Abstract: This paper presents a novel robustness algorithm. It uses an improved curvature scale space technology to obtain a few robust corner points to rebuild the image suffered geometric attacks. Two chirp signals as the watermark are selected and embedded into the discrete Fractional Fourier Transform(FRFT) domain of the image. The transform orders and the matrices of the position where the watermark is embedded are used as encryption keys of the algorithm. With the help of the property that a chirp signal can produce a strong impulse in the FRFT domain, the watermark can be detected conveniently. Simulation experimental results show that this algorithm assures good imperceptibility, security and robustness.

Key words: image watermarking, chirp signal, corner points, curvature scale space, Fractional Fourier Transform(FRFT)

摘要： 提出一种新的鲁棒性水印算法，利用改进的曲率尺度空间技术，提取图像的少量较顽强角点，用于重建受几何攻击的图像。选取2个chirp信号作为水印信息，将其嵌入图像的分数傅里叶变换(FRFT)域，选用变换阶次和水印嵌入位置矩阵作为算法密钥，根据chirp信号在FRFT域呈现的冲激特征检测水印信号。仿真实验结果表明，该算法具有较高不可见性、安全性和鲁棒性。

关键词: 图像水印, chirp信号, 角点, 曲率尺度空间, 分数傅里叶变换

CLC Number: 

• TP309