CAJ | 학술논문

为了提高水印算法的抗几何攻击能力，提出一种基于图像特征的彩色图像水印算法。首先利用加速鲁棒性特征SURF （ Speed-Up Robust Features）算法提取出图像中稳定的特征点，并为其构造SURF特征描述算子，使提取出的特征点具有尺度不变性；然后根据特征尺度的自适应性获得局部特征区域并对其进行归一化处理；之后对局部特征区域进行斜变换，并获得其中频系数；最后结合奇异值分解SVD（ Singular Value Decomposition）算法完成水印的嵌入。实验结果表明，嵌入水印后的图像在经过噪声以及几何攻击后，其提取的水印与原水印仍具有较高的相似度，该算法具有较好的鲁棒性。
위료제고수인산법적항궤하공격능력，제출일충기우도상특정적채색도상수인산법。수선이용가속로봉성특정SURF （ Speed-Up Robust Features）산법제취출도상중은정적특정점，병위기구조SURF특정묘술산자，사제취출적특정점구유척도불변성；연후근거특정척도적자괄응성획득국부특정구역병대기진행귀일화처리；지후대국부특정구역진행사변환，병획득기중빈계수；최후결합기이치분해SVD（ Singular Value Decomposition）산법완성수인적감입。실험결과표명，감입수인후적도상재경과조성이급궤하공격후，기제취적수인여원수인잉구유교고적상사도，해산법구유교호적로봉성。
To improve the ability of watermarking algorithm in resisting geometrical attacks, we propose an image feature-based watermarking algorithm for colour images.First, it uses Speed Up Robust Features ( SURF) algorithm to extract steady feature points in the image, and constitutes SURF feature description operator for it, which makes the extracted feature points have scale invariance;Secondly, it obtains local feature region according to the adaptivity of the feature scale and normalises it;Then the algorithm applies slant transformation on local feature region, and gets the frequency coefficient in it; Finally, it completes the watermark embedding by combining singular value decomposition (SVD) algorithm.Experimental results show that the image with the watermark embedded still has higher similarity of its extracted watermark to the original one’ s after suffering the noise and geometrical attacks, this algorithm has quite good robustness.