Without being drawn into a long protracted argument about this (this is my last post on this topic), the simple act of scanning the formatted page pre-OCR infringes the copyright of the publisher, if the publishers rights have not yet expired.
You can't OCR from nothing, at some point you need a direct copy, be it in memory or on disk, of thing being copied.
You can't OCR from nothing, at some point you need a direct copy, be it in memory or on disk, of thing being copied.