基于整數(shù)變換的自適應(yīng)圖像可逆水印方法

邱應(yīng)強 余 輪

?

基于整數(shù)變換的自適應(yīng)圖像可逆水印方法

邱應(yīng)強*①②余 輪①

①(福州大學(xué)數(shù)字媒體研究院 福州 360002)②(華僑大學(xué)信息科學(xué)與工程學(xué)院 廈門 361021)

為保證水印圖像質(zhì)量并提高嵌入水印數(shù)據(jù)量，該文提出一種基于整數(shù)變換的自適應(yīng)圖像可逆水印方法。該方法定義了一種新的一般化整數(shù)變換算法，對于任意像素點組成的圖像塊向量進行簡單的整數(shù)變換后，將產(chǎn)生一定的冗余數(shù)據(jù)可用于嵌入水印數(shù)據(jù)。該方法進一步根據(jù)圖像塊向量自身方差大小自適應(yīng)選擇參數(shù)值進行整數(shù)變換，在平滑圖像塊中嵌入更多數(shù)據(jù)，同時避免復(fù)雜圖像塊中引入較大失真，從而保證具有較高的嵌入容量和較好的水印圖像質(zhì)量。與同類算法對比的實驗結(jié)果表明，該方法增大了最大數(shù)據(jù)嵌入容量，以Lena為宿主圖像時有效載荷可達2.36 bpp。該文整數(shù)變換算法運算簡單，通過自適應(yīng)選擇參數(shù)進行整數(shù)變換并嵌入數(shù)據(jù)可保證水印圖像質(zhì)量并實現(xiàn)較大的有效載荷。

圖像處理；可逆水??；整數(shù)變換；有效載荷；圖像質(zhì)量

1 引言

圖像水印可分為可逆水印和不可逆水印兩類。與傳統(tǒng)的不可逆水印相比，可逆水印在提取宿主圖像中嵌入的水印信息后還能無失真恢復(fù)原宿主圖像。利用宿主圖像可逆恢復(fù)的特性，可逆水印技術(shù)可廣泛應(yīng)用于對圖像質(zhì)量要求較高的軍事情報、醫(yī)學(xué)診斷和法律論證等領(lǐng)域，近年來得到了廣泛的發(fā)展。

2 自適應(yīng)可逆水印算法

2.1 可逆整數(shù)變換算法

由于經(jīng)兩次變換后滿足

2.2 自適應(yīng)數(shù)據(jù)嵌入

步驟6 所有向量代表的圖像塊重組最終得到水印圖像。

2.3 提取數(shù)據(jù)及無失真恢復(fù)宿主圖像

接收端提取水印并無失真恢復(fù)原宿主圖像過程是發(fā)送端水印嵌入的逆過程，具體實現(xiàn)步驟為：

3 實驗與結(jié)果分析

3.1 不同圖像塊大小的算法性能比較

圖1 實驗測試圖像

3.2 最大取值不同時的算法性能比較

3.3 與其他文獻方法的性能比較

與本文整數(shù)變換公式：

對比可得兩種方法整數(shù)變換結(jié)果相差不大，文獻[20]直接將數(shù)據(jù)嵌入包含在整數(shù)變換表達式中，而本文方法利用整數(shù)變換產(chǎn)生的冗余數(shù)據(jù)用于數(shù)據(jù)嵌入并采用了多層二值位置圖方式，最大數(shù)據(jù)嵌入容量也略高于文獻[20]方法。本文方法在Lena, Barbara, Baboon, Airplane(F-16), Goldhill和Lake圖像中的最大單位像素有效載荷依次為2.36, 1.65, 1.49, 2.21, 1.95和1.64 bpp，而文獻[20]方法依次為2.34, 1.61, 1.46, 2.19, 1.90和1.59 bpp。

表1 不同宿主圖像在不同圖像塊參數(shù)情況下最大單位像素有效載荷(bpp)

圖2 不同大小圖像塊時Lena的性能曲線

圖3 不同取值時Lena的性能曲線

圖4 不同算法的性能比較

圖5 部分測試圖像性能曲線放大圖

4 結(jié)束語

[1] Barton J M. Method and apparatus for embedding authentication information within digital data[P]. USA, Patent, 5646997, 1997.

[2] Tian J. Reversible data embedding using a difference expansion[J]., 2003, 13(8): 890-896.

[3] Alattar A M. Reversible watermark using the difference expansion of a generalized integer transform[J]., 2004, 13(8): 1147-1156.

[4] Thodi D M and Rodriguez J J. Expansion embedding techniques for reversible watermarking[J]., 2007, 16(3): 721-730.

[5] Kim H J, Sachnev V, Shi Y Q,. A novel difference expansion transform for reversible data embedding[J]., 2008, 3(3): 456-465.

[6] 孫鴻睿, 李光強, 朱建軍, 等. 改進的差值擴張和平移矢量地圖可逆水印算法[J]. 武漢大學(xué)學(xué)報(信息科學(xué)版), 2012, 37(8): 1004-1007.

