OCR algorithms typically aren’t targeting heavily blurred text. It’s more about handling all the ways letters can look, including ligatures, determining paragraph breaks, detecting tables, ignoring staples and coffee stains, etc. than about correcting for bad scans/photos.