Machine Vision

Definition

The digitization of the physical world in order to make a decision on how to inspect or interact with a process or task. This typically will involve hardware, to gather information and software to process this information.

This is analogous to the way our eyes capture light from the world and our brain determines what we are looking at.

Classes of machine vision

Machine vision hardware can broadly be split into two primary categories: 2D and 3D. 2D devices capture 2-dimensional information typically in the form of light intensity (contrast) and 3D devices capture 3-dimensional information most often relating to the physical structure of whatever is being viewed. Read more about when to use 3D over 2D machine vision here.

3D machine vision using geometric measurement techniques

3D machine vision is made possible using different measurement methods. Two major techniques are commonly employed in industrial automation: geometric and light-based techniques.

3D measurement devices using geometric measurement techniques include stereo vision, active stereo vision, laser triangulation, coded light, and structured light (non-exhaustive). Of which, laser triangulation is deemed the most reliable and produces reliable and accurate results. Industrial 3D scanners are often used to measure a target object’s dimensions, as well as to detect objects, and identify features. The most powerful application lies in processing the 3D data for production process optimization.

3D machine vision using light-based measurement techniques

Common light-based measurement techniques include time of flight, interferometry, Michelson interferometry, and confocal displacement (non-exhaustive).

« Back to Glossary Index