Sun Hong-rui, Li Guang-qiang, Zhu Jian-jun,. Improved reversible watermarking algorithm for vector map based on difference expansion and shifting[J]., 2012, 37(8): 1004-1007.

[7] Weng S W, Pan J S, and Gao X. Reversible watermarking combining pre-processing operation and histogram shifting[J]., 2012, 3(4): 320-326.

[8] Hu Y J, Lee H K, and Li J W. DE-based reversible data hiding with improved overflow location map[J]., 2009, 19(2): 250-260.

[9] Li Xiao-long, Yang Bin, and Zeng Tie-yong. Efficient reversible watermarking based on adaptive prediction-error expansion and pixel selection[J]., 2011, 20(12): 3524-3533.

[10] Coltuc D. Low distortion transform for reversible watermarking[J]., 2012, 21(1): 412-417.

[11] 劉芳, 汪玉凱. 基于插值預(yù)測和塊分類低失真可逆數(shù)據(jù)隱藏[J]. 光電子·激光, 2013, 24(7): 1392-1398.

Liu Fang and Wang Yu-kai. Reversible data hiding with low distortion based on interpolation prediction and dividing blocks model[J]., 2013, 24(7): 1392-1398.

[12] Ni Z C, Shi Y Q, Ansari N,. Reversible data hiding[J]., 2006, 16(3): 354-362.

[13] Tai W L, Yeh C M, and Chang C C. Reversible data hiding based on histogram modificaiton of pixel differences[J]., 2009, 19(6): 906-910.

[14] Hong W. Adaptive reversible data hiding method based on error energy control and histogram shifting[J]., 2012, 285(2): 101-108.

[15] 王俊祥, 倪江群, 潘金偉. 一種基于直方圖平移的高性能可逆水印算法[J]. 自動化學(xué)報, 2012, 38(1): 88-96.

Wang Jun-xiang, Ni Jiang-qun, and Pan Jin-wei. A high performance reversible watermarking scheme based on histogram shifting[J]., 2012, 38(1): 88-96.

[16] Li Xiao-long, Zhang Wei-ming, Gui Xin-lu,. A novel reversible data hiding scheme based on two-dimensional difference-histogram modification[J]., 2013, 8(7): 1091-1100.

[17] Coltuc D and Chassery J M. Very fast watermarking by reversible contrast mapping[J]., 2007, 14(4): 255-258.

[18] Wang Xiang, Li Xiao-long, Yang Bin,. Efficient generalized integer transform for reversible watermarking[J]., 2010, 17(6): 567-570.

[19] Wang Chao, Li Xiao-long, and Yang Bin. High capacity reversible image watermarking based on integer transform[C]. Proceeding of 2010 IEEE 17th International Conference on Image Processing, Hong Kong, China, Sept. 26-29, 2010: 217-220.

[20] Peng Fei, Li Xiao-long, and Yang Bin. Adaptive reversible data hiding scheme based on integer transform[J]., 2012, 92(1): 54-62.

[21] 張秋余, 孫媛, 晏燕. 基于分塊自適應(yīng)壓縮感知的可逆水印算法[J]. 電子與信息學(xué)報, 2013, 35(4): 797-804.

Zhang Qiu-yu, Sun Yuan, and Yan Yan. A reversible watermarking algorithm based on block adaptive compressed sensing[J].&, 2013, 35(4): 797-804.

邱應(yīng)強： 男，1981年生，博士生，講師，研究方向為信息隱藏、數(shù)字水印和圖像處理等.

余 輪： 男，1952年生，教授，博士生導(dǎo)師，研究方向為通信與信息系統(tǒng)、圖像處理等.

Adaptive Reversible Image Watermarking MethodBased on Integer Transform

Qiu Ying-qiang①②Yu Lun①

①(,,360002,)②(&,,361021,)

To ensure the quality of watermarked image and improve the embedding capacity of watermarkings, an adaptive image reversible watermarking method based on interger transform is proposed in this paper, which defines a new generalized integer transform algorithm. Through the use of the method the image blocks of arbitrary sized are transformed, producing certain redundancy data that can be used for watermarking embedding. In addition, the parameterused for integer transform is adaptively selected according to the variance of every image block, hence allowing for embedding more data bits into the smooth blocks while avoiding large distortion generated by complex ones, and thus the algorithm ensures a higher embedding capacity and better quality of watermarked image. Compared with similar algorithms, the experimental results show that the proposed method has larger maximal embedding capacity and taking Lena as a host image, the real payload can reach up to 2.36 bpp. The proposed integer transform algorithm is simple; through adaptively interger transforming and data embedding, the quality of watermarked image can be assured and the method offers a large real payload.

Image processing; Reversible watermarking; Integer transform; Real payload; Image quality

TP391

A

1009-5896(2014)06-1278-07

10.3724/SP.J.1146.2013.01528

邱應(yīng)強 yqqiu@hqu.edu.cn

2013-10-08收到，2014-01-08改回

國家自然科學(xué)基金(61372107, 61302094)資助課題