Hi I have a stability matter with my Visual inspection system. In the OCR control I have to read a string printed on a label. Characters are often stuck, then the Autoslit function does not work. I developed a new algorithm to separate characters and we can assume that it works.
In order to improve the stability and the confidence of the reading I used the following system (see picture):
- For each iteration of the For cycle I set the text pattern (Set Valid Characters). It can be Decimal Digit o Uppercase Letter with the expected character.
- For each iteration of the For cycle, I pass the ROI for each character to the IMAQ OCR Read Text 3.
In spite of all these solutions I still have several errors, as you can see in the pictures.
Do I using in wrong way the text pattern? Because, I extract the ROI for a"3", I suggest a 3 as a pattern, but it reads "8". Ok, they are similar I cannot believe that a "3" matches more with an "8" than other "3".
Daos anyone know how the text pattern algorithm work? Because it seem to return not the char which has the higher score, but the first one has a score enough high.
More over, if the character that has been red is not the one suggested in the text pattern I will expect the "?" and not another character.
Thanks a lot for your patience.
