Glossary · Machine vision
Pixel resolution
Also known as: Spatial resolution, Object pixel size
German: Pixelauflösung
In machine vision, pixel resolution is either the number of pixels of an image sensor (for example 2448 x 2048) or, more usefully for applications, the size of the object area one pixel represents, obtained by dividing the field of view by the pixel count.
- Machine vision
In one sentence
Pixel resolution is the sensor's pixel count or, for applications, the object size one pixel covers: field of view divided by pixels.
Example
With a 200 mm field of view on 2,000 pixels, the pixel resolution is 0.1 mm, enough to detect scratches of about 0.3 mm.
How it applies
- Engineering: A common rule of thumb is that a feature to be detected should cover several pixels; for measurement, subpixel methods improve precision, but optics and lighting limit real accuracy.
- Commissioning: Pixel resolution is checked with a calibration target; lens distortion and perspective change it across the image unless corrected.
- Documentation: The documentation team should state which meaning is used (sensor pixels or object pixel size) and give both where relevant. Measurement capability should be expressed with Measurement uncertainty, not pixel size alone.
Pixel resolution vs. measurement accuracy
Pixel resolution describes how finely the image samples the object. Accuracy describes how close measurements come to the true value, which also depends on optics, calibration and the algorithm.