IMAQ ColorMatch Score of -1000 using internal ROI ID

[StefanW]

Hi at all,
the ColorMatch seems to have a problem when you input a ROI cluster containing contours with internal edges. For example if you use MaskToROI with internal edges activated and if you want to evaluated the color of a ring. For an internal ID the function returns a match score of -1000, as well as the outer ring is evaluated as it would be a circle and not a ring.

The behavior can be seen by running the attached example with the attached image.

Has any one the same problem? IMAQ ColorLearn seems to be able to handle internal ID's correctly.

Yours, stefan

[GuenterMueller]

Hi Stefan.
I confirm the behaviour (I have never used an internal contour in conjunction with IMAQ ColorMatch before). The documentation does not mention a Match Score of -1000. Instead, values between 0 and 1000 should be reported. I have been creating different ROIs with internal contours -> Match Score is always -1000.

But: The documentation says, that a Match Score for each region is calculated (!!!)  Obviously, IMAQ ColorMatch uses the -1000 to indicate a problem with a specific region. This is smarter than reporting an error. (BTW: Good job of the code of "IMAQ ColorMatch".)

"ColorMatching" your "ring" will require an ROI Descriptor with one(!) contour.

Could someone from NI please give me a link or a note about Match Score = -1000? Am I wrong? Is there anything else to consider? Thanks in advance.
Guenter

[StefanW]

Hi Guenter,

that was as well our solution generating one contour by hand. I wanted to report this obvious bug, that it can be fixed in the future. For IMAQ ColorLearn it works correctly, so it shouldn't be a big thing.

Greetings, Stefan

[Elmar_NI]

Hi all,

i created a car for this behaviour with the number "CAR 4E87ONAQ".


Kind regards,

Elmar