One of my friends has a company in Iran and they sell security cameras to the shops, banks, residential buildings and so on. The quality of captured images of these cameras is usually low and they need to use software to analyse these images to be able to recognise the Persian letters and digits on the plates. However, the software in the market cannot recognise Iranian letters and digits on the cars’ plate. Thus, can anybody help them by introducing specific software that they can buy? Or, is there anybody to program this kind of software for them?
Hi, so your friend may have already done some research on google and ended up posting on google groups. I've personally used Tesseract-OCR with English and as you may read in the link that there kind-of is Persian language support but not entirely there.
An option to explore is to use Tesseract and build their own training data for persian language based on whatever fonts is used on Iranian license plates. It's what I'd do anyways if I had funding to go down that route.
https://groups.google.com/forum/#!topic/tesseract-ocr/dUR0bSbYIYM
You could, potentially, build a machine learning system that could do this for you.
http://ieeexplore.ieee.org/xpls/abs_all.jsp?arnumber=1699586
Hi Mehdi,
I would agree with George. Tesseract-OCR would be a viable option for doing the OCR part. However, you would probably need to build the pre-processing components your self for isolating the licence plate and letters/numbers.
I have previously used Tesseract-OCR successfully for a new script (Thaana) and you can find the details in the following article.
Regards
Ahmed
Article Dhivehi OCR: Character Recognition of Thaana Script using Ma...
- Badges
- Science topic