You could parse the HTML that is uploaded either manually or using an XML validator (although good luck handling poorly formatted HTML). Its a good idea especially if you want to make sure people don't upload malicious code, as an example, such parsers usually remove all but the most basic HTML tags like bold/italic/etc. One its parsed, finding the img tag is then trivial.
Another method is just to convert the entire html file into a single
string and try using regular expressions to match an image tag.
Really for uploading files though, I recommend a form upload button to explicitly upload the files. There's no telling how many files or how big they will be if you start uploading from external sites.
[ August 16, 2006: Message edited by: Scott Selikoff ]