基于QDCT和HVS的双彩色图像水印算法

doi:10.16180/j.cnki.issn1007-7820.2020.08.004

摘要/Abstract

摘要：

针对水印中信息容量、不可见性和鲁棒性相互制约的问题,文中提出了一种基于QDCT和HVS的双彩色图像水印算法。该算法首先将彩色载体图像和彩色水印图像进行QDCT;然后根据HVS的特性决定水印信息的嵌入位置与嵌入强度,并结合广义Arnold变换和冗余嵌入策略, 将水印QDCT系数嵌入到载体图像QDCT系数的实部和虚部。实验结果表明,通过该算法嵌入水印后的彩色图像在PSNR为35 dB时,在常见攻击下提取出的水印大部分NC值高于0.95,水印信息具有更好的不可见性与鲁棒性。

关键词: 四元数, 四元数离散余弦变换, 人类视觉系统, 广义Arnold变换, 冗余嵌入, 双彩色图像水印

Abstract:

Aiming at the mutually restricted problem of information capacity, invisibility and robustness in watermarking, a dual color image watermarking algorithm based on QDCT and HVS was proposed. With this algorithm, a colored carrier image and a colored watermarking image were subject to QDCT. The embedding location and strength of watermarking information were decided based on the characteristics of HVS. Based on the generalized Arnold transform and redundant embedding strategy, the QDCT coefficient of the watermarking image was embedded into the real part and imaginary part of the QDCT coefficient of the carrier image. The experiment result showed with this algorithm, most of NC values of the watermark extracted under common attacks were higher than 0.95 and the watermark information possessed better invisibility and robustness for the colored watermark-embedded image when PSNR was 35 dB.

Key words: quaternion, QDCT, HVS, generalized Arnold transform, redundant embedding, dual color image watermarking

中图分类号:

TN918

经正俊,齐刚. 基于QDCT和HVS的双彩色图像水印算法[J]. 电子科技, 2020, 33(8): 21-27.

JING Zhengjun,QI Gang. Dual Color Image Watermarking Algorithm Based on QDCT and HVS[J]. Electronic Science and Technology, 2020, 33(8): 21-27.

