Certainly an OCR interface utility would solve the problem. A GOOD OCR package is neither simple nor cheap.

Nowadays, OCR is just part of the OS. Apple's OSs (desktop, tablet, and phone), for example, offer transparent OCR and pull text from any photo you are viewing as a matter of course. You can just select the text on a photo and it will copy to the clipboard.

You can also translate text in place, replacing the text in the photo with the new language.

Things have changed.

 

Evidently things changed a great deal. My experience was advising a friend about 20 years ago. And an OCR program that would be able to read the fonts in scans of paper and magazine articles was quite something.
I have wondered about the next thing, which would be the opposite: Speech to text. I see quite a few technical presentations with a speaker talking rapidly and nonstop, delivering an hour long presentation in 20 minutes. That makes the whole thing a waste of time, when it is a detailed tech topic that needs to be understood. What is the motivation for such speed?? and is there speech recognition actually able to translate that fast??