On #2: in general, only bitmap images can reliably be extracted from a PDF. Not always, though: not every bitmap is "an image". Consider, for example, a bitmapped font or that nasty Word sub-function that inserts Symbol characters as 8x8 images. And sometimes, bitmap images are used as fills for vector objects.
Acrobat Pro provides (provided?) a command "Extract all images" that asks for a minimum size to prevent lots of irrelevant little files being created.
Vector images can only be extracted under very specific circumstances. In most cases, the vector data is embedded on a page together with "regular" page content, and so there is no real difference between lines that form an image and lines that draw an underline under some plain text.