基于修改概率转换和非加性嵌入失真的视频隐写方法

李林聪; 姚远志; 张晓雅; 张卫明; 俞能海

doi:10.11999/JEIT200001

基于修改概率转换和非加性嵌入失真的视频隐写方法

doi: 10.11999/JEIT200001

中国科学技术大学信息科学技术学院合肥 230027

基金项目: 国家自然科学基金(61802357, U1636201)，国家重点研发计划项目(2018YFB0804102)，中央高校基本科研业务费专项资金(WK3480000009)

详细信息

作者简介:
李林聪：男，1996年生，博士生，研究方向为信息隐藏、深度学习

姚远志：男，1989年生，副研究员，研究方向为信息隐藏、视频编码

张晓雅：女，1998年生，硕士生，研究方向为信息隐藏、计算机视觉

张卫明：男，1976年生，教授，博士生导师，研究方向为信息隐藏、多媒体安全、隐私保护

俞能海：男，1964年生，教授，博士生导师，研究方向为图像视频处理与分析、计算机视觉与模式识别、信息隐藏与媒体内容安全、信息检索与数据挖掘

通讯作者:
姚远志　yaoyz@ustc.edu.cn

中图分类号: TN918
计量
- 文章访问数: 1940
- HTML全文浏览量: 494
- PDF下载量: 111
- 被引次数: 0
出版历程
- 收稿日期: 2020-01-02
- 修回日期: 2020-08-04
- 网络出版日期: 2020-08-12
- 刊出日期: 2020-10-13

Video Steganography Based on Modification Probability Transformation and Non-additive Embedding Distortion

School of Information Science and Technology, University of Science and Technology of China, Hefei 230027, China

Funds: The National Natural Science Foundation of China (61802357, U1636201), The National Key Research and Development Program of China (2018YFB0804102), The Fundamental Research Funds for the Central Universities (WK3480000009)

摘要

摘要: 近年来，基于运动矢量的视频隐写引起了信息隐藏领域研究者的广泛关注。许多视频隐写方法通过合理地对运动矢量定义加性嵌入失真函数获得了良好的性能，然而这些方法忽略了载体元素之间的相互嵌入影响。该文提出的利用非加性嵌入失真的视频隐写方法为运动矢量设计了可以反映相互嵌入影响的联合嵌入失真，并通过分解联合失真实现修改概率的转换，从而动态、合理地在运动矢量的水平分量和垂直分量分配秘密消息。实验结果表明，与使用加性嵌入失真方法相比，该方法能获得更好的安全性和率失真性能。
- 信息隐藏 /
- 视频隐写 /
- 运动矢量 /
- 联合嵌入失真 /
- 修改概率
Abstract: In recent years, research on motion vector-based video steganography has attracted considerable attention from researchers in the field of information hiding. Many video steganographic methods incorporating motion vector-based additive embedding distortion functions have achieved good performance. However, the mutual embedding impact between cover elements in video steganography is neglected in these additive embedding distortion functions. In this paper, joint distortion which reflects the mutual embedding impact for motion vectors is designed. By decomposing joint embedding distortion, modification probability transformation can be achieved and embedding payloads can be dynamically and reasonably allocated in horizontal components and vertical components of motion vectors. Therefore, the video steganography method using non-additive embedding distortion is proposed. Experimental results demonstrate that the proposed method can enhance the security performance significantly compared with the typical methods using additive embedding distortions and obtain the relatively better video coding quality as well.
- Information hiding /
- Video steganography /
- Motion vector /
- Joint embedding distortion /
- Modification probability

HTML全文

图 1 本文和文献[9]的方法的失真函数选定不同参数时抵抗隐写分析方法S1和S2的性能

下载: 全尺寸图片幻灯片

图 2 5种隐写方法抵抗3种隐写分析方法的性能

下载: 全尺寸图片幻灯片

图 3 5种隐写方法抵抗3种隐写分析方法的综合性能

下载: 全尺寸图片幻灯片

表 1 5种嵌入率下5种隐写方法抵抗3种隐写分析方法的最小平均误分类概率(%)

嵌入率(bpmv)	文献[4]方法			文献[5]方法			文献[6]方法			文献[9]方法			本文方法
嵌入率(bpmv)	S1	S2	S3	S1	S2	S3	S1	S2	S3	S1	S2	S3	S1	S2	S3
0.1	46.54	37.19	17.56	46.29	16.02	16.12	40.36	40.95	17.26	45.55	41.10	21.02	48.71	42.14	21.51
0.2	41.69	26.01	11.82	39.86	8.36	11.77	32.00	37.29	13.85	42.68	37.78	14.94	46.54	41.30	16.96
0.3	37.29	23.54	9.50	34.08	6.08	9.74	26.21	33.88	12.41	38.38	34.03	12.71	42.98	41.20	15.68
0.4	34.52	17.11	5.93	26.16	4.55	6.73	17.16	28.04	9.79	34.32	25.82	10.04	40.06	36.84	15.18
0.5	31.75	13.95	4.95	22.26	3.07	5.64	12.31	24.13	7.96	31.95	21.17	7.81	37.24	34.82	12.81

