A Simple Method for Filtering Image Spam

Abstract

Image spam can easily circumvent widely text spam filters. It is a new trend of spam, and poses a great threat to email communications. In this paper, we give the definition of image spam firstly. Then we propose a simple method for filtering image spam, which utilizes file properties and histogram (gray histogram or color histogram). Finally, a preliminary experimental evaluation of our approach is reported on a personal dataset and Princeton benchmark dataset of spam images publicly available.