Industrial Vision: seeing is believing

Seeing is believing, and the same applies to technology. The manufacturing industry is keen to get a grip on their processes, in various manifestations. This is usually solved by measuring variables before (pre-inspection) and after (post-inspection) the process. Certain product data may also be needed for the production process. Consider, for instance, the product coordinates so that a robot can grab it (positioning).

A widely used method for obtaining this data is 'Industrial Vision'. This is a solid camera set-up which, through the right exposure, can process images. By taking a number of pictures, measurements and checks can be performed through image analysis. The power of this system lies in the image analysis software and the lighting used. The software determines what happens to the image. It determines, for example, whether a product has been manufactured to the desired specifications by comparing this image to an example. Lighting is important, because the object to be inspected must always be clearly displayed without it being affected by ambient lighting. Think of sunlight hitting an object at a specific time, for instance.

What makes vision technology so powerful is that instead of a local measurement of a laser sensor or physical sensor, for example, you can perform multiple measurements at once with a single photo. Vision technology is also used for reading barcodes (Track and Tracing). One disadvantage of a Vision solution is that it requires a lot of specific knowledge, and the
solution is often more costly than individual sensor technology.


There are two types of cameras: smart cameras and industrial cameras. As the name suggests, the difference is in intelligence. Smart cameras have an integrated processor and memory containing the image processing software. The software is programmed externally (on a PC) and downloaded onto the camera. Because the software is integrated in the camera, you do not need an external PC. Smart cameras often have digital outputs for a correct/error message, or a communication protocol that communicates the measurement results. A disadvantage of this system is that the purchase price of a smart camera is often high because of the intelligence it contains.

An industrial camera is the variant without integrated image processing software. The camera transmits direct images through a specific protocol and an external PC has to process these images. Because the software runs on a PC, this is often camera-independent. The vision library in the PC software is often more extensive than that on a smart camera. In addition, multiple cameras can be connected to the PC. The disadvantage of this system is that it requires a separate PC, but the advantage is that the average industrial camera is much cheaper than a smart camera.


Upcoming trends in the Vision industry are, in particular, 3D Vision. Normally, a flat image is created when a picture is taken. This provides a lot of information, but not the height: it is a 2D image. By applying new techniques, a 3D image can be created so that the height component can also be obtained. A technique for this is stereovision. By using two cameras in a fixed set-up, the depth can also be mapped and then measured after calibration. With 3D image information, 3D algorithms can also be used, for example comparing a 3D image with a 3D CAD file.