基于分段STC的音频隐写算法  

An audio steganography algorithm based on segment-STC

作　　者：张雪垣 王让定[1] 严迪群[1] 林昱臻 ZHANG Xueyuan;WANG Rangding;YAN Diqun;LIN Yuzhen(College of Information Science and Engineering,Ningbo University,Ningbo 315211,China)

机构地区：[1]宁波大学信息科学与工程学院

出　　处：《电信科学》2019年第7期115-123,共9页Telecommunications Science

基　　金：国家自然科学基金资助项目(No.U1736215,No.61672302);浙江省自然科学基金资助项目(No.LZ15F020002,No.LY17F020010);宁波市自然科学基金资助项目(No.2017A610123);宁波大学学科基金资助项目(No.XKXL1509,No.XKXL1503);浙江省移动网应用技术重点实验室开放基金资助项目(No.F2018001)~~

摘　　要：在隐写领域中,STC(syndrome-trellis code)隐写编码是时下最为先进的隐写算法之一。提出了一种基于STC的隐写算法--分段STC。该算法利用分段、局部优化,可有效降低载体失真。首先对音频最低有效位(least significant byte,LSB)载体流进行了分段切割,随后利用最佳子校验矩阵将分段的密信嵌入分段载体中。该隐写算法在现有的语音样本库TIMIT库以及通过网络歌曲自制的样本库上进行了隐写实验。实验结果表明,该思路下的算法可以有效地降低载体的失真元素个数和提高隐写算法的不可检测性。本文算法相较于传统STC算法,失真元素个数的优化程度最高可达到24%;对比同类音频隐写算法,本文算法生成的隐写后音频具备更加优秀的不可感知性能。In steganography field, syndrome-trellis code is a state-of-the-art steganography algorithm. A steganography algorithm of Segment-STC was proposed. Segment-STC could locally optimize the distortion by segmentation.Firstly, the LSB-stream(least significant bit, LSB) of audio cover was segmented as few sub-segments. Then, the segmented message in each sub-segment was embedded to the optimal submatrix. The TIMIT dataset and author collected online songs as databases was used to run the experiments. Results show that segment-STC can effectively reduce the number of distortion elements of cover and improve the undetectability of the stego. Segment-STC algorithm can reduce the distortion over 24%, compared with the original STC algorithm. It also has a better imperceptible performance compare with other similar audio steganography algorithms.

关 键 词：分段STC 载体密信分配 最佳子校验矩阵 WAV隐写 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]