下载: 导出CSV

表 2 嵌入率为0.5bpmv时5种隐写方法得到的载密视频亮度分量的平均PSNR (dB)

视频序列	载体视频	文献[4]方法	文献[5]方法	文献[6]方法	文献[9]方法	本文方法
Bus	34.425	34.185	34.235	34.265	34.270	34.295
City	34.676	34.412	34.432	34.464	34.472	34.494
Coastguard	34.410	34.212	34.234	34.256	34.262	34.286
Crew	36.948	36.798	36.790	36.816	36.814	36.844
Flower	34.520	34.270	34.362	34.440	34.425	34.460
Football	35.938	35.842	35.848	35.838	35.845	35.860
Foreman	36.080	35.814	35.794	35.868	35.876	35.900
Harbour	34.046	33.864	33.854	33.898	33.892	33.924
Ice	39.558	39.282	39.270	39.382	39.380	39.405
Mobile	33.616	33.454	33.430	33.506	33.494	33.524
Paris	33.518	35.375	35.243	35.436	35.401	35.450
Soccer	35.664	35.496	35.504	35.504	35.518	35.548
Stefan	35.340	35.070	35.030	35.210	35.190	35.240
Tempete	34.520	34.358	34.320	34.388	34.392	34.418
Waterfall	34.688	34.250	34.338	34.360	34.388	34.405

下载: 导出CSV

表 3 嵌入率为0.5bpmv时5种隐写方法得到载密视频的平均比特率扩张(%)

视频序列	文献[4]方法	文献[5]方法	文献[6]方法	文献[9]方法	本文方法
Bus	12.81	14.03	3.48	5.77	3.39
City	22.92	20.47	9.02	10.34	7.01
Coastguard	6.78	9.72	3.92	4.64	2.94
Crew	3.99	4.95	3.10	3.15	1.95
Flower	14.87	15.26	2.93	5.14	1.74
Football	2.15	3.27	1.84	2.17	1.23
Foreman	11.92	18.45	5.97	6.52	4.63
Harbour	6.97	7.34	3.15	4.13	2.72
Ice	16.05	18.03	4.71	6.53	4.17
Mobile	9.95	13.10	3.49	5.64	3.00
Paris	10.44	19.15	2.51	6.59	2.46
Soccer	5.34	7.78	3.98	4.57	2.98
Stefan	13.95	14.89	3.63	5.13	2.39
Tempete	8.89	12.93	3.84	4.39	2.71
Waterfall	18.78	20.20	9.11	8.73	6.70

下载: 导出CSV

表 4 文献[9]方法和本文方法在嵌入消息和编码视频步骤的平均执行时间(s)

隐写方法	嵌入消息	编码视频	总计
文献[9]方法	3.39	92.26	95.65
本文方法	3.38	92.66	96.04

下载: 导出CSV

参考文献(22)

HUSSAIN M, WAHAB A W A, IDRIS Y I B, et al. Image steganography in spatial domain: A survey[J]. Signal Processing: Image Communication, 2018, 65: 46–66. doi: 10.1016/j.image.2018.03.012

吴志军, 李常亮, 李荣. 基于随机位置选择和矩阵编码的语音信息隐藏方法[J]. 电子与信息学报, 2020, 42(2): 355–363. doi: 10.11999/JEIT181163

WU Zhijun, LI Changliang, and LI Rong. Speech information hiding method based on random position selection and matrix coding[J]. Journal of Electronics &Information Technology, 2020, 42(2): 355–363. doi: 10.11999/JEIT181163

LIU Yunxia, LIU Shuyang, WANG Yonghao, et al. Video steganography: A review[J]. Neurocomputing, 2019, 335: 238–250. doi: 10.1016/j.neucom.2018.09.091

XU Changyong, PING Xijian, and ZHANG Tao. Steganography in compressed video stream[C]. The 1st International Conference on Innovative Computing, Information and Control-Volume I (ICICIC'06), Beijing, China, 2006: 269–272. doi: 10.1109/ICICIC.2006.158.

ALY H A. Data hiding in motion vectors of compressed video based on their associated prediction error[J]. IEEE Transactions on Information Forensics and Security, 2011, 6(1): 14–18. doi: 10.1109/TIFS.2010.2090520

CAO Yun, ZHAO Xianfeng, FENG Dengguo, et al. Video steganography with perturbed motion estimation[C]. The 13th International Conference on Information Hiding, Prague, Czech Republic, 2011: 193–207. doi: 10.1007/978-3-642-24178-9_14.

