I have been using CAM::PDF and also "pdftohtml",
and I would like to suggest the latter for your task.
As Chris mentioned, "you'd be doing a lot of heuristic coding", but according to my experience you are better off with the output of "pdftohtml -xml", supposing you are fine with xml parsing.
I used XML::Simple for that purpose, but maybe you are more experienced.
There is no such concept as tables in PDF, and you would have to do a lot of guessing (-> heuristics) as to what are columns and rows and cells on these pages.
If you are really progressing on this, let us now!
Me personally, I am quite interested in that kind of utility as well, maybe I am going to work on it as well.