensity, color, or both?
  6. What interface and PC/grabber architecture already exists on the line?
  7. What line-light geometry creates contrast on real samples?
  8. Who owns continuous inspection software and recipe management?

Building a working line-scan setup

Vision hardware stack with camera and controller
A working web station pairs the camera with line optics, lighting, encoder and compute.

Design the camera, optics, lighting and motion measurement together:

  1. Camera, pixels across, line rate, interface matched to speed and width.
  2. Lens, magnification and resolution class for µm/pixel across the web.
  3. Line lighting, uniform intensity, often high power; geometry matched to surface finish.
  4. Mechanics, rigid mount, vibration isolation, precise alignment perpendicular to travel.
  5. Encoder & sync, calibrated pulses per millimeter of travel.
  6. Compute & software, grabber or NIC, continuous inspection recipes, reject marking or logging.

Common first-system mistakes

  • Buying “8k” without calculating µm/pixel at the actual web width.
  • Ignoring encoder sync, then wondering why images look stretched.
  • Under-lighting the line and compensating with gain that hides small defects in noise.
  • Mounting the camera at a slight angle so the optical line is not square to travel.
  • Choosing an interface that cannot sustain peak line rate for a full shift.
  • Testing only clean, flat samples and missing the surface and presentation variation found in production.

How to read a line scan product page

On Sedeco product pages for line scan families you will typically see cues such as sensor resolution(pixels across), max. line rate, mono/color, and interface. Read them as answers to the eight questions above:

  • Pixels / resolution, samples across the web; combine with optics for µm/pixel.
  • Max. line rate, ceiling under stated conditions; confirm for your format and interface.
  • Data interface, how lines reach the PC or controller continuously.
  • Mono/color, intensity vs color decisions on print or coatings.
  • Encoder / trigger features, whether motion sync is native or needs grabber support.

Use the calculation below to establish your pixel count and line frequency before comparing cameras in the camera catalogue. Do not treat catalogue numbers as a substitute for a speed-and-resolution calculation on your web.

Worked example: line rate and sampling

Suppose you inspect a plastic film that is 1 200 mm wide. The critical defect is a 0.5 mm hole. You want at least 5 pixels across that hole across the web.

Required spatial sampling across web: 0.5 mm / 5 = 0.1 mm per pixel. Across 1 200 mm you need 1 200 / 0.1 = 12 000 pixels, so an 8k sensor alone is not enough unless you narrow the FOV, use multiple cameras, or accept fewer pixels on the defect. This is why width and defect size drive pixel count before brand preference.

Now speed: the web runs at 300 m/min = 5 m/s. If you also want 0.1 mm sampling along travel, you need a line every 0.1 mm → 5 / 0.0001 = 50 000 lines per second. That line rate must be available at your bit depth and interface, and lighting must deliver enough photons in each short exposure.

This is an illustrative calculation, not a demonstrated detection result. Five pixels across the hole is an assumed design input that must be tested with the actual optics, contrast and inspection software. Prepare representative samples using the feasibility checklist.

Rule of thumb

Across the material: pixel size = field width ÷ active pixels. Along travel: line spacing = transport speed ÷ line frequency. Use consistent units and check both directions against the required defect size.

Industry snapshots

The following examples show which process information accompanies the image. They are not reported installation results.

Converting and flexible packaging

For film or print, associate detected defects with their lengthwise and cross-web position. Include splices and changes in the permitted print or coating.

Metals and steel

For metal, test surface reflections and the required inspection width. Preserve defect position through any downstream cutting or rejection step.

Paper, pulp and textiles

For wide paper or textile, test any overlap between camera fields. Distinguish a stitching boundary from an actual material defect.

Logistics and high-speed sortation

Logistics and conveyor environment for Computer Vision
On fast belts, line scan can cover width without stopping parcels for a full-frame shot.

For conveyed objects, check height changes and gaps between products. Movement does not itself establish whether line scan or area scan provides the required view.

Machine builders and OEMs

For repeated machines, retain the encoder scaling, optics and acquisition settings with reference images. Test restoration on the next build.

Integrating with PLC, encoder, and IT

A line scan camera rarely stands alone. Typical industrial loops look like this:

  • Encoder in, pulses proportional to travel drive line acquisition.
  • Acquire, continuous lines under stable line light; stream to PC or vision controller.
  • Decide, software finds defects, maps them to meter marks, and grades severity.
  • Act, PLC marks, rejects, slows, or logs a roll report to MES.

Test continuous acquisition at maximum speed and during speed changes. Record missing lines, processing delays and buffer use. Include image storage and machine communication in the test so that the complete inspection chain is loaded.

If you are unsure whether processing should live on a smart device or a PC, settle architecture early, via the smart vs PCdecision page, then freeze the line-scan interface.

Cost, quality, and what good enough means

Image defects can originate in the camera, optics, illumination or motion measurement. Establish which combinations cause missed defects or false rejects by testing known samples and checking image geometry. Record the conditions so that maintenance staff can reproduce the test.

Include the camera, lens, lighting, encoder, mounting, acquisition hardware, software and commissioning in the investment estimate. Estimate running costs from cleaning, calibration, storage and maintenance. Compare these with measured manual inspection time, scrap, rework and stops; use your own production figures.

Agree acceptance limits for each defect class before testing. Include good material, known defects, borderline cases and expected surface variation at minimum, nominal and maximum speed. Report missed defects and false rejects separately, with the inspected quantity and test conditions. Also test restart, missing encoder pulses and downstream reject tracking.

Short glossary

Line rateThe number of lines captured per second. Required line frequency equals transport speed divided by the desired distance between lines, using consistent units.
Encoder syncUsing a motion pulse train so each line maps to a known distance of travel.
Web / stripContinuous material moving through the process (film, paper, metal, textile).
Across-web resolutionSpatial sampling perpendicular to travel, set by pixels and optics.
TDITime Delay Integration, multi-line sensors that accumulate light while tracking motion.
Frame grabberHardware that receives high-bandwidth camera data (common with Camera Link / CXP).
Line lightIllumination shaped as a bright line matched to the sensor’s view of the web.

FAQ: line scan cameras

Is a line scan camera the same as a Computer Vision camera?

It is one family within Computer Vision. Computer Vision also includes area scan, 3D and smart cameras. Line scan specifically means capturing successive lines synchronized to motion.

When should I choose area scan instead?

When the required inspection area fits a rectangular image at sufficient detail and exposure. Continuous or pulsed illumination can be suitable. See the area scan guide.

Do I always need an encoder?

An encoder is useful when speed changes would otherwise distort the image beyond the allowed tolerance. Fixed-frequency acquisition may be sufficient for a sufficiently stable speed. Verify actual material movement, encoder placement and slip; a motor signal does not always represent movement at the inspection point.

Why is lighting so critical for line scan?

The selected exposure must supply enough signal across the full line. Compare contrast at the fastest acquisition condition; required light depends on the surface, lens, sensor and permitted noise.

What does 8k mean?

About eight thousand pixels across the line, not “8 megapixels” in the area-scan sense. Convert to µm/pixel using your web width and optics.

Can one line scan camera do color print and tiny holes?

Sometimes, if resolution, optics and lighting support both. Demanding color and fine mono defects may need separate stations or dual sensors.

What software do line scan cameras use?

Typically PC or vision-controller packages that handle continuous acquisition, encoder mapping and defect classification. GenICam / interface standards apply on many modern models.

Next step

Which camera should I choose?

Use the camera-selection guide to turn the required image detail, movement and working distance into a choice of camera, optics and lighting.

Selecting cameras for Computer Vision