摘 要:幾何形變一直是各類數(shù)字水印算法較難抵抗的攻擊手段。針對此問題,提出一種抗幾何攻擊數(shù)字圖像水印算法。采用周期性的水印嵌入圖像,強度由NVF函數(shù)確定。在檢測時,用白化濾波器對圖像進行預(yù)處理。由于水印數(shù)據(jù)是隨機的,而圖像數(shù)據(jù)高度相關(guān),白化濾波器起到增加水印數(shù)據(jù)能量,同時降低圖像數(shù)據(jù)能量的作用。由理論計算得出選擇濾波器的原則。然后,通過自相關(guān)檢測,得到反映幾何形變的峰值點。對實驗結(jié)果的分析證明,該算法對各類幾何攻擊都能夠精確定位。
關(guān)鍵詞:數(shù)字圖像水印; 幾何攻擊; 周期水印; 白化濾波器; 自相關(guān)檢測
中圖分類號:TN911.73 文獻(xiàn)標(biāo)識碼:A
文章編號:1004-373X(2010)12-0114-03
Image Watermarking Algorithm for Resisting Geometric Attack
DENG Xiao-ying, CHEN Wei-feng
(College of Physics and Technology, Yangzhou University, Yangzhou 225002, China)
Abstract:An anti-geometric attack digital image watermarking algorithm, in which the periodic watermarking-embedded images whose intensity are determined by NVF function are adopted, is proposed to resist the geometric deformation which is difficult for many watermarking algorithm. The whitening filter is used in image pre-processing during inspection. As the watermark data is random and the image data are highly relevant, the whitening filter plays an important part in increasing the energy of watermark data and reducing the energy of image data. The principle of filter selection can be obtained by the theoretical calculation. The peak points, which reflect the geometric deformation, can be deduced by the self-correlation detection. The experimental results prove that the algorithm can pinpoint various geometric attacks.
Keywords: digital image watermarking; geometric attack; periodic watermarking; whitening filter; self-correlation detection
0 引 言
數(shù)字水印技術(shù)是用于數(shù)字產(chǎn)品保護的一種技術(shù)。多數(shù)應(yīng)用要求水印算法對各種處理和攻擊具有較高的魯棒性和較強的抗攻擊性。現(xiàn)在,基于擴頻思想的水印算法已經(jīng)能夠很好地抵抗多種針對圖像的攻擊,但一般難以抵抗由于旋轉(zhuǎn)、縮放和平移處理等幾何處理所造成的去除同步攻擊。針對幾何攻擊,人們提出以下幾種嵌入水印思路:在圖像的幾何不變域嵌入水印;基于模板匹配的水印算法;基于圖像特征的水印算法;嵌入周期性水印。本文提出的算法屬于第4類。
周期性的水印之所以能夠抗幾何攻擊,在于檢測器采用自相關(guān)或自卷積運算,不管對圖像如何操作,水印信息的周期性不會變化,所以檢測器輸出的局部峰值也呈周期性排列[1-3]。……