Here's an idea for blocking image spam: What about taking the idea of SURBL and DNSRBls and extending it to images. My proposal is to hash the image and do a DNS query using the hash value and domain hosting the image RBL.
This eliminates the need to OCR the graphic, and if they obscure the text, it doesn't matter. If they send the same image to many recipients, it'll get detected blocked. Maybe combine this with OCR, doing the hash/lookup first since it should be low cpu cost (but slow on network lookup unless it's already cached). Odds of a false positive would depend on hash length which would really only be limited by the RFCs for DNS. How long can any element of a DNS name be? Total length of a fully qualified name? A collegue suggested calling such a system Gerbil, taking the sound of the last 5 letter of "image RBL". --- "Solving the spam problem is like curing cancer. It's not one disease but 100 diseases, each with their own issues." -- John Levine co-chairman of the independent Antispam Research Group, part of the Internet Engineering Task Force. William Brown Web Development & Messaging Services Technology Services, WNYRIC, Erie 1 BOCES (716)821-7285 _______________________________________________ NOTE: If there is a disclaimer or other legal boilerplate in the above message, it is NULL AND VOID. You may ignore it. Visit http://www.mimedefang.org and http://www.roaringpenguin.com MIMEDefang mailing list MIMEDefang@lists.roaringpenguin.com http://lists.roaringpenguin.com/mailman/listinfo/mimedefang