[FIXED] pyTesseract recognize a pattern of text

Issue

I'm trying to do a simple license plate recognizer. Currently my problem comes from Tesseract messing some readings (for example 5 as S). I know the images are always going to be three uppercase characters, followed by three digits, in the form AAA 999 or so. Is there any way I can give this info to the OCR?

Solution

Tesseract allows to whitelist specific characters using the tessedit_char_whitelist parameter.

A way to address your license plate identification problem would be to split your detection window in two "subwindows", and:

• whitelist letters for the first subwindow (tessedit_char_whitelist=ABCDEFGHIJKLMNOPQRSTUVWXYZ)
• whitelist numbers for the second subwindow (tessedit_char_whitelist=0123456789)

Answered By - Sheldon