根据脊波能刻画图像的线状奇异性,但不能刻画图像中的曲线奇异性的特点,提出一种嵌入到分块图像脊波系数能量最大方向的乘性水印算法。通过加水印的不可见性和可检测性约束,建立基于峰值信噪比受限和能量最大方向保持下的水印强度数学估计模型,给出了水印嵌入强度的可靠性取值区间;并采用无需原始图像的相关检测器实现水印的盲检测,理论估计了给定虚警概率小于10^-8的情况下使漏警概率最小的最优检测阈值。实验结果表明所提出的水印算法具有较高的稳健性和不可见性。
According to the properties that ridgelet can capture the linear singularity but can not represent the curve singularity, a robustness watermarking algorithm based on ridgelet transform of image block is proposed. The watermarks are embedded into the ridgelet coefficients in the highest energy direction of each block. Under the restraints of the invisibility and detectability for watermarks, a mathematical model on the waterniarking strength is proposed according to the restraints of the Peak Signal Noise Ratio and the invariance of the highest energy direction. The value interval of watermarking strength is also given. The watermarks can be blindly detected by correlation function and the optimal threshold is also estimated by minimizing the probability of missing alarm with respect to the given probability of false alarm less than 10^-8. Experimental results show that the proposed algorithm can achieve a better tradeoff between the robustness and transparency.