Detection method for echo hiding based on convolutional neural network framework

doi:10.11772/j.issn.1001-9081.2019081400

Journal of Computer Applications ›› 2020, Vol. 40 ›› Issue (2): 375-380.DOI: 10.11772/j.issn.1001-9081.2019081400

• DPCS 2019 • Previous Articles Next Articles

Detection method for echo hiding based on convolutional neural network framework

Jie WANG, Rangding WANG(), Diqun YAN, Yuzhen LIN

Faculty of Electical Engineering and Computer Science，Ningbo University，Ningbo Zhejiang 315211，China

Received:2019-07-31 Revised:2019-08-28 Accepted:2019-09-19 Online:2019-10-14 Published:2020-02-10
Contact: Rangding WANG
About author:WANG Jie， born in 1996， M. S. candidate. His research interests include multimedia communication， information security.
YAN Diqun， born in 1979， Ph. D.， associate professor. His research interests include multimedia communication， information security.
LIN Yuzhen， born in 1994， M. S. candidate. His research interests include multimedia communication， information security.
Supported by:
the National Natural Science Foundation of China(U1736215);the Natural Science Foundation of Zhejiang Province(LZ15F020002);the Natural Science Foundation of Ningbo(2017A610123);the Subject Foundation of Ningbo University(XKXL1509);the Open Foundation of the Mobile Network Application Technology Key Laboratory of Zhejiang Province(F2018001)

基于卷积神经网络框架的回声隐藏检测方法

王杰, 王让定(), 严迪群, 林昱臻

宁波大学信息科学与工程学院，浙江宁波 315211

通讯作者: 王让定
作者简介:王杰（1996—），男，浙江嘉兴人，硕士研究生，主要研究方向：多媒体通信、信息安全
严迪群（1979—），男，浙江余姚人，副教授，博士，CCF会员，主要研究方向：多媒体通信、信息安全
林昱臻（1994—），男，浙江宁波人，硕士研究生，主要研究方向：多媒体通信、信息安全。
基金资助:
国家自然科学基金资助项目(U1736215);浙江省自然科学基金资助项目(LZ15F020002);宁波市自然科学基金资助项目(2017A610123);宁波大学学科基金资助项目(XKXL1509);浙江省移动网应用技术重点实验室开放基金资助项目(F2018001)

Abstract

Abstract:

Echo hiding is a steganographic technique with audio as carrier. Currently， the steganalysis methods for echo hiding mainly use the cepstral coefficients as handcrafted-features to realize classification. However， when the echo amplitude is low， the detection performance of these traditional methods is not high. Aiming at the low echo amplitude condition， a steganalysis method for echo hiding based on Convolutional Neural Network （CNN） was proposed. Firstly， Short-Time Fourier Transform （STFT） was used to extract the amplitude spectrum coefficient matrix as the shallow feature. Secondly， the deep feature was extracted by the designed CNN framework from the shallow feature. The network framework consisted of four convolutional blocks and three fully connected layers. Finally， the classification results were output by Softmax. The proposed method was steganographically evaluated on three classic echo hiding algorithms. Experimental results indicate that the detection rates of the proposed method under low echo amplitude are 98.62%， 98.53% and 93.20% respectively. Compared with the existing traditional handcrafted-features based methods and deep learning based methods， the proposed method has the detection performance improved by more than 10%.

Key words: echo hiding, steganalysis, Convolutional Neural Network (CNN), Short-Time Fourier Transform (STFT), deep learning

摘要：

回声隐藏是一种以音频为载体的隐写技术，目前针对回声隐藏的隐写分析方法主要以倒谱系数作为手工特征进行分类。然而，这些传统方法普遍在回声幅度较低时检测性能不高。针对回声幅度较低的情况，提出一种基于卷积神经网络（CNN）的回声隐藏隐写分析方法。首先利用短时傅里叶变换（STFT）提取音频的幅度谱系数矩阵作为浅层特征，然后设计了一个卷积神经网络框架对浅层特征进行进一步的深度特征提取，网络框架中包含了四个卷积模块以及三层全连接层，最后分类结果以Softmax进行输出。在三种经典的回声隐藏算法上对提出的方法进行了隐写分析实验评估，实验结果表明，该方法在低回声幅度条件下的检测率分别为98.62%、98.53%和93.20%，与目前所提出的传统基于手工特征的方法和基于深度学习的方法相比，检测性能提升10%以上。

关键词: 回声隐藏, 隐写分析, 卷积神经网络, 短时傅里叶变换, 深度学习

CLC Number:

TP391

Jie WANG, Rangding WANG, Diqun YAN, Yuzhen LIN. Detection method for echo hiding based on convolutional neural network framework[J]. Journal of Computer Applications, 2020, 40(2): 375-380.

