|
M |
achine vision has progressed a lot since the early 1980's when we first starting using some of the early GE Optovision systems to sort objects by measuring area and perimeter. Today vision systems can do a lot more – just don’t expect them to be as good as a human.
|
Take the Machine Vision Aanalysis Tutorial available online from High Tech Services. |
Machine vision is easy if you can get a good image into the system. Getting a good image requires a lot of patience using different optics and lighting. You have to try an endless combination of cameras, lenses, filters, lights, and some other weird and unordinary devices in order to evaluate machine vision applications. It is amazing how some applications that appear to be simple turn out to be difficult and how some difficult applications turn out to be easy. You usually never know until you sit in front of the camera with the application. What makes it so frustrating is that engineers think that if a little is a little good then a lot should be a lot of good. With optics and lighting it can be direct and sometimes it is indirect relationship. It can be very frustrating.
What makes a good vision application?
A good application is one with large differences in contrast. For example, looking at a black part on a white background, or a white part on a black background. Sometimes this can be achieved by backlighting – putting the light behind the object being viewed and viewing the silhouette of the object.
Note that there are color cameras available. So looking for distinct colors is often easy to do.
Note that there are a lot of ways to increase the contrast of the part of the object you are trying to view. You can play with the gain on the frame grabber, you can run algorithms in software, you can use filters, or you can use stains for biological applications.Image is everything
Click here to view Machine Vision Aanalysis Tutorial in its entirety.
ARTICLES ON MACHINE VISION
VISION NEWS
VISION SYSTEMS WHITE PAPERS
VISION PRODUCTS
