What are the main types of vision inspection systems?

Vision inspection systems are typically grouped into 1D (line profile), 2D (area image), and 3D (depth / point cloud) systems. Camera choices split further into area scan and line scan, with deep-learning models layered on top for defect detection on variable parts.

What is the difference between area scan and line scan cameras?

Area scan cameras capture a full 2D frame in one exposure - ideal for discrete parts on indexed lines. Line scan cameras capture one row of pixels at a time and build an image as the part moves - ideal for continuous webs, rolls, cylinders, and high-resolution surface inspection.

When do you need a 3D vision inspection system?

Use 3D vision when defects or measurements depend on shape and depth - weld bead profile, dent and dimple detection, volume measurement, robotic bin picking, or assembly verification where parts can shift in Z.

Types of Vision Inspection Systems

1D, 2D, and 3D vision systems each solve a different inspection problem on the factory floor. Picking the right combination - and the right camera class behind it - is what makes automated inspection actually catch defects.

The three families of vision inspection

Industrial vision inspection systems are grouped by the kind of data they capture: a 1D profile, a 2D image, or a 3D depth map. The grouping matters because each family has its own cameras, lighting, calibration, and software pipeline.

Type	What it captures	Typical cameras	Best fit
1D vision	Captures a line profile - variation along a single axis.	Line scan + laser displacement	Continuous webs (paper, film, foil), labels, glass, thickness gauging.
2D vision	Captures a flat image - the workhorse of modern factories.	Area scan; line scan for high-resolution surfaces	Surface defect detection, presence/absence, OCR, barcode, assembly verification.
3D vision	Captures depth - shape, height, volume, orientation.	Structured light, laser triangulation, stereo, time-of-flight	Weld bead profile, dent/dimple, bin picking, dimensional QA on flexible parts.

Area scan vs line scan cameras

Within 2D inspection, the camera choice usually comes down to area scan vs line scan. Area scan grabs a full rectangular image in one exposure - the right tool for discrete, indexed parts: castings, PCBs, assemblies, bottles. Line scan captures one pixel row at a time and stitches a tall, high-resolution image as the part moves under it - the right tool for continuous webs and cylindrical parts where a single area-scan frame would miss most of the surface.

Line scan also wins for very high-resolution surface inspection: it's easier to push 16k pixels across a moving web than to build an equivalent area sensor.

Where deep learning fits

The system type determines what data you can capture. Deep learning determines what you can detect in that data. Modern factories combine the two: area-scan 2D cameras feeding a deep-learning model for surface defects on automotive panels; line-scan cameras feeding a model for coating flaws on rolled metal; 3D structured-light scanners feeding a model for weld bead quality. Advitiix AI-Inspect runs those deep-learning models on edge hardware next to the line - sub-100 ms inference, on-prem, retrainable as defects and SKUs evolve.

How to choose

Continuous material (web, roll, cylinder) → 1D / line scan.
Discrete parts, flat presentation (PCB, casting, bottle, package) → 2D area scan.
Shape, depth, or volume matters (weld, seal, robotic pick, dent) → 3D.
Defects are subtle or variable (porosity, scratches, contamination) → deep-learning layer on top of whatever capture system you choose.

Types of Vision Inspection Systems

The three families of vision inspection

Area scan vs line scan cameras

Where deep learning fits

How to choose

Not sure which vision system fits your line?