I have a situation where I found a TON of scanned PDF files. These PDFs are effectively thin wrappers over a wasteful image format. Each file is several MB in size. The total size of the archive is over 1TB. I know there was no OCR work done... the PDFs only have the images, with not text layer or search capability.
I'd like to reduce the size of this archive. I know that if I had the original images, they would compress easily from 3-6MB each down to a few hundred KB each without noticeable loss of quality... a 10 times reduction in size. What I don't know is how to get back from these PDFs to the original image, how to recreate a PDF file from the reduced image (the user really likes PDF for some reason), and how to do all this in a reasonably automated way.
As an additional wrinkle, there are a few rare cases where some newer PDFs in the archive came from a separate and may have a searchable text layer that we don't want to lose.
Any ideas on how I can efficiently process these files?