FRIDRICH J, GOLJAN M, LISONEK P, et al. Writing on wet paper[J]. IEEE Transactions on Signal Processing, 2005, 53(10): 3923–3935. doi: 10.1109/TSP.2005.855393

FILLER T, JUDAS J, and FRIDRICH J. Minimizing additive distortion in steganography using syndrome-trellis codes[J]. IEEE Transactions on Information Forensics and Security, 2011, 6(3): 920–935. doi: 10.1109/TIFS.2011.2134094

YAO Yuanzhi, ZHANG Weiming, YU Nenghai, et al. Defining embedding distortion for motion vector-based video steganography[J]. Multimedia Tools and Applications, 2015, 74(24): 11163–11186. doi: 10.1007/s11042-014-2223-8

ZHANG Hong, CAO Yun, and ZHAO Xianfeng. Motion vector-based video steganography with preserved local optimality[J]. Multimedia Tools and Applications, 2016, 75(21): 13503–13519. doi: 10.1007/s11042-015-2743-x

ZHU Baolin and NI Jiangqun. Uniform embedding for efficient steganography of H. 264 video[C]. The 25th IEEE International Conference on Image Processing (ICIP), Athens, Greece, 2018: 1678–1682. doi: 10.1109/ICIP.2018.8451214.

SU Yuting, ZHANG Chengqian, and ZHANG Chuntian. A video steganalytic algorithm against motion-vector-based steganography[J]. Signal Processing, 2011, 91(8): 1901–1909. doi: 10.1016/j.sigpro.2011.02.012

CAO Yun, ZHAO Xianfeng, and FENG Dengguo. Video steganalysis exploiting motion vector reversion-based features[J]. IEEE Signal Processing Letters, 2012, 19(1): 35–38. doi: 10.1109/LSP.2011.2176116

WANG Keren, ZHAO Hong, and WANG Hongxia. Video steganalysis against motion vector-based steganography by adding or subtracting one motion vector value[J]. IEEE Transactions on Information Forensics and Security, 2014, 9(5): 741–751. doi: 10.1109/TIFS.2014.2308633

ZHANG Hong, CAO Yun, and ZHAO Xianfeng. A steganalytic approach to detect motion vector modification using near-perfect estimation for local optimality[J]. IEEE Transactions on Information Forensics and Security, 2017, 12(2): 465–478. doi: 10.1109/TIFS.2016.2623587

LI Bin, WANG Ming, LI Xiaolong, et al. A strategy of clustering modification directions in spatial image steganography[J]. IEEE Transactions on Information Forensics and Security, 2015, 10(9): 1905–1917. doi: 10.1109/TIFS.2015.2434600

汤光明, 姜明明, 孙艺. 失真代价动态更新的自适应彩色图像隐写算法[J]. 电子与信息学报, 2019, 41(3): 656–665. doi: 10.11999/JEIT180388

TANG Guangming, JIANG Mingming, and SUN Yi. Adaptive color image steganography based on dynamic distortion modification[J]. Journal of Electronics &Information Technology, 2019, 41(3): 656–665. doi: 10.11999/JEIT180388

LI Weixiang, ZHANG Weiming, CHEN Kejiang, et al. Defining joint distortion for JPEG steganography[C]. The 6th ACM Workshop on Information Hiding and Multimedia Security, Innsbruck, Austria, 2018: 5–16. doi: 10.1145/3206004.3206008.

ZHANG Weiming, ZHANG Zhuo, ZHANG Lili, et al. Decomposing joint distortion for adaptive steganography[J]. IEEE Transactions on Circuits and Systems for Video Technology, 2017, 27(10): 2274–2280. doi: 10.1109/TCSVT.2016.2587388

周洋, 吴佳忆, 陆宇, 等. 面向三维高效视频编码的深度图错误隐藏[J]. 电子与信息学报, 2019, 41(11): 2760–2767. doi: 10.11999/JEIT180926

ZHOU Yang, WU Jiayi, LU Yu, et al. Depth map error concealment for 3D high efficiency video coding[J]. Journal of Electronics &Information Technology, 2019, 41(11): 2760–2767. doi: 10.11999/JEIT180926

FILLER T and FRIDRICH J. Gibbs construction in steganography[J]. IEEE Transactions on Information Forensics and Security, 2010, 5(4): 705–720. doi: 10.1109/TIFS.2010.2077629

KODOVSKY J, FRIDRICH J, and HOLUB V. Ensemble classifiers for steganalysis of digital media[J]. IEEE Transactions on Information Forensics and Security, 2012, 7(2): 432–444. doi: 10.1109/TIFS.2011.2175919

施引文献

资源附件(0)

访问统计