How to detect font type[bold, italic] and font color while scanning text through OCR?


#1

I want to implement the feature in iOS to get Questions and answers from the image taken by a camera from the phone.

To exemplify, http://prntscr.com/kvnbca (see below). I need to fetch questions and answers from that. And especially the main focus is to detect the bold answer or answer with red color. So I would like to know is there any way to detect the bold or red color fonts from that through OCR while scanning and fetching text?

download


#2

Detecting font information (bold, italic, cursive, font name) is a popular feature request, and we fully understand that. Unfortunately our solution can not detect bold as an attribute to a word yet. So the word will be OCR’ed ok, but the “bold” attribute/information is lost. And to my knowledge, no current OCR solution in the market can detect a bold/italic attribute or even the font name.

However, things look better with finding the red answer. If your text is like in your example, just with a red answer, you can pre-process the image before feeding it into the OCR API. During pre-processing remove all black (and dark grey?) colors from the image, so only the red text remains.

For removing the black color, have a look at the free command line tool imagemagick. Here is a related StackOverflow answer that shows what I mean: Remove all except one colour from an image


#3

Thank you for your brief response.