I am wondering if there is a way to split a scanned copy of a page (i.e. an image) into images of its constituent words. The book which is being scanned is in an Indian language for which OCR (Optical Character Recognition) would not work.
The goal is to have a digital version of the book (in the same language), and for that, I was thinking of doing the following -
- scan the book
- split the contents of a page into its constituent words
- send the images of the words to transcribers who can type the word in the image
- reassemble the words in each page
Since image to image matching is reasonably good, images of the same words can be identified, and thus wouldn't have to be transcribed repeatedly.
Any help would be great!