摘要： 通过对自然文本统计模型和特性的分析，指出隐藏消息后可能对文本统计特性带来的变化，提出基于支持向量机的通用检测算法。将文本的5个基本统计特征量作为分类特征向量，采用支持向量机对自然文本和载密文本进行有效分类检测。实验结果证明，该算法具有较好的适用性和可靠性。

关键词: 隐写分析, 文本, 支持向量机

Abstract: This paper studies the statistical models of natural texts, and points out that embedding messages in texts may change the features of them by analyzing the statistical features. Based on these changing, a blind detecting method is designed using Support Vector Machine(SVM). It extracts five basic parameters of texts as distinguished feature vectors to discriminate natural texts and stego-texts effectively using SVM. Experimental results show the high accuracy and reliability of the method.

Key words: steganalysis, text, Support Vector Machine(SVM)

中图分类号: 

• TP391