Knowledge · Cameras
To inspect a complete part, the camera must capture the required width and height with enough detail. An area scan camera uses a rectangular pixel array to form that 2D image. This guide explains image formation, shutter timing and the limits imposed by field of view and movement.
Use the worked example to relate feature size to pixels, then test the actual image before choosing a model.

Compare full-frame and line-based acquisition when the required area cannot be captured with sufficient detail in one image. Then continue with model selection.
An area scan camera is an industrial camera with a two-dimensional array of pixels. It records a complete image covering both width and height. With a global shutter the pixels share the exposure interval; with a rolling shutter different rows are exposed at different times, which can distort moving objects.
Area scan is worth evaluating when the required inspection area fits in one image and the relevant feature is visible from that view. Products can be stationary or moving, provided exposure and shutter timing meet the image-quality requirement. Processing may run in the camera or on an external computer; that is a separate architectural choice.
In product catalogues you will also see labels such as matrix camera, 2D industrial camera, or simply Computer Vision camera. Compare available models under Cameras. Confirm the image dimensions and shutter type rather than relying on the category label.
In one sentence
Area scan describes a two-dimensional sensor layout. It does not by itself specify simultaneous exposure, inspection accuracy or onboard processing.
Follow the image from the illuminated product through the lens, sensor and data connection. Each stage can limit the usable detail.

A supported trigger requests an image when the product is in position. During exposure, movement spreads an edge across pixels. For example, a permitted displacement of 0.08 mm at 800 mm/s gives a maximum exposure of 0.08 ÷ 800 = 100 µs. This is a motion budget; the light must still deliver sufficient signal in that time.
The lens maps the object area onto the sensor. Check field width and height, focus across the allowed height variation and sharpness at the corners. For the optical calculations, use the lenses guide.
Each sensor pixel accumulates a signal during exposure. A monochrome sensor records intensity; a typical Bayer colour sensor samples different colour bands at neighbouring pixels and reconstructs colour. Obtain active sensor width, height and pixel pitch from the datasheet. Optical-format labels such as 1 inch are not literal sensor diagonals in inches.
After readout, the camera transfers images through its data interface to processing software. Check the selected camera, driver and software combination for the required pixel formats, trigger modes and transfer rate. A standard interface helps compatibility, but does not guarantee that every camera feature is available in every software package.
| Stage | What happens | What this means |
|---|---|---|
| Trigger | External or free-run start of exposure | Match PLC / encoder timing |
| Exposure | Sensor integrates light | Motion blur vs light budget |
| Readout | Pixels become a digital frame | Frame rate & ROI limits |
| Transfer | Image leaves the camera | Cable length, bandwidth, PC load |
Teams often ask for “an industrial camera” when they actually need a family decision. Use this comparison as a filter before you open a datasheet.
| Family | Captures | Best when | Watch-outs |
|---|---|---|---|
| Area scan | Full 2D frame per trigger | Discrete parts in a FOV; most QC stations | Fast continuous webs may need line scan |
| Line scan | One line at a time, built into a 2D image | High-speed conveyors, film, metal strip, print | Needs precise motion sync & lighting |
| Smart camera | Image + onboard processing | Standard ID, OCR, presence, simple QC | Less flexible than a full PC stack |
| 3D camera | Shape, height, volume, pose | Bin picking, volume, height defects | Overkill if contrast in 2D is enough |
Area scan vs line scan: area scan captures a two-dimensional field. A line-scan system assembles successive image lines using controlled relative movement. Compare line scan when a continuous surface or the required width/detail combination makes a full-frame approach difficult. Speed alone does not decide the choice.
Area scan vs smart camera: a “plain” area scan camera delivers images to a PC or controller. A smart camera embeds tools on the device. Architecture comes first, see also our smart camera guide and smart camera vs PC-based vision, then you pick the sensor family.
Area scan vs 3D: a single 2D image does not generally establish an unknown surface height. If the decision requires depth or three-dimensional position, compare 3D.

The following tasks can use an area image when the required feature is visible at sufficient contrast:
For each task, first identify the viewing direction and the smallest relevant detail. A hidden underside, unknown height or feature outside the field may require another view or measurement method.
Use these parameters to interpret the image-forming limits. The dedicated selection guides cover model and architecture comparisons in more detail.
Image dimensions state the horizontal and vertical pixel counts. Divide the required field width by its horizontal count to calculate object-space sampling. A feature must span enough pixels for the chosen algorithm and contrast. More pixels also increase image data and may increase transfer or processing time; validate the complete inspection.
Compare the actual sensor diagonal with the lens’s usable image circle. Verify corner sharpness as well as coverage. A matching C-mount thread alone does not prove optical compatibility.
Check the achievable image rate at the selected image size, pixel format and exposure. Cropping the active image region may improve the rate on some models, but only if the device and transport support it. Measure dropped images and trigger acceptance under production load.
Use colour when colour differences are part of the decision. If the task relies on intensity or edges, compare monochrome imaging under the proposed light. Bayer colour reconstruction changes spatial sampling, so validate the smallest feature rather than comparing pixel counts alone.
GigE Vision transports image data over Ethernet. Check link speed, cable specification, available bandwidth and any shared switch paths. USB3 Vision uses a USB connection. Check the supported cable assembly and whether cameras share a host controller. Other interfaces may suit different transfer or installation requirements; use the required data rate and cable route to compare them.
With a global shutter, pixels share the exposure interval. With a rolling shutter, row exposure starts at different times. Motion can therefore change geometry even when individual rows appear sharp. Test the actual movement and any pulsed lighting with the chosen shutter; readout order alone does not determine exposure timing.
Ring, bar, dome and backlighting create different contrast. Continuous or pulsed operation is a separate timing choice. Test whether the relevant feature remains visible under product variation using the lighting page when you design the station.

