基于回声隐藏改变自然语言倒谱分布这一特点,提出了一种新的针对回声隐藏的隐写分析算法。该算法无需估计隐写分段大小,通过构造倒谱和差分方差统计量VDSC(variants of difference of sum of cepstrum)来反映隐写对自然语音音频的影响,从而判定该文件是否为回声隐藏文件。实验结果显示,该算法正确率为88%。
Dynamical running coding was previously proposed as a method for construction of steganographic schemes in digital steganography.In this article,a new approach of fast coding is proposed to improve steganographic efficiency.Original embedding method is discussed in two cases of constructing different generating matrix;however,the proposed method generalizes the problem to only one case by changing the construction method from the original method.Therefore,the complexity of the proposed method is less than the original method.Meanwhile,both theoretical analysis and experiment results show better performance in terms of distortions and alterations of the cover objects when embedding the same amount of information data bits,which proves a better resistance against steganalysis.
