HTML image only spam and OCR
ka at pacific.net
Fri Mar 10 23:05:48 GMT 2006
Why not use a checksum of the image attached, assuming the spammers
don't customize images for each recipient, you should be able to use
DCC, razor, pyzor type approach to block these if you just look at the
.gif attachments separate from the bayes poison. You'd probably FP on
some commonly used 'stationary' if you aren't careful though. The
MailScanner custom scanner interface is an ideal place to plug in such a
> On 3/9/06, Ian <cobalt-users1 at fishnet.co.uk> wrote:
>> After reading this bit I had though about maybe using ocr when these types of messages are
>> A (not-so) quick experiment using netpbm and gocr on a linux machine here produces some
>> ASCII output from one of these gif images.
>> The question is: how can I get MailScanner / SpamAssassin to use this method?
>> The command line I am using is:
>> giftopnm test.gif | gocr -
>> which then produces the text on stdout.
>> Thoughts anyone?
> MS supports both a custom spam scanner and a generic virus scanner.
> Look in MailScanner.conf for more info.
More information about the MailScanner