Industrial Inspection Camera Selection: Shutter, Resolution, Frame Rate and Interface

The following sequence provides a structured approach to evaluating camera requirements for industrial inspection applications:

Inspection task → motion → smallest feature → acquisition speed → optics / lighting → camera position → interface → validation

The camera should only be selected after these requirements are understood.

Shutter type becomes particularly important when the object or camera moves during exposure.

Global shutter sensors expose all pixels simultaneously, helping minimize motion-induced geometric distortion.
Rolling shutter sensors expose successive rows of pixels at different times, which can introduce geometric distortion when the camera or subject is moving. Rolling shutter remains suitable for applications where these effects are negligible or can be controlled.

Consider global shutter sensors when:
•products move continuously through the inspection area;
•dimensional accuracy is required while the object moves;
•object position must remain geometrically correct;
•capture is synchronized with motion or external triggering;
•multiple cameras require coordinated acquisition.

A rolling shutter sensor may fit when:
•the product is stationary during acquisition;
•movement can be stopped or controlled;
•the application prioritizes high-resolution static imaging;
•motion distortion is not critical to the inspection result.

A global shutter sensor does not automatically eliminate motion blur. If the object moves significantly during the exposure period, the image can still become blurred.

For a more detailed explanation of how the two shutter technologies affect moving and stationary applications, see Global Shutter vs Rolling Shutter Cameras.

The required resolution should be based on the smallest feature that must be detected across the required field of view.
Higher megapixel counts are not automatically better.

A useful starting point is:
Pixels per mm = horizontal image resolution ÷ horizontal field of view

Then:
Pixels across feature = feature size × pixels per mm
Example
Suppose an inspection system covers a 200 mm field of view using a camera with 4000 horizontal pixels.
4000 ÷ 200 = 20 pixels/mm
If the smallest defect is 0.5 mm:
0.5 × 20 = 10 pixels
The defect would therefore cover approximately 10 pixels in the image.


Whether this is sufficient depends on:
•defect contrast;
•optics;
•lighting;
•inspection algorithm;
•required reliability.

The calculation should therefore be used as a starting point rather than a universal threshold.

The important selection principle is:
Smaller feature → more required image detail → potentially higher resolution → more image data → greater bandwidth and processing demand.
The camera-selection material similarly treats required detail and processing speed together rather than recommending resolution independently. 

Frame rate determines how frequently the camera can provide images to the inspection system.

The required frame rate depends on:

•production speed;
•object spacing;
•inspection frequency;
•triggering;
•exposure time;
•number of cameras;
•required processing cycle.

A fast production line does not necessarily mean the camera must run continuously at a very high frame rate.
For example, if each product triggers one image at a defined inspection position, the important requirement may be the required trigger frequency rather than the maximum continuous FPS.

For continuous image acquisition, frame rate becomes more directly related to how quickly objects move through the field of view.

Resolution and frame rate must also be considered together:
Higher resolution + higher FPS = more image data per second
That affects both interface bandwidth and processor load.

These examples illustrate how inspection requirements translate into camera decisions.

• Conveyor defect inspection
  Requirement: Detect small defects on continuously moving products.

Camera implications:
A global shutter may be preferred because the product moves during exposure. The resolution must provide enough pixels across the smallest relevant defect, while the frame rate must keep pace with the required inspection frequency.

Short exposure times may also increase the lighting requirement.

Interface implication:
If the camera is integrated close to the processor, MIPI CSI-2 may fit. For remote installation, GMSL2 or FPD-Link III may be more appropriate.

• High-resolution stationary surface inspection
  Requirement: Inspect fine surface features while the product is stationary.

Camera implications:
Resolution and image quality may be more important than high frame rate. Because the object does not move during acquisition, a rolling shutter may also be suitable.

System implication:
Optics and lighting can become more important than maximum camera speed because the main challenge is making the defect clearly visible.

• Remote inspection inside industrial equipment
  Requirement: Inspect a component several meters from the processing system.

Camera implications:
Shutter, resolution, and FPS still depend on the inspection task.

Interface implication:
The physical distance between camera and processor becomes an additional constraint. A GMSL2 or FPD-Link III SerDes architecture may be more appropriate than direct MIPI CSI-2 integration.