王杰, 王让定, 严迪群, 林昱臻. 基于卷积神经网络框架的回声隐藏检测方法[J]. 《计算机应用》唯一官方网站, 2020, 40(2): 375-380.

Figures/Tables 12

Fig. 1 Schematic diagram of echo hiding with traditional kernel

Fig. 2 Schematic diagram of echo hiding with positive and negative kernels

Fig. 3 Schematic diagram of echo hiding with backward and forward kernels

Fig. 4 Average difference of audio frequency points between original audio and three kernels

Fig. 5 Proposed CNN framework

Fig. 6 Flowchart of echo hiding steganalysis method based on CNN

Tab. 1 Parameter setting of echo kernels

128或256或512

（随机选取）

α

=0.3，0.4，0.5

d 0

d 1

=21/23或

d 0

d 1

=20/25或

d 0

d 1

=30/40（随机选取）

α 1

α 2

=0.1/0.3，0.2/0.4，0.3/0.5

d 1

d 2

=21/23和

d 1

d 2

=20/25

α

=0.1，0.2，0.3

d 0

d 1

=21/23或

d 0

d 1

=20/25或

d 0

d 1

=30/40（随机选取）

Tab. 1 Parameter setting of echo kernels

回声核

分段长度

回声幅度

延迟时长

128或256或512

（随机选取）

α

=0.3，0.4，0.5

d 0

d 1

=21/23或

d 0

d 1

=20/25或

d 0

d 1

=30/40（随机选取）

α 1

α 2

=0.1/0.3，0.2/0.4，0.3/0.5

d 1

d 2

=21/23和

d 1

d 2

=20/25

α

=0.1，0.2，0.3

d 0

d 1

=21/23或

d 0

d 1

=20/25或

d 0

d 1

=30/40（随机选取）

Tab. 2 Accuracy comparison of three methods

回声核	回声幅度	准确率/%
回声核	回声幅度	本文方法	文献［7］方法	文献［9］方法
K1	0.3	98.62	86.00	90.35
	0.4	99.28	95.00	93.65
	0.5	99.45	98.00	94.12
K2	0.1/0.3	98.53	85.00	89.43
	0.2/0.4	99.47	94.00	96.63
	0.3/0.5	99.77	97.00	96.65
K3	0.1	93.20	78.00	79.22
	0.2	97.50	97.00	93.50
	0.3	98.72	100.00	97.83

Fig. 7 Training accuracy of K3 kernel with echo amplitude of 0.1

Tab. 3 Accuracy comparison of different network structures

回声核	回声幅度	准确率/%
回声核	回声幅度	本文框架	网络深度	激活函数	权重初始化	卷积核尺寸
K1	0.3	98.62	98.40	96.58	97.55	97.32
	0.4	99.28	98.99	97.76	98.65	98.81
	0.5	99.45	99.35	98.32	99.07	99.23
K2	0.1/0.3	98.53	98.08	98.41	97.23	97.98
	0.2/0.4	99.47	99.00	99.23	98.46	99.10
	0.3/0.5	99.46	99.77	99.59	99.08	99.57
K3	0.1	93.20	89.32	92.27	88.95	88.52
	0.2	97.50	96.77	97.18	94.83	95.56
	0.3	98.72	97.46	97.98	96.57	98.31

Fig. 8 Training accuracy of K3 kernel with 0.1 echo amplitude when increasing network depth

Tab. 4 Accuracy comparison of cross-over testing

测试回声核	回声幅度	准确率/%
		K1			K2			K3
		0.3	0.4	0.5	0.1/0.3	0.2/0.4	0.3/0.5	0.1	0.2	0.3
K1	0.3	—	96.22	94.92	73.08	63.08	58.58	93.63	85.75	77.63
	0.4	99.20	—	98.97	76.08	66.55	61.89	94.82	89.05	84.20
	0.5	99.30	99.30	—	76.75	68.45	63.67	95.13	89.90	87.13
K2	0.1/0.3	87.73	79.00	69.68	—	97.20	92.57	94.23	90.63	80.82
	0.2/0.4	88.22	79.43	71.50	99.48	—	98.95	94.35	91.85	84.95
	0.3/0.5	89.43	81.58	73.78	99.60	99.73	—	94.82	92.93	87.80
K3	0.1	72.95	63.90	55.77	64.70	56.88	53.73	—	77.35	62.42
	0.2	94.72	88.55	81.47	86.25	73.43	74.73	97.25	—	94.62
	0.3	97.95	96.10	95.02	95.28	86.70	76.37	97.72	99.40	—

