09-12-2011 01:21 AM
Good day again NI Experts,
I would like to seek help about improving the VBAI's OCR character recognition. In my application, some dust/dirt manages to slip into an IC where I need to get the data through OCR. That's when my problem occurs. In one of my example where texts read "706MMG122", a dirt slips between "1" and "2" and touches "1" (please refer to the image). When I apply an OCR ROI, OCR can't separate "1" from "2" although "2" is separated from "1" and the dirt, as shown in "OCR_ROI_1,2_mod2".
In the next example where texts read "DCTQD", a dirt slips between and above "D" and "C". When I apply an OCR ROI, OCR can't separate the dirt from "C" although the dust is far above it. For some reason, I can't minimize the size of the ROI. Please advise me on how to deal with these dot-like dusts and its effects.
Thank you very much,
Andrei K.
09-12-2011 03:38 AM
DCTQD's OCR ROI
09-12-2011 08:17 AM
Something that might help is to specify the min width, height, and size for characters to help keep characters from getting joined together. See the attached screenshot for a sample. To avoid having the 1 and 2 combined, the min width also prevented to the full M from being read, but I would bet with the amount of the M being used, it would still work alright.
Hope this helps,
Brad