It is often not possible to mirror only images, because HTTrack must follow links on the pages (HTML) to find all the images you want.
The good method is to crawl a site with default, or general, rules (stricter rules may be better, but very dependant on the actual site) and use a Build Structure (see
Options > Build tab) like:
Html in web/, images/other in web/xxx, where xxx is the file extension
This way, when HTTrack saves the files to your hard drive, the images are separated into their own folder.
Also ensure that you have appropriate image filters (
Options > Scan rules). The following will accept any images found even if they are linked from other servers:
+*.gif +*.jpg +*.png +*.bmp