Vision systems can be categorized in several ways, both in terms of their complexity and the type of inspection performed.
First, there is a distinction between 2D and 3D vision systems. 2D systems are suitable for tasks such as pattern recognition, object identification, and size measurement - all tasks where you need to "count pixels" - while 3D systems are used for more complex applications that require spatial understanding, such as cracks or other surface defects that are not visible in a standard 2D image.
There is also "traditional vision", which is based on an algorithm programmed to find specific defect types, versus vision systems based on artificial intelligence, where advanced neural networks are trained to identify and classify many different defect types at once.
The most robust solutions often consist of combinations of multiple technologies - for example, traditional 2D vision combined with machine learning. In this way, "hybrid vision" - as we call it - is created, which, through solid engineering, brings the best approaches together.
In addition, it makes sense to distinguish between standard machine vision and customized solutions. For simpler inspection tasks, it is possible to buy standard solutions that include both hardware and software in one package, making it relatively cheap and easy to get started with visual inspection.
However, if the type of quality control to be performed is more complex, a customized solution adapted to the specific use case will often be required. Learn more about how far you can get with a standard machine vision solution here.