References 16

1	BENDER W， GRUHL D， MORIMOTO N， et al. Techniques for data hiding［J］. IBM Systems Journal， 1996， 35（3/4）： 313-336. 10.1147/sj.353.0313
2	OH H O， SEOK J W， HONG J W， et al. New echo embedding technique for robust and imperceptible audio watermarking［C］// Proceedings of the 2001 IEEE International Conference on Acoustics， Speech， and Signal Processing. Piscataway： IEEE， 2001：1341-1344. 10.1109/icassp.2001.941176
3	KIM H J， CHOI Y H. A novel echo-hiding scheme with backward and forward kernels［J］. IEEE Transactions on Circuits and Systems for Video Technology， 2003， 13（8）： 885-889. 10.1109/tcsvt.2003.815950
4	ZENG W， AI H， HU R， et al. An algorithm of echo steganalysis based on Bayes classifier［C］// Proceedings of the 2008 International Conference on Information and Automation. Piscataway： IEEE， 2008： 1667-1670. 10.1109/icinfa.2008.4608272
5	杨榆，雷敏，钮心忻，等. 基于回声隐藏的VDSC隐写分析算法［J］. 通信学报， 2009， 30（2）：83-88. 10.3321/j.issn:1000-436X.2009.02.014
	YANG Y， LEI M， NIU X X， et al. VDSC steganalysis algorithm based on echo hiding［J］. Journal on Communications， 2009， 30（2）： 83-88. 10.3321/j.issn:1000-436X.2009.02.014
6	XIE C， CHENG Y， WU F. A new detection scheme for echo hiding［C］// Proceedings of the 2010 IEEE International Conference on Information Theory and Information Security. Piscataway： IEEE， 2010： 242-246. 10.1109/icitis.2010.5688776
7	王昱洁，郭立，王翠平. 基于功率倒谱统计特征的音频回声隐写分析方法［J］. 中国科学技术大学学报， 2010， 40（10）：1074-1081. 10.3969/j.issn.0253-2778.2010.10.012
	WANG Y J， GUO L， WANG C P. An audio steganalysis method for echo hiding based on statistical features of power cepstrums［J］. Journal of University of Science and Technology of China， 2010， 40（10）：1074-1081. 10.3969/j.issn.0253-2778.2010.10.012
8	CHEN B， LUO W， LUO D. Identification of audio processing operations based on convolutional neural network［C］// Proceedings of the 6th ACM Workshop on Information Hiding and Multimedia Security. New York： ACM， 2018： 73-77. 10.1145/3206004.3206005
9	CHEN B， LUO W， LI H. Audio steganalysis with convolutional neural network［C］// Proceedings of the 5th ACM Workshop on Information Hiding and Multimedia Security. New York： ACM， 2017： 85-90. 10.1145/3082031.3083234
10	LIN M， CHEN Q， YAN S. Network in network［EB/OL］. ［2018-03-04］. . 10.1109/icicta.2014.118
11	WANG Y， YANG K， YI X， et al. CNN-based steganalysis of MP3 steganography in the entropy code domain［C］// Proceedings of the 6th ACM Workshop on Information Hiding and Multimedia Security. New York： ACM， 2018： 55-65. 10.1145/3206004.3206011
12	SEICHTER D， CUCCOVILLO L， AICHROTH P. AAC encoding detection and bitrate estimation using a convolutional neural network［C］// Proceedings of the 2016 IEEE International Conference on Acoustics， Speech and Signal Processing. Piscataway： IEEE， 2016： 2069-2073. 10.1109/icassp.2016.7472041
13	LUO D， YANG R， LI B， et al. Detection of double compressed AMR audio using stacked autoencoder［J］. IEEE Transactions on Information Forensics and Security， 2017， 12（2）： 432-444. 10.1109/tifs.2016.2622012
14	IOFFE S， SZEGEDY C. Batch normalization： accelerating deep network training by reducing internal covariate shift［EB/OL］. ［2018-03-02］. .
15	SIMONYAN K， ZISSERMAN A. Very deep convolutional networks for large-scale image recognition［EB/OL］. ［2018-04-10］. . 10.5244/c.28.6
16	GAROFOLO J S， LAMEL L F， FISHER W M， et al. DARPA TIMIT： acoustic-phonetic continuous speech corpus［R］. Gaithersburg， MD： National Institute of Standards and Technology， 1993. 10.6028/nist.ir.4930

