3D Machine Vision
A new and upcoming trend in Machine Vision has been 3D analysis.Recently at the Vision Show in Stuttgart, I noticed a marked increase in products and solutions related to 3d Machine Vision. Right from Lasers used for 3D triangulation to 3D smart cameras – 3D was everywhere. But what are the basic principle behind 3d analysis in Machine Vision?
Basically there are two broad techniques applied in 3d analysis:
- Stereo Vision
- LED/Laser Triangulation
In Stereo Vision two cameras are placed at two different angles at the same object/area of inspection thus grabbing two different perspectives of the same object. The images are then super-imposed using specialized algorithms and a point cloud can be obtained from this. This is show in the below figure. The principle of this technique can be compared with the way the human eye sees (offering two different perspectives of the same object).
The downside to this techniques however is that two cameras are required and can raise the cost of the solution. Stereo vision is common in robotics and surveillance applications.
In this technique a light pattern is projected onto an object. A camera is positioned to view this object from an angle to extrapolate the 3rd dimension based on the image captured. This is a very common technique used in measurement or applications where inspection has to happen on a 3rd dimensional place across the projection of the laser.
LEDs have a more consistent light pattern and is recommended to be used when the inspection accuracy is more demanding.
Also to be very successful in 3d analysis – one must calibrate the camera and the system to a high degree of accuracy. In order to prevent parallax errors it also recommended that LED pattern projects with a telecentric lens (to project the light pattern) is used. The principle is explained below in the figures: