一种基于宏块复杂度的视频隐写算法

doi:10.19665/j.issn1001-2400.2022.02.019

摘要/Abstract

摘要：

基于运动矢量的视频隐写算法通常会破坏运动矢量的局部最优性,这种统计特性的破坏容易被对应的隐写分析算法所检测,致使算法的抗隐写检测性能降低,隐写安全性受到威胁。为降低对运动矢量局部最优性的破坏,在分析运动矢量的修改对视频质量和运动矢量局部最优性影响的基础上,提出一种基于宏块复杂度的视频隐写算法,选择低复杂度的宏块运动矢量作为隐写载体,有效地保持了嵌入信息后运动矢量的局部最优性。首先利用希尔伯特填充曲线对宏块像素进行扫描并定义复杂度,然后统计宏块复杂度,根据待嵌信息长度动态确定嵌入阈值,最后选择复杂度低于嵌入阈值的宏块运动矢量进行随机匹配修改以嵌入秘密信息。实验结果表明,以1 000 kb/s的压缩率对视频进行压缩并嵌入信息时,新算法隐写视频的峰值信噪比和结构相似性值下降分别不超过0.30 dB和0.04,比特率增长不超过0.97 %。与同类算法对比结果显示,新算法的隐写视频质量高,比特率增长小,算法具有较高的抗隐写检测能力和安全性。

关键词: 运动矢量, 隐写, 希尔伯特曲线, 宏块复杂度, 隐写分析

Abstract:

The video steganography based on the motion vector (MV) usually destroys the local optimality MV,and the destruction of such statistical properties is easily detected by the corresponding steganography analytical algorithms,resulting in a reduced performance of anti-stegoanalysis and steganography security.In order to reduce the damage to the local optimality of MVs,a video steganography algorithm based on macroblock complexity is proposed through the analysis of the influence of MV modification on the video quality and the local optimality of MV,and the low-complexity macroblock motion vectors are selected as carriers to effectively maintain the local optimality after embedding information.The proposed algorithm first introduces the Hilbert filling curve to scan macroblock pixels and defines macroblock complexity,then the macroblock complexity distribution is counted and the embedding threshold is dynamically determined according to the length of to-be-embedded data,and finally selects the MV of macroblock whose complexity is lower than the embedding threshold for random matching modification to embed secret information.Experimental results show that the stego video PSNR and SSIM degradation of the proposed algorithmare no more than 0.30 dB and 0.04,respectively,and the bit rate increase does not exceed 0.97 % when the video is compressed and embedded with a compression rate of 1000 Kb/s.Its comparison with related algorithms show that the stego video of the proposed algorithm has a high-level visual quality and a low-level bit rate growth,and that the proposed algorithm has good anti-steganalysis detection capability and security.

Key words: motion vector, steganography, Hilbert curve, macroblock complexity, steganalysis

中图分类号:

TP309.7

杨晓元,唐洪琼,钮可,张英男. 一种基于宏块复杂度的视频隐写算法[J]. 西安电子科技大学学报, 2022, 49(2): 164-172.

YANG Xiaoyuan,TANG Hongqiong,NIU Ke,ZHANG Yingnan. Video steganography based on macroblock complexity[J]. Journal of Xidian University, 2022, 49(2): 164-172.