图/表 8

图1

图2

Table 1

表2

表3

图3

表4

Table 5

参考文献 17

[1]	徐欢. 面向数字化图像资源版权保护的数字水印技术研究[D]. 大连:辽宁师范大学, 2018.
	Xu Huan. Research on watermarking technology for copyright protection of digital image resources[D]. Dalian:Liaoning Normal University, 2018.
[2]	苏庆堂. 基于盲提取的彩色图像数字水印算法研究[D]. 上海:华东理工大学, 2013.
	Su Qingtang. Research on color image watermarking algorithm based on blind extraction[D]. Shanghai: East China University of Science and Technology, 2013.
[3]	Su Qingtang. Novel blind colour image watermarking technique using Hessenberg decomposition[J]. IET Image Processing, 2016,10(11):817-829. doi: 10.1049/iet-ipr.2016.0048
[4]	王欢欢, 周霁婷. 基于DD-PCA的可靠性水印改进算法[J]. 电子测量技术, 2017,40(11):108-113.
	Wang Huanhuan, Zhou Jiting. DD-PCA based improved watermarking with reliability[J]. Electronic Measurement Technology, 2017,40(11):108-113.
[5]	Su Qingtang, Niu Yugang, Zou Hailin, et al. A blind dual color images watermarkinging based on singular value decomposition[J]. Applied Mathematics and Computation, 2013,219(16):8455-8466. doi: 10.1016/j.amc.2013.03.013
[6]	Su Qingtang, Niu Yugang, Zou Hailin, et al. A blind double color image watermarking algorithm based on QR decomposition[J]. Multimedia Tools and Applications, 2014,72(1):987-1009. doi: 10.1007/s11042-013-1653-z
[7]	刘凡, 杨洪勇, 苏庆堂. 基于矩阵Schur分解的彩色图像盲水印算法[J]. 计算机应用研究, 2017,34(10):211-215.
	Liu Fan, Yang Hongyong, Su Qingtang. Color imageblind watermarking algorithm based on Schur decomposition[J]. Application Research of Computers, 2017,34(10):211-215.
[8]	王晓红, 孙业强. 基于QR分解的强鲁棒性双彩色盲水印算法[J]. 光电子·激光, 2017,28(9):88-96.
	Wang Xiaohong, Sun Yeqiang. Robust dual color blindwatermarking algorithm based on QR decomposition[J]. Journal of Optoelectronicso Laser, 2017,28(9):88-96.
[9]	万志银, 唐国红, 李桂林. 基于HVS和DWT的双彩色数字图像水印技[J]. 电子科技, 2016,29(1):145-148.
	Wan Zhiyin, Tang Guohong, Li Guilin. Double digital watermarking technique for color images based on DWT and HVS characteristic[J]. Electronic Science and Technology, 2016,29(1):145-148.
[10]	熊祥光. 基于HVS和关系的DCT域彩色图像水印方案[J]. 计算机工程与科学, 2014,36(2):311-316.
	Xiong Xiangguang. Color image watermarking scheme based on HVS and relaitionship in DCT domain[J]. Computer Engineering and Science, 2014,36(2):311-316.
[11]	Panda J, Uppal A, Nair A S, et al. Genetic algorithm based optimized color imagewatermarking technique using SVD and DWT[C]. Hyderabad:IEEE 7^th International Advanced Computing Conference , 2017.
[12]	Giri K J, Bashir R. A block based watermarking approach for color images using discrete wavelet transformation[J]. International Journal of Information Technology, 2017,10(2):139-146. doi: 10.1007/s41870-017-0075-y
[13]	Li Jianzhong, Lin Qun, Yu Chuying, et al. A QDCT- and SVD-based color image watermarking scheme using an optimized encrypted binary computer-generated hologram[J]. Soft Computing, 2018,22(1):47-65. doi: 10.1007/s00500-016-2320-x
[14]	郭进元, 马永强. 基于四元数傅里叶变换的彩色图像数字水印算法[J]. 包装工程, 2017(3):165-169.
	Guo Jinyuan, Ma Yongqiang. Color image digital watermarking algorithm based on quaternion fourier transform[J]. Packaging Engineering, 2017(3):165-169.
[15]	Chen Beijing, Zhou Chunfei, Jeon B, et al. Quaternion discrete fractional random transform for color image adaptive watermarking[J]. Multimedia Tools and Applications, 2018,77(16):20809-20837. doi: 10.1007/s11042-017-5511-2
[16]	Chen Beijing, Coatrieux G, Chen Gang, et al. Full 4-D quaternion discrete Fourier transform based watermarking for color images[J]. Digital Signal Processing, 2014,28(1):106-119. doi: 10.1016/j.dsp.2014.02.010
[17]	Liu Hui, Zhao Bo, Huang Linquan. Quantum image encryption scheme using arnold transform and S-box scrambling[J]. Entropy, 2019,21(4):343-358. doi: 10.3390/e21040343

Su等^[5]	Su等^[6]	刘等^[7]	Chen等^[16]	本文

PSNR 35.157 0	35.366 7	35.008 8	38.346 4	40.246 6

NC 1.000 0	0.999 8	0.999 6	1.000 0	1.000 0

中值滤波 (3×3)	均值滤波 (3×3)	高斯滤波 (3×3)	JPEG压缩 (q=30)	直方图均衡化
0.990 8	0.991 1	0.999 0	0.943 9	0.983 8
高斯噪声 (σ=0.1)	椒盐噪声 (σ=0.1)	平移 (100,0)	剪切 50%	亮度降低 30%
0.876 6	0.936 4	0.930 4	0.982 4	0.995 8

算法	载体图像图1(a)		测试载体图像集
	水印图2(a)	水印图2(b)	水印图2(a)	水印图2(b)
	水印图2(a)	水印图2(b)	PSNR_AVE	PSNR_AVE
Su等^[5]	35.157 0	35.140 6	33.166 9	33.189 4
Su等^[6]	35.366 7	35.425 0	35.440 8	35.436 9
刘等^[7]	35.008 8	35.017 9	34.933 2	34.959 6
Chen等^[16]	38.346 4	38.273 5	38.229 0	38.171 9
本文	40.246 6	42.455 6	39.422 1	41.299 0

