PDF files from scan are much larger than the images files therein

I previously complained about the exorbitant file size of documents scanned with the Xerox WorkCenter 7855.  I'll paraphrase the explanation that was provided: The file size depends on the entropy of the information in the image rather than the pixel density (http://forum.support.xerox.com/t5/Copying-Faxing-Scanning/72-dpi-grayscale-almost-as-big-as-600-dpi-colour/td-p/214506)

Inefficient scanning is a big issue for me, as I am trying to go paperless in a much smaller office space.  I eventually took the time to track down the ongoing problem of large file size.  I did two trials of scanning of a two-page document, one trial at 75dpi and another at 100dpi, both in grayscale.  They were both 350KB, which seemed awfully suspicious.  From a websearch, I found that I could use Adobe Pro to extract the image files contained within the PDF files.  I'm not in the office right now, but from what I can recall, each page was in the low umpteen KBs, if not below 10KB, and lower pixel density did in fact yield smaller file sizes.

Xerox's 350KB seems to be a minimum PDF container size, regardless of the images contained therein.  That's more than an order of magnitude bigger the actual scanned image files therein.  I could use analytical software to further decrease the bits per pixel so that the files were *well* below 10KB without out noticable degradation in readability.  You don't need many levels of grayscale for scanned hand notes.

The scanning capability is of limited utility for going paperless if it creates such unnecessarily exorbitant file sizes.  Users need to be able to scan documents, often times notes of a few pages, without a second thought.  In an operational setting, it is impractical (and often impossible) to shuttle documents to other networks where analysis tools can be brought to bear to recreate the scanned documents with the correct sizes.  I'm hoping that there is actually a solution to this?

As well, is there a way to unmark the response as the answer in the thread linked to above?