Do you have a flatbed scanner or multi-function machine that came with OCR software (and most have in recent years)?
If so, and you are able to select those graphics, I'd simply select them, paste each on a single page, print them and scan them using the scanner and OCR. So long as you have a system for knowing which belongs where (which could include a single additional annotation) that could work.
That or just copying the image, creating a separate word file including an annotation regarding source, pasting it in and using the screen reader OCR on that second document which would just contain pages of the images to make it easy for the OCR software to get only what it needs.
Brian - Windows 10 Pro, 64-Bit, Version 1809, Build 17763
Puritanism: The haunting fear that someone, somewhere, may be happy.
~ H.L. Mencken