max min math function in Vision Builder calculator

[Lorne H.]

Greg,

I have an answer to your problem.  It turns out there is bug with our subpixel edge detection algorithm.  This is supposed to be fixed in the next version of Vision Builder for Automated Inspection.

I am sorry it took so long to figure this out.

If I were you, I would continue to use pattern match instead of edge detection.

Lorne Hengst
Application Engineer
National Instruments

[Greg Shearer]

Lorne,

Thanks again for your response. I'd pretty much convinced myself there must have been a bug, as the edge detection based graphs I'd generated appeared to have a resolution of only about 1 pixel. Fortunately a similar bug wasn't present in the pattern matching algorithm, as it works fine!!! I'm about to go on leave for two weeks, so don't expect to install the system until the last week of January. This may be brought forward if there are problems getting the old system to operate correctly next week, once production recommences. It can be very temperamental. Just have to wait and see!

From the output obtained from the pattern matching test arrangement I'm now very confident that the replacement system will perform as a drop-in replacement as I originally planned. Once again, thanks so much for your efforts in resolving this issue. All else is pretty much fine as far as this project goes.

Can you suggest the best way to make suggestions for future enhancements? I have at least one! Would be nice to have a simple flat file writing facility. In my situation, this would allow characteristics such as calibration factors (e.g. camera image separation distance) to be written once by the calibration inspection, and then repeatedly read by subsequent inspections. In my current system, this figure is determined by the calibration inspection, but must then be edited into the normal inspection calculations. Can't make use of storage via the DataSocket mechanism (other than by editing an initial value), as this would be lost whenever the system is rebooted. A simple flat file writing mechanism would allow the whole calibration process to be automated, with no manual transfer of calibration data.

I wouldn't be surprised if this feature is already available .... but I haven't found it anywhere! Anyway, with a bit of luck this will be the last you'll hear from me on this thread. Following the success of this first vision project I'm on the lookout for others. I believe there may be opportunities for additional extensometry systems, as well as profiling systems in one of our other businesses.

Greg

[Lorne H.]

Greg,

"Can you suggest the best way to make suggestions for future enhancements? I have at least one! Would be nice to have a simple flat file writing facility. In my situation, this would allow characteristics such as calibration factors (e.g. camera image separation distance) to be written once by the calibration inspection, and then repeatedly read by subsequent inspections. In my current system, this figure is determined by the calibration inspection, but must then be edited into the normal inspection calculations. Can't make use of storage via the DataSocket mechanism (other than by editing an initial value), as this would be lost whenever the system is rebooted. A simple flat file writing mechanism would allow the whole calibration process to be automated, with no manual transfer of calibration data."

Sounds like a great idea.  The developers over here really take to heart suggestions made by customers.  You can make a product suggestio by going to...

www.ni.com and selecting the "Contact NI" link in the top left corner of the screen.  When the new page pops up there will be a link for "Product Feedback" in the bottom left of the screen.

or follow the link below...

http://sine.ni.com/apps/utf8/nicc.call_me

It was great working with you.

Lorne Hengst
Application Engineer
National Instruments