Design the image acquisition and processing components together:
On Sedeco product pages for area scan families you will typically see options such as sensor model, resolution, and max. frame rate, plus interface cues in the SKU or specifications (GigE vs USB, mono vs color). Read them as answers to the eight questions above:
Use the requirements established here to compare the exact model specifications in the camera catalogue.
Illustrative inputs: an 80 mm field must reveal a 0.4 mm burr. Start with a provisional requirement of five pixels across the burr, then validate that assumption on real images. Position variation may require a field wider than the product itself.
Required sampling is 0.4 ÷ 5 = 0.08 mm/pixel. Across 80 mm this requires at least 1,000 horizontal pixels. A hypothetical 1,280 × 1,024 image at that field gives 0.0625 mm/pixel horizontally and a 64 mm vertical field with square pixels. Check that the product and its movement fit both dimensions. At a 160 mm horizontal field, the same five-pixel target requires 2,000 horizontal pixels.
This example assumes five pixels across the defect; it is a design input to validate, not demonstrated detection performance. Test the actual lens, illumination, part variation and inspection software before selecting the camera. Prepare the trial using the feasibility checklist.
Rule of thumb
Sampling describes pixel spacing on the object. It is not measurement uncertainty or a guarantee that a defect will be detected. Test optical contrast, focus and the inspection decision separately.
Area scan shows up differently by sector. The sensor family stays the same; the lighting and acceptance criteria change.
For packaging, distinguish visible print or assembly defects from seal strength or leaks, which an ordinary surface image may not establish. Test every relevant format and surface finish.
For component inspection, include actual position and height variation from the feeder or robot. Dimensional measurement also requires calibration and an independent reference.
For electronics, a visible polarity mark or missing pin may be suitable for 2D inspection. A height or coplanarity requirement may need another measurement method. Verify visibility before increasing resolution.
For print and code inspection, agree whether the task is reading, content verification or formal print-quality grading. These require different acceptance evidence and software functions.
For repeat machines, retain camera, lens, lighting and software versions with the settings and accepted images. Test restoration after a camera replacement instead of assuming that the same model name guarantees the same configuration.
An area scan camera rarely stands alone. Typical industrial loops look like this:
Measure trigger-to-result time including image acquisition, transfer, processing and communication. The result deadline may occur before the next product arrives. Test communication loss, trigger-while-busy behaviour and association of each result with the correct product.
If you are unsure whether processing should live on a smart camera or a PC, settle that architecture question before freezing the area-scan SKU, again via the smart vs PCdecision page.
False rejects and missed defects can originate in optics, lighting, product presentation or processing. Test the complete setup with known good products, defects and borderline cases. Record the inspected quantity and conditions, and report missed defects separately from false rejects.
Include the camera, lens, lighting, mounts, enclosure, cables, processing hardware and commissioning in the investment estimate. Account for cleaning, calibration and maintenance. Compare the trial results with measured manual inspection effort, scrap and rework to assess the business case.
Use independent validation samples, including allowed contamination, reflections, product positions and heights. Agree separate limits for missed defects, false rejects and result time. Retain images and expected outcomes so that the test can be repeated after maintenance.
In everyday language, often yes, but Computer Vision also includes line scan, 3D, and smart cameras. Area scan specifically means a 2D matrix sensor capturing a full frame per exposure.
When product moves continuously at high speed and you need fine defects along a long web or strip, or when a single area-scan FOV cannot cover the width at the required resolution.
Only if color is part of the decision. Otherwise mono is usually sharper and more light-efficient for industrial inspection.
GigE for longer cables and multi-camera machines; USB3 for short distances and simple single-camera cells next to the PC. Bandwidth needs and PC ports also matter.
Yes, if resolution, optics, and lighting support both tasks, sometimes in one recipe, sometimes as sequential inspections. Extremely demanding code reading may still warrant a dedicated reader.
Start from the smallest feature and required pixels across it, then compute sensor resolution from FOV. Validate on real samples; do not buy “the highest MP” by default.
Most industrial models follow GenICam / GigE Vision or USB3 Vision and work with vendor SDKs or independent libraries. Smart cameras use onboard environments; PC-based systems use full vision packages.
You now have the category model: area scan captures a full 2D frame for discrete inspection, measurement, ID, and guidance, provided optics and lighting are designed with the camera.
Compare camera types using the required image and processing task.
FOV, working distance and distortion for your sensor format.
Backlight, ring, dome, bar and strobe for stable contrast.
Send part dimensions, inspection criteria, cycle time and representative samples.
Related knowledge hub: Kennis / Knowledge. Continue with line scan, 3D cameras, smart cameras, code readers, lenses and lighting.
Next step
Use the selection guide to turn the tested field of view, feature size and movement into requirements for the camera, optics and trigger interface.