We read the image with an OCR engine and lay an invisible text layer over it. The page looks identical, but now you can search it, copy from it — and translate it.