https://www.google.com/search?as_sitesearch=processing.org&as_q=ocr
Nothing specific to Processing, AFAIK.

yes, this is reason why I want open this topic again. Don't tell me that nobody was solving this since that time.

In the old forum (see the search), I wrote that making a good OCR is hard, and there is nearly no open source engine, except Google's Tesseract. Which is native code, so would need JNI / JNA to use it from Java. Perhaps there is already a wrapper.
My statement is still valid, I think...