CAJ | 학술논문

有效抵抗几何攻击的水印算法是目前水印技术研究的难点之一,很小的几何变形就可能使大多数水印算法失效.对图像Contourlet变换和Harris-Affine算子的特性进行了分析和讨论,在此基础上提出了一种以仿射不变特征点的基于Contourlet的抗几何攻击水印算法.首先,用Contourlet变换提取出多尺度、多方向的纹理信息;再用Harris-Affine检测算子从变换域的中、低频方向子带中提取出仿射不变特征点,结合自适应局部结构的特征尺度确定特征区域,并用U变换对其归一化处理,水印就自适应地嵌入到归一化后的区域中;水印嵌入采纳矢量量化的策略,将水印信息重复嵌入到不同方向子带、多个不相交的局部仿射不变特征区域.检测时,利用特征点实现水印的重同步,采用"多数原则"来提取水印的位信息,无需求助于原始图像.仿真结果表明,算法不仅具有较好的透明性,而且对常规信号处理和去仿射攻击均具有较好的鲁棒性
유효저항궤하공격적수인산법시목전수인기술연구적난점지일,흔소적궤하변형취가능사대다수수인산법실효.대도상Contourlet변환화Harris-Affine산자적특성진행료분석화토론,재차기출상제출료일충이방사불변특정점적기우Contourlet적항궤하공격수인산법.수선,용Contourlet변환제취출다척도、다방향적문리신식;재용Harris-Affine검측산자종변환역적중、저빈방향자대중제취출방사불변특정점,결합자괄응국부결구적특정척도학정특정구역,병용U변환대기귀일화처리,수인취자괄응지감입도귀일화후적구역중;수인감입채납시량양화적책략,장수인신식중복감입도불동방향자대、다개불상교적국부방사불변특정구역.검측시,이용특정점실현수인적중동보,채용"다수원칙"래제취수인적위신식,무수구조우원시도상.방진결과표명,산법불부구유교호적투명성,이차대상규신호처리화거방사공격균구유교호적로봉성
Image watermarking resistant to geometrical attacks is the hotspot and challenging task in the state-of-the-art research on watermarking. Even an invisible affine transformation can invalidate most existing watermarking algorithms. This the authors analyse and discuss the characteristic of contourlet transform and Harris-Affine detector, and on this basis, propose a novel contourlet-domain watermarking algorithm against geometric attacks based on the affine-invariant feature points and characteristics of regions. First, the proposed scheme uses contourlet transform to extract multidirectional and multiscale texture information, and then obtains the feature points invariant to the affine transformation from the low and middle directional subbands by Harris-Affine detector. Second, the feature regions are adaptively computed by the feature scale of the local structure and normalized by U transform. The watermarking is emdedded adaptively in those characteristic regions which are normalized. Finally, by vector quantization, several copies of the watermark are embedded into the no overlapped local feature regions in different directional subbands. During detection, the resynchronization is achieved through feature points. And the embedded sequence can be extracted by "majority principles" without resorting to the original host image. Experimental results show that the proposed algorithm has good transparency and is very robust to common image processing and geometric attacks.