It’s probably because the OCR isn’t scanning the signatures as text. They’re being scanned as a separate image layer on top of the background, while the OCR is picking up most of the rest of the text as searchable.
Are you suggesting that OCR picked up some, but not all, of the signatures? Why would this make any sense? Furthermore, why would OCR fail on some of the most obvious letters (such as the H, the L, the K, and the S)? OCR is imperfect, but it doesn’t work the way you think it does.
Thanks for that. Hey! Just checked your bio and it indicates you had a little “Freedom Fighter” recently. Big congrats!