this bug has been reported here:

https://bugs.edge.launchpad.net/poppler/+bug/320006

"The command pdfimages in poppler-utils 10.0.3-0ubuntu1 (9.04 Jaunty) still ignores the option "-j" to extract ".jpg". It always extracts ".ppm". The same applies to poppler-utils 0.8.7-1ubuntu0.1 (8.10 Intrepid)."

"I deal with PDFs that are produced by gscan2pdf 0.9.27 (9.04 Jaunty) which contain one page-filling JPG per page. I attach here a Testscan.pdf of one empty paper. (BTW: I use i386 packages)

With xpdf-utils 3.02-1.4ubuntu1, I get what I expect:
$ pdfimages -j Testscan.pdf xpdf-image
$ file xpdf-image-000.jpg
xpdf-image-000.jpg: JPEG image data, JFIF standard 1.01, comment: " SANE data follows"
$ ls -l xpdf-image-000.jpg
-rw-r--r-- 1 oliver oliver 16634 2009-01-23 09:52 xpdf-image-000.jpg

With poppler-utils 0.10.3-0ubuntu1, I get:
$ pdfimages -j Testscan.pdf poppler-image
$ file poppler-image-000.ppm
poppler-image-000.ppm: Netpbm PPM "rawbits" image data
$ ls -l poppler-image-000.ppm
-rw-r--r-- 1 oliver oliver 6524897 2009-01-23 09:53 poppler-image-000.ppm"

Pdf file:

http://launchpadlibrarian.net/21557296/Testscan.pdf

Thanks you,