CAJ | 학술논문

为了提高水印的稳定性和抗几何攻击能力，提出了一种结合图像归一化技术和分形理论的零水印算法。该算法首先利用归一化技术将原始图像映射到几何不变空间中，再以几何中心为中点提取归一化图像的重要区域；然后对该区域进行分形编码，由拼贴误差获得重要区域的边缘信息，即图像特征信息。最后用该信息与水印做相应的处理后即可得到提取水印的密钥矩阵。仿真实验结果表明，算法对常规图像处理和几何攻击具有较强的鲁棒性。
위료제고수인적은정성화항궤하공격능력，제출료일충결합도상귀일화기술화분형이론적령수인산법。해산법수선이용귀일화기술장원시도상영사도궤하불변공간중，재이궤하중심위중점제취귀일화도상적중요구역；연후대해구역진행분형편마，유병첩오차획득중요구역적변연신식，즉도상특정신식。최후용해신식여수인주상응적처리후즉가득도제취수인적밀약구진。방진실험결과표명，산법대상규도상처리화궤하공격구유교강적로봉성。
In order to improve the watermark’s stability and the ability to resist geometric attacks, this paper puts forward a kind of zero-watermarking algorithm combined image normalization and fractal theory. Firstly, the image’s geometricaly invariant space is constructed by using the image normalization method, the significant area is gained from the normalized image by using the geometric center. Then the significant area is encoded by fractal coding. The edge information wil be obtained from the colage error of the image’ s fractal coding. That is the image feature information. Finaly, do certain processing on the watermark and feature information to achieve the key matrix. It wil be used in detecting watermark. The experimental results show that this approach has good performance to the image processing and geometric attack..