Tools for real-world machine vision

The machine vision sales and marketing people work hard to make vision look easy, but as any one who’s actually developed an application knows, it isn’t. One of the issues I’ve wrestled with on many occasions is the need to mask out “don’t care” regions in an image. If you’re looking at parts that are 100% identical this isn’t a problem, but as soon as you have part-to-part variation you’ll want to tailor the mask for each image acquired.

European imaging specialists Euresys claim to have a solution to this. Their “Open eVision 1.1” suite of image analysis software now has a “Flexible Mask” tool which, it is claimed, “provide(s) a powerful way of restricting the processing to freely shaped parts of an image.” It sounds good, but I haven’t downloaded the evaluation version yet because it really demands C++ programming expertise to use and I lack the time, energy, and enthusiasm to get in to that. Give me a graphical environment any time!

But my lack of programming skills not withstanding, the Euresys site is worth a few minutes of your time. As well as software, they offer both analog and digital framegrabbers and they’re pretty much camera-agnostic, so there’s plenty of flexibility to work with.