Revision history [back]

I am beginner my self, this can totally be OCR'ed with some processing to gather the text out of the image, or if the language you are working with isn't supported you can use something like DB clustering provided with correct text representations it would effectively stitch the whole thing according to the pixels from the images provided! try this link : https://www.naftaliharris.com/blog/visualizing-dbscan-clustering/

try smiley in it with default settings you'll see how it stitches the whole things with scattered pixels, we can use this to get definite text patterns out of this thing then we can use some image manipulation techniques to colour the bound area. pardon me if this is a bad approach, still learning :)