结合字典序和排序线性拟合的音频隐藏方法

发布时间：2018-03-30 16:35

本文选题：音频信息隐藏　切入点：秘密音频　出处：《小型微型计算机系统》2017年12期

【摘要】：传统音频隐藏方法通常是将秘密音频或信息直接嵌入到待隐藏的载体中,由此带来了较大的信息嵌入量也会对嵌入载体造成较大影响,导致嵌密载体听觉质量下降和易于受到攻击.针对此问题,提出一种结合字典序和排序线性拟合的音频隐藏方法.在嵌入时,首先将秘密音频和公开音频划分为不等长小段序列并一一对应;然后取与秘密音频小段序列等数量的公开音频小段序列起始元素和秘密音频小段序列进行排序线性拟合并通过字典序记录秘密音频排列位置索引以及对应的拟合变换参数;最后将位置索引和拟合参数进行q进制存储并通过改进EMD方法嵌入到公开音频小段序列中.在提取时,首先从公开音频小段序列中提取出位置索引和拟合参数,然后将与秘密音频小段序列等数量的公开音频小段序列起始元素恢复为秘密音频小段序列并重构秘密音频.理论和实验表明,同现有方法相比,采用排序拟合使得信道中公开传输音频具有较高信噪比,同时仅对字典序、q进制表示的变换参数按改进EMD方法进行嵌入,在减少对载体音频修改的同时也降低了秘密音频被泄露的风险且恢复出的秘密音频拥有一定的可辨识听觉质量.
[Abstract]:The traditional audio hiding methods usually embed secret audio or information directly into the carrier to be hidden, which brings a large amount of information embedding and has a great impact on the embedded carrier. As a result of the deterioration of audio quality and the vulnerability to attack, a novel audio hiding method combining dictionary order and sorting linear fitting is proposed. Firstly, secret audio and public audio are divided into unequal short segment sequences and one-to-one correspondence. Then, the starting elements of the public audio sequence and the secret audio sequence are sequenced and merged to record the location index of the secret audio sequence and the corresponding fitting transformation parameters through dictionary order. Finally, the location index and fitting parameters are stored in q-ary and embedded in the open audio sequence by the improved EMD method. Firstly, the location index and fitting parameters are extracted from the open audio sequence. Then, the starting elements of the open audio sequence, such as the number of secret audio sequences, are restored to the secret audio sequences and the secret audio sequences are reconstructed. Theoretical and experimental results show that, compared with the existing methods, By using sort fitting, the audio transmission in the channel has a higher SNR, and only the transformation parameters of the dictionary order Q ary representation are embedded according to the improved EMD method. While reducing the modification of carrier audio, it also reduces the risk of secret audio being leaked, and the recovered secret audio has a certain recognizable auditory quality.
【作者单位】：陕西师范大学计算机科学学院;
【基金】：国家自然科学基金项目(61100239)资助陕西省自然科学基金项目(2011JQ8009,2016JM6065)资助中央高校基本科研业务费支持项目(GK201402036,GK201703057)资助
【分类号】：TP309

【相似文献】