基于DCT—SVD抗几何变换的扩频水印算法研究

提出了一种基于奇异值分解（SVD）的扩频水印算法，该算法将原始图像分块DCT变换，对每块DCT系数进行SVD分解，得到奇异值。水印采用重复码，用秘钥产生的伪随机序列对其进行扩频得到扩频后的水印序列。通过使用量化步长，修改最大奇异值实现水印的嵌入，同时结合人类视觉系统模型来平衡水印的鲁棒性和透明性这两种主要的性能指标。实验证明，由于使用了扩频技术，水印提取正确率很高，具有很好的鲁棒性，对噪声、滤波、压缩和剪切都有很强的抵抗力。对于旋转攻击，通过估算旋转角度，运用扩频技术和SVD使得水印提取正确率得到很大提高。

Digital Image Spread Spectrum Watermarking Algorithm Based on DCT-SVD for Anti-geometric Attacks

A spread spectrum watermarking algorithm based on singular value decomposition （SVD） has proposed. The original image block was transformed in DCT block, then SVD decomposition, the singular value. Watermark adopts repetition code, and pseudo-random generator generates a pseudo-random se- quence with a secret key to spread spectrum the watermark. Watermark embedding is achieved by modif- ying the maximum singular value. And combine with a model of the human visual system to balance these two main performance indicators of the robustness and transparency of the watermark. The experiments show that, due to the use of spread spectrum technology, the watermark extraction accuracy rate is very high ,with good robustness to noise, filtering, compression and shear. For spin attack, it makes the water- mark extraction accuracy rate greatly improve by estimating the angle of rotation, the use of spread spec- trum technology and SVD.