基于倒谱集成学习的语音回声隐藏分析算法

回声信息隐藏是语音信息隐藏的一种常用方法,针对语音回声信息隐藏的分析问题,提出了一种基于倒谱集成学习的隐藏分析方法.分析了回声隐藏及集成学习的原理,构建了基于倒谱AdaBoost集成学习算法的回声隐藏分析系统.对倒谱特征、倒谱一阶差分、二阶差分、梅尔倒谱特征、直方图联合特征、高阶矩特征分别采用支持向量机和AdaBoos...

Echo hiding steganalysis method based on cepstrum ensemble learning

Echo information hiding is a commonly-used method of speech information hiding.Focusing on the analytical problem of speech echo hiding,a steganalysis method based on cepstrum ensemble learning was proposed.By analyzing the principle of the cepstrum and ensemble learning,the echo hiding steganalysis system based on AdaBoost ensemble learning algorithm was built.The features that contained cepstrum,the first-order and second-order cepstrum,the mel-cepstrum,combined histogram and moments of high-order statistics was simulated by using SVM and AdaBoost ensemble learning algorithm.Simulation results show that the classification effects of the ensemble learning are better than the SVM for the cepstrum,the first-order and second-order cepstrum and combined histogram features,especially the accuracy rate of cepstrum detection reaches to 91.9 %.