基于随机蕨的集成分类图像隐写分析算法

An image steganalysis algorithm based on random ferns and ensemble learning

提出了一种基于随机蕨(random ferns)和集成学习的图像隐写分析算法。首先利用图像高维特征构建蕨特征,采用成对采样策略构造样本子集,生成若干个基分类器;然后计算出训练样本在基分类器中各个蕨的先验概率并集成各个基分类器,进行隐写检测判别。实验结果表明,本文算法复杂度低,能有效降低隐写检测错误率。

A novel image steganalysis algorithm based on random ferns and ensemb le learning is proposed.At first,the random ferns are constructed with high-dimensional steganalysis features,and th e sample subsets are built based on paired sampling stategy,to generate some base classifiers.Then the prior probability of f erns of traning samplings in base classifiers is calculated,which is ensemble to obtain the final results.The experimental r esults show that the proposed algorithm has low complexity,and can effectively r educe the detection error rate.