[1]	Yunchuan HUANG, Yongquan JIANG, Juntao HUANG, Yan YANG. Molecular toxicity prediction based on meta graph isomorphism network [J]. Journal of Computer Applications, 2024, 44(9): 2964-2969.
[2]	Shunyong LI, Shiyi LI, Rui XU, Xingwang ZHAO. Incomplete multi-view clustering algorithm based on self-attention fusion [J]. Journal of Computer Applications, 2024, 44(9): 2696-2703.
[3]	Yexin PAN, Zhe YANG. Optimization model for small object detection based on multi-level feature bidirectional fusion [J]. Journal of Computer Applications, 2024, 44(9): 2871-2877.
[4]	Yun LI, Fuyou WANG, Peiguang JING, Su WANG, Ao XIAO. Uncertainty-based frame associated short video event detection method [J]. Journal of Computer Applications, 2024, 44(9): 2903-2910.
[5]	Jing QIN, Zhiguang QIN, Fali LI, Yueheng PENG. Diagnosis of major depressive disorder based on probabilistic sparse self-attention neural network [J]. Journal of Computer Applications, 2024, 44(9): 2970-2974.
[6]	Xiyuan WANG, Zhancheng ZHANG, Shaokang XU, Baocheng ZHANG, Xiaoqing LUO, Fuyuan HU. Unsupervised cross-domain transfer network for 3D/2D registration in surgical navigation [J]. Journal of Computer Applications, 2024, 44(9): 2911-2918.
[7]	Hong CHEN, Bing QI, Haibo JIN, Cong WU, Li’ang ZHANG. Class-imbalanced traffic abnormal detection based on 1D-CNN and BiGRU [J]. Journal of Computer Applications, 2024, 44(8): 2493-2499.
[8]	Yuhan LIU, Genlin JI, Hongping ZHANG. Video pedestrian anomaly detection method based on skeleton graph and mixed attention [J]. Journal of Computer Applications, 2024, 44(8): 2551-2557.
[9]	Yanjie GU, Yingjun ZHANG, Xiaoqian LIU, Wei ZHOU, Wei SUN. Traffic flow forecasting via spatial-temporal multi-graph fusion [J]. Journal of Computer Applications, 2024, 44(8): 2618-2625.
[10]	Qianhong SHI, Yan YANG, Yongquan JIANG, Xiaocao OUYANG, Wubo FAN, Qiang CHEN, Tao JIANG, Yuan LI. Multi-granularity abrupt change fitting network for air quality prediction [J]. Journal of Computer Applications, 2024, 44(8): 2643-2650.
[11]	Zheng WU, Zhiyou CHENG, Zhentian WANG, Chuanjian WANG, Sheng WANG, Hui XU. Deep learning-based classification of head movement amplitude during patient anaesthesia resuscitation [J]. Journal of Computer Applications, 2024, 44(7): 2258-2263.
[12]	Dongwei WANG, Baichen LIU, Zhi HAN, Yanmei WANG, Yandong TANG. Deep network compression method based on low-rank decomposition and vector quantization [J]. Journal of Computer Applications, 2024, 44(7): 1987-1994.
[13]	Huanhuan LI, Tianqiang HUANG, Xuemei DING, Haifeng LUO, Liqing HUANG. Public traffic demand prediction based on multi-scale spatial-temporal graph convolutional network [J]. Journal of Computer Applications, 2024, 44(7): 2065-2072.
[14]	Zhi ZHANG, Xin LI, Naifu YE, Kaixi HU. DKP： defending against model stealing attacks based on dark knowledge protection [J]. Journal of Computer Applications, 2024, 44(7): 2080-2086.
[15]	Yiqun ZHAO, Zhiyu ZHANG, Xue DONG. Anisotropic travel time computation method based on dense residual connection physical information neural networks [J]. Journal of Computer Applications, 2024, 44(7): 2310-2318.

Detection method for echo hiding based on convolutional neural network framework

基于卷积神经网络框架的回声隐藏检测方法

RichHTML

PDF

Knowledge

Abstract

Cite this article

share this article

Figures/Tables 12

References 16

Related Articles 15

Recommended Articles

Metrics