I would think that if you can use edge detection (sobel/laplacian) should be fine (DoG too), you would have the outline of the characters. Then it's a matter of detecting the closed paths and filling them. If you account for the radial (get it!?) nature of the text alignment, then bucket fill the background, white, and when you cross a gradient, fill the other side with black (if not already filled with white) you should be able to fill the colors accordingly. Lorenzo also recently posted this link that might help: https://www.quora.com/How-do-I-fill-holes-in-image-using-image-processing/answer/V-Sri-Chakra-Kumar
You biggest signal is the gradient around the letters. Use it. On Monday, May 20, 2019 at 2:20:44 PM UTC-4, David Bess wrote: > > Hi All, > > I am facing challenges reading DOT from a tire. As a person, I can clear > make it out with no difficulty. See attached image. In this example the > DOT is PJ40 KU1R 2011. I have already tried inverting, and binarisation, > but I just do not get any output. Did a lot of research online, and it > seems like a changeling problem, so I thought I would check with the smart > folks here on the forum. Thanks in advance for any help or advice. > > Best regards, > > David Bess > -- You received this message because you are subscribed to the Google Groups "tesseract-ocr" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at https://groups.google.com/group/tesseract-ocr. To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/a562c2bc-f969-46ed-a38b-971697af9698%40googlegroups.com. For more options, visit https://groups.google.com/d/optout.
