摘 要: 基于離散余弦變換(DCT)的特點,提出一種自適應(yīng)水印算法。選取DCT域三個中頻系數(shù)及主對角線系數(shù),并求它們各自的均值,比較嵌入位與這兩個平均值的關(guān)系,調(diào)整嵌入位的大小自適應(yīng)實現(xiàn)水印嵌入。利用Arnold變換對水印信息進行預(yù)處理,增強水印的安全性。實驗表明,該算法水印提取效果較好,對剪切、旋轉(zhuǎn)、修改、JPEG等攻擊具有較好的穩(wěn)健性和魯棒性。
關(guān)鍵詞: 自適應(yīng)算法; 數(shù)字圖像水??; Arnold變換; 離散余弦變換
中圖分類號: TN911.72?34; TP309.7 文獻標(biāo)識碼: A 文章編號: 1004?373X(2013)15?0076?03
Adaptive digital image watermarking technology based on DCT domain
FANG Chun?cheng1, TAN Zhong?ming2, LIN Ruo?bo1, XING Nan?liang3
(1. Department of Mechanical and Electrical Engineering, Jieyang Vocational and Technical College, Jieyang 522051, China;
2. Library, Jieyang Vocational and Technical College, Jieyang 522051, China;
3. Center of Training and Information, Jieyang Vocational and Technical College, Jieyang 522051, China)
Abstract: An adaptive digital watermarking algorithm is proposed based on the characteristics of discrete cosine transformation (DCT). Three medium frequency DCT coefficients and main diagonal coefficients are selected, and their means are calculated. The relation between the embedded coefficient and these two means are analyzed to achieve watermarking embedding adaptively by adjusting the value of embedded coefficient. The watermarking is preprocessed by Arnold transformation before being embedded to enhance the security of the algorithm. The experimental result shows that the algorithm has a good effect in watermarking extraction, as well as enough toughness and robustness for some common attacks, such as cutting, rotation, altering, JPEG compression, etc.
Keywords: adaptive algorithm; digital image watermarking; Arnold transformation; discrete cosine transformation
0 引 言
隨著互聯(lián)網(wǎng)技術(shù)和計算機技術(shù)的廣泛應(yīng)用,網(wǎng)上辦公、網(wǎng)上商務(wù)活動越來越多,信息安全越來越顯示其重要性。網(wǎng)上商務(wù)活動的數(shù)據(jù)必須做得可以防篡改、偽造,數(shù)據(jù)完整性也必須得到保障。加密技術(shù)雖然一定程度上能保護數(shù)據(jù)的安全性,但在防篡改、偽造方面卻存在一定的缺陷;于是,數(shù)字水?。―igital Watermarking)技術(shù)應(yīng)運而生[1?2]。加密技術(shù)與水印技術(shù)的區(qū)別在于,前者對載體的修改攻擊者在感官上能較容易感知,而后者則要求攻擊者對隱蔽載體的改動在感官上不能感知或者不容易感知[2]。水印算法按不同嵌入域,可分為空域算法和變換域算法[1],當(dāng)前的研究熱點是變換域算法。如根據(jù)人類視覺模型特性的DCT(Discrete Cosine Transformation,離散余弦變換)域水印算法等。DCT算法與JPEG(Joint Photographic Experts Group,聯(lián)合圖像專家組)國際數(shù)據(jù)壓縮標(biāo)準(zhǔn)兼容,便于實現(xiàn),應(yīng)用也比較多[2?3]。本文提出的算法為DCT域數(shù)字水印算法。
1 水印預(yù)處理
水印信號都代表一定的意義,如版權(quán)、序列號等,水印嵌入前一般先做預(yù)處理,以增強其安全性?!?br>