You may want to search the forums there are many documents that deal with Optical Character Recognition (OCR) and training. If you are planning on using the Vision addon I would take a look here: http://sine.ni.com/devzone/cda/epd/p/id/547
Hi prashantpatel21, As ShotSimon recommended, we have the NI Vision Development Module (VDM). VDM contains lots of VIs and tools for image processing and machine vision applications, including OCR. It has an OCR training application found in Start » Programs » National Instruments » Vision » OCR Training. If you want to perform a similar task in LabVIEW using the VIs from the VDM, you could use the OCR Train VI to do this. For more info on the theory behind our software, the NI Vision Concepts Manual has an entire chapter (Chapter 18) on OCR implementation using VDM.
The OCR Train interface only expects 1 character at a time. You are
giving it an index of 0 and only calling it once even if you have
multiple characters selected in your ROI, so only the first character
(such as the U) is getting added to the character list.. I think you
need to put OCR Train within a for loop instead and have it loop the
same number of times as threshold-detected characters, but I haven't
figured out a way to determine how many characters (blocks) the OCR
Train VI is actually detecting.
I believe the issue is that you are deleting the character, but you are not writing the set back to file within the case structure before reading the set from file again. If you look at the case where you add a character to the set, you see that you write it out again before you leave the case structure and read it again. I've attached a diagram to show what I mean, although I don't know exactly how you're handling your index control anymore since you got your character insertion code fixed.
When you get that working, would you mind posting your code again for
the benefit of others? The code you've developed so far is pretty cool.
Message Edited by VGA CD-ROM on 05-22-200810:38 AM