攻击类型
中值滤波 (3×3)	图2(a)	0.811 9	0.877 8	0.966 6	0.891 0	0.990 4
中值滤波 (3×3)	图2(b)	0.808 1	0.836 3	0.938 3	0.885 3	0.990 0
均值滤波 (3×3)	图2(a)	0.808 0	0.883 1	0.953 4	0.895 1	0.991 3
均值滤波 (3×3)	图2(b)	0.811 6	0.843 9	0.914 4	0.882 0	0.990 2
高斯滤波 (3×3)	图2(a)	0.992 0	0.992 5	0.991 3	0.996 5	0.998 9
高斯滤波 (3×3)	图2(b)	0.980 6	0.982 3	0.978 3	0.992 5	0.998 7
JPEG压缩 (q=30)	图2(a)	0.905 2	0.848 9	0.825 0	0.909 8	0.947 9
JPEG压缩 (q=30)	图2(b)	0.856 9	0.819 3	0.808 5	0.873 7	0.952 3
高斯噪声 (σ=0.1)	图2(a)	0.848 4	0.795 2	0.800 1	0.877 1	0.931 4
高斯噪声 (σ=0.1)	图2(b)	0.821 8	0.795 4	0.798 3	0.871 8	0.951 2
椒盐噪声 (σ=0.1)	图2(a)	0.926 8	0.912 4	0.847 9	0.872 4	0.952 4
椒盐噪声 (σ=0.1)	图2(b)	0.884 3	0.8674	0.822 3	0.875 0	0.968 0
平移 (100,0)	图2(a)	0.803 8	0.835 1	0.836 8	0.821 8	0.937 6
平移 (100,0)	图2(b)	0.834 8	0.760 9	0.758 1	0.816 1	0.945 9
裁剪50%	图2(a)	0.966 7	0.852 4	0.855 2	0.873 5	0.981 0
裁剪50%	图2(b)	0.931 8	0.801 7	0.7947	0.856 5	0.983 0
直方图均衡化	图2(a)	0.997 0	0.787 2	0.809 7	0.932 7	0.990 3
直方图均衡化	图2(b)	0.990 7	0.793 2	0.795 7	0.901 7	0.967 7
亮度降低 30%	图2(a)	0.999 3	0.839 2	0.823 6	0.997 4	0.995 7
亮度降低 30%	图2(b)	0.997 9	0.824 0	0.808 1	0.989 8	0.995 9

攻击类型	Su等^[5]	Su等^[6]	刘等^[7]	Chen等^[16]	本文
中值滤波 (3×3)
中值滤波 (3×3)	0.814 1	0.894 6	0.983 2	0.897 0	0.990 0
均值滤波 (3×3)
均值滤波 (3×3)	0.813 9	0.888 1	0.957 4	0.902 3	0.991 8
高斯滤波 (3×3)
高斯滤波 (3×3)	0.996 3	0.995 5	0.993 2	0.996 5	0.998 9
JPEG 压缩 (q=30)
JPEG 压缩 (q=30)	0.896 2	0.845 2	0.829 5	0.912 8	0.946 2
高斯噪声 (σ=0.1)
高斯噪声 (σ=0.1)	0.848 8	0.792 6	0.802 1	0.870 4	0.949 2
椒盐噪声 (σ=0.1)
椒盐噪声 (σ=0.1)	0.923 6	0.907 0	0.857 2	0.869 0	0.977 7
剪切50%
剪切50%	0.968 0	0.857 8	0.854 1	0.866 3	0.982 9
平移 (100,0)
平移 (100,0)	0.8028	0.837 7	0.838 1	0.824 6	0.931 6
直方图均衡化
直方图均衡化	0.9980	0.7746	0.782 9	0.924 9	0.992 1
亮度降低30%
亮度降低30%	1.0000	0.8303	0.826 1	0.998 1	0.995 4