Title :
Image spam — ASCII to the rescue!
Author :
Nielson, Jordan ; Aycock, John ; De Castro, Daniel Medeiros Nunes
Author_Institution :
Dept. of Comput. Sci., Univ. of Calgary, Calgary, AB
Abstract :
We take an unorthodox approach to image spam detection, by applying existing software and decades-old technology: ASCII art. Our technique is straightforward and gets good levels of detection over a corpus with 1159 ham and 1492 spam images, with a tolerable amount of misclassifications. Furthermore, we only look at the images themselves, meaning that this method can be trivially enhanced by combining it with existing anti-spam techniques.
Keywords :
image classification; security of data; unsolicited e-mail; ASCII; antispam techniques; image spam detection; Art; Bayesian methods; Character recognition; Computer science; Drives; Image converters; Mathematics; Optical character recognition software; Optical filters; Unsolicited electronic mail;
Conference_Titel :
Malicious and Unwanted Software, 2008. MALWARE 2008. 3rd International Conference on
Conference_Location :
Fairfax, VI
Print_ISBN :
978-1-4244-3288-2
Electronic_ISBN :
978-1-4244-3289-9
DOI :
10.1109/MALWARE.2008.4690859
