Glossary · Machine vision
3D vision
Also known as: 3D machine vision, 3D imaging
German: 3D-Bildverarbeitung
In machine vision, 3D vision is the acquisition and evaluation of three-dimensional shape information, such as depth maps or point clouds, using methods like stereo vision, laser triangulation, structured light or time-of-flight measurement.
- Machine vision
In one sentence
3D vision captures and evaluates depth or point-cloud data using stereo, laser triangulation, structured light or time-of-flight.
Example
3D vision measures the height of each box in a mixed pallet so the robot can pick the top boxes without prior knowledge of the layout.
How it applies
- Engineering: The method is chosen by accuracy, speed, measuring volume and surface properties. Shiny, dark or transparent surfaces are difficult for most methods; motion of the part favors single-shot methods.
- Commissioning: 3D sensors need calibration, often a Hand-eye calibration to the robot, and tuning of exposure and filtering. Results depend strongly on ambient light for some methods.
- Documentation: The documentation team should state the measuring principle, the measuring range, the laser class if a laser is used, and the conditions under which results were validated.
3D vision vs. 2D vision
2D vision evaluates intensity images and finds position in a plane. 3D vision adds height and orientation, needed for bin picking, volume measurement and inspecting surface shape.