Sunday, December 16, 2012

Who needs programming skills?

All the hard-core machine vision people I know pride themselves on their ability to generate thousands of lines of C# code at the drop of a hat. If Pierantonio Boriero of Matrox Imaging is correct, they may be taking the wrong approach.

Pierantonio is the author of, “Vision Library or Vision-Specific IDE?”published December 2012 on the Quality Magazine daughter site, Vision & Sensors. The article is a “compare and contrast” exercise that weighs the pros and cons of writing code that ties together tools from a vision library against using a menu-driven, drag-and-drop type of technique.

If you’re struggling to see the difference, take a quick look at MIL. This is a heavy-duty vision suite that requires the application developer be a dab hand with C++. Or he/she can get to work with Matrox Design Assistant, which is full of menus and drag-n-drop type functionality, making it easy for anyone to use, (but is only on the Iris smart camera, or so I believe.)

Or if you want another example, hope over to Cognex. Their VisionPro product is the same kind of highly-capable vision library that Matrox offer in MIL, but if you don’t want to code they have a QuickBuild IDE that aims to make life easier.

And which approach is best?

Pierantonio offers an interesting explanation. Read the Quality magazine article for the whole thing, but I’d summarize it thus: vision libraries work really well when you’re (a) really experienced, and (b) replicating an application. Everyone else should take the IDE route.

No comments: