General Category > General Discussion
OCR Software
<< < (2/2)
Starbuck777:

--- Quote from: sheep2001 on May 30, 2015, 11:57:09 AM ---Just google the first line in each language.  8 out of 10 times its already available.  Oce is going to be ready hit and miss on those small fonts, especially with stuff in the background.

--- End quote ---

OCR
     This brings up a good point; what best practices does the cover designers use?
          I want to be able to have box text (be text) on it's own layer...
               I read a post that the person would "extract/build" each letter - Vector-ize each letter..
The two problems for me would be "FONTS" and "FOREIGN LANGUAGES, Symbols" I think vector-izing the fonts could be good along as the vector font matches the original source. With foreign languages/symbols it could be similar unless there are font packs available to match the original source.

OCR - the smaller the text the more difficult like "sheep2001" mentioned in his post

So what is the best practice when dealing with these issues?

Thanks Again
wiggy:
I’ve translated several manuals using this method. There are a handful of good websites that’ll do it for ya (JPG-to-TXT).

I’ve yet to find type that was too small to be detected. It’s just a matter of having clean source material.
Starbuck777:
Thanks Wiggy & Sheep2001 for the suggestions, I'll try them out.
wiggy:
Sure thing.


I get the best results with this site.

https://www.onlineocr.net
Navigation
Message Index
Previous page

Go to full version