What is 3D Machine Vision?
3D vision is becoming more popular and more mainstream within machine vision circles. Why? Because it is a powerful technology capable of providing more accuracy for localisation, recognition, and inspection tasks that traditional 2D machine vision systems cannot reliably or repeatably succeed at.
As machine vision applications grow more complex, more creative solutions are required to solve more difficult problems in machine vision. 3D machine vision comprises an alternative set of technologies to 2D machine vision which aim to process these issues in greater depth and provide solutions to difficulties that 2D systems cannot solve.
How do we reliably and repeatably determine quantity in this context?
3D machine vision systems utilise 4 main forms of technology to generate 3-dimensional images of an object: Stereo Vision, Time of Flight (ToF), Laser Triangulation (3D Profiling), and Structured Light.
A 3D vision system furthers the analogy of machine vision as the ‘eyes’ of a computer system, as the addition of accurate depth perception functions more similarly to human eyes.
Stereo vision, for example, utilises two side-by-side cameras, calibrated and focused on the same object to provide full field of view 3D measurements in an unstructured and dynamic environment, based on triangulation of rays from multiple perspectives.
Laser triangulation, by contrast, measures the alteration of a laser beam when projected onto the object using a camera perpendicular to the beam. Where stereo vision can be used to capture stationary objects, laser triangulation requires a continuous linear motion, which can be achieved with a conveyor belt for example. This constraint is resolved in other ways, however, as laser triangulation can provide a spectacularly detailed point cloud map of the object.
Laser Triangulation as demonstrated with the Matrox AltiZ Camera
Time of Flight (ToF), alternatively, measures the time it takes for light from a modulated illumination source to reach the object, generating a point cloud based on these recorded times.
Time of Flight as demonstrated with the Helios2 Camera from LUCID Vision Labs
3D technology has allowed for many creative solutions to the question of depth, and so there are a variety of options to choose from when considering 3D machine vision systems. Before deciding on which of the 4 main 3D technologies to choose, there are things to consider in your intended machine vision application.
Is 3D Machine Vision Right for Me?
3D systems are intrinsically more complicated than 2D systems, which are far more common for most applications, not to mention cheaper. But looking past the price tag and setup, you will find a system that can achieve far more powerful results than any 2D camera.
3D machine vision can be useful for applications that require more accuracy of the size, texture, and depth of the object in question.
For example, agriculture, manufacturing, inspection, and quality control can all benefit from 3D vision, but deciding between 3D technologies will ultimately depend on factors such as the level of accuracy required, speed of measurement, whether your object is fixed or moving, and the reflectivity and texture of the surfaces on your object.
For more information on the differences between 3D machine vision technologies, take a look at our e-book.
Interested in designing a 3D vision system? Check out our video presentation above.
2D vs 3D Machine Vision Systems
The traditional two-dimensional machine vision system when used in tandem with imaging library software has been proven to be very successful in applications such as barcode reading, presence detection, and object tracking, and these technologies are only improving with time.
However, since 2D cameras simply take an image of light reflected from the object, changes in illumination can have adverse effects on accuracy when taking measurements. Too much light can create an overexposed shot, leading to light bleeding or blurred edges of the object, and insufficient illumination can adversely affect the clarity of edges and features that appear on the 2-dimensional image.
In applications where illumination cannot be easily controlled, and therefore cannot be altered to fix the shot, this creates a problem within 2D machine vision systems.
Black text on a black object: 3D systems in tandem with 2D image processing can solve this issue
3D machine vision cameras can offset this by having the capability of recording accurate depth information, thus generating a point cloud, which is a far superior object in terms of accuracy.
Every pixel of the object is accounted for in space, and the user is provided with X, Y and Z plane data as well as the corresponding rotational data for each of the axes.
This makes 3D machine vision an exceptional option compared to 2D in the context of applications involving dimensioning, space management, thickness measurement, Z-axis surface detection and quality control involving depth. Traditional 2D image processing can still be used with the collected images, creating an implementable solution to many machine vision problems.
For further information on the above feel free to consult our informative e-book on 3D Imaging Techniques. Specifications for different 3D imaging solutions can be found in the data sheets of our cameras, available on our website to help you make the decision when choosing the optimal 3D machine vision camera model for your industrial application.