Survey on image-based spam filtering
图像型垃圾邮件过滤技术综述

This paper analyzed the difficulties of detecting image-based spam with the features of images in detail.The features of spam images were divided into eight categories such as file attribute features,image metadata and so on.Then,discussed and compared five classification algorithms which have been used in image-based spam filtering were outlined,including support vector machines,decision tree method,maximum entropy model,the DS evidence theory,Bayesian algorithm and the effect of these algorithms.Finally,gave some future directions of research on the techniques of image-based spam filtering.