图/表 7

图1

图2

图3

图4

表1

图5

表2

参考文献 18

[1]	KUTTER M, JORDAN F, EBRAHIMI T. Proposal of A Watermarking Technique for Hiding/Retrieving Data in Compressed And Decompressed Video:Technical Report M2281,JTC1/SC29/WG11[R]. Geneva:ISO/IEC, 1997.
[2]	XU C Y, PING X J, ZHANG T. Steganography in Compressed Video Stream[C]// Proceedings of the 1^st International Conference on Innovative Computing,Information and Control.Piscataway:IEEE, 2006:269-272.
[3]	FANG D Y, CHANG L W. Data Hiding for Digital Video with Phase of Motion Vector[C]// Proceedings of IEEE International Symposium on Circuits and Systems.Piscataway:IEEE, 2006:1422-1425.
[4]	ALY H. 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
[5]	CAO Y, ZHAO X F, FENG D G, et al. Video Steganography with Perturbed Motion Estimation[C]// Proceedings of the 13^th International Conference on Information Hiding.Heidelberg:Springer, 2011:193-207.
[6]	CAO Y, ZHANG H, ZHAO X F, et al. Video Steganography Based on Optimized Motion Estimation Perturbation[C]// Proceedings of the 3^rd ACM Workshop on Information Hiding and Multimedia Security. New York: ACM, 2015:25-31.
[7]	FILLER T, JUDAS J, 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
[8]	钮可, 李军, 刘佳, 等. 博弈模型上的策略自适应隐写方法[J]. 西安电子科技大学学报, 2015, 42(4):165-170.
	NIU Ke, LI Jun, LIU Jia, et al. Strategic Adaptive Steganography Based on the Game Theory Model[J]. Journal of Xidian University, 2015, 42(4):165-170..
[9]	YAO Z Y, ZHANG W M, YU N H, 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
[10]	WANG P P, ZHANG H, CAO Y, et al. A Novel Embedding Distortion for Motion Vector-Based Steganography Considering Motion Characteristic,Local Optimality and Statistical Distribution[C]// Proceedings of the 4^th ACM Workshop on Information Hiding and Multimedia Security, New York: ACM, 2016:127-137.
[11]	王丽娜, 徐一波, 翟黎明, 等. 基于宏块复杂度的自适应视频运动矢量隐写算法[J]. 计算机学报, 2017, 40(5):1044-1056.
	WANG Lina, XU Yibo, ZHAI Liming, et al. An Adaptive Video Motion Vector Steganography Based on Macroblock Complexity[J]. Chinese Journal of Computers, 2017, 40(5):1044-1056.
[12]	ZHAI L M, WANG L N, REN Y Z. Multi-Domain Embedding Strategies for Video Steganography by Combining Partition Modes and Motion Vectors[C]// 2019 IEEE International Conference on Multimedia and Expo,Piscataway:IEEE, 2019:1402-1407.
[13]	王丽娜, 王旻杰, 翟黎明, 等. 基于相关性异常的H.264/AVC视频运动矢量隐写分析算法[J]. 电子学报, 2014, 42(8):1457-1464. doi: 10.3969/j.issn.0372-2112.2014.08.001
	WANG Lina, WANG Minjie, ZHAI Liming, et al. H.264/AVC Video Steganalysis Algorithm Based on Motion Vector Abnormal Correlation[J]. Acta Electronica Sinica, 2014, 42(8):1457-1464. doi: 10.3969/j.issn.0372-2112.2014.08.001
[14]	张弘, 尤玮珂, 赵险峰. 视频隐写分析技术研究综述[J]. 信息安全学报, 2018, 3(6):13-27.
	ZHANG Hong, YOU Weike, ZHAO Xianfeng. A Survey of Video Steganalysis[J]. Journal of Cyber Security. 2018, 3(6):13-27.
[15]	ZHAI L M, WANG L N, REN Y Z. Universal Detection of Video Steganography in Multiple Domains Based on the Consistency of Motion Vectors[J]. IEEE Transactions on Information Forensics and Security, 2020, 15:1762-1777. doi: 10.1109/TIFS.2019.2949428
[16]	WANG K R, ZHAO H, WANG H X. 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
[17]	ZHAI L M, WANG L N, REN Y Z. Combined and Calibrated Features for Steganalysis of Motion Vector-Based Steganography in H.264/AVC[C]// Proceedings of the 5^th ACM Workshop on Information Hiding and Multimedia Security. New York: ACM, 2017:135-146.
[18]	CHANG C, LIN C. LIBSVM:A Library for Support Vector Machines[J]. ACM Transactions on Intelligent Systems and Technology, 2011, 3:1-27.

压缩方式	实际比特率	嵌入率	平均比特率
压缩方式	实际比特率	嵌入率	文献[2]	文献[4]	文献[5]	文献[11]	文中算法
固定比特率压缩(1 000 kb/s)	966.19	0.1	979.56	980.71	968.54	966.41	964.76
		0.2	990.57	993.09	971.39	968.42	967.22
		0.3	996.44	1 000.28	974.18	971.99	970.93
		0.4	1 005.3	1 004.91	975.52	976.68	975.54
固定量化参数(为4)	1 524.65	0.1	1 618.52	1 631.61	1 546.47	1 535.88	1 519.94
		0.2	1 737.57	1 736.07	1 570.79	1 550.87	1 540.50
		0.3	1 842.05	1 853.96	1 591.23	1 573.62	1 571.99
		0.4	1 944.54	1 968.19	1 601.79	1 605.42	1 617.87

检测特征	比特率	嵌入率	准确率
检测特征	比特率	嵌入率	文献[2]	文献[4]	文献[5]	文献[11]	本文算法
局部最优性特征	500 kb/s	0.1	0.91	0.91	0.65	0.57	0.54
		0.2	0.96	0.97	0.73	0.70	0.66
	1 000 kb/s	0.1	0.92	0.90	0.69	0.66	0.60
		0.2	0.97	0.97	0.79	0.74	0.71
相邻相关性特征	500 kb/s	0.1	0.84	0.76	0.67	0.71	0.71
		0.2	0.94	0.87	0.83	0.86	0.84
	1 000 kb/s	0.1	0.82	0.78	0.65	0.69	0.70
		0.2	0.93	0.88	0.77	0.80	0.81