| 摘要: |
| 目的 为了提高水印的抗几何攻击能力， 以DCT(余弦变换)变换生产全息水印， 并通过DCT嵌入宿主图像的中频区域， 结合Randon变换技术， 使得水印拥有良好的抗裁切和几何形变性能。方法 首先， 根据FFT(傅里叶变换)和DCT全息的成像特点介绍全息水印。然后， 通过计算机系统将原始图像与参考光干涉， 生成全息图像后经 DCT 变换， 利用 DCT 嵌入到宿主图像的 Y 通道中。最后， 结合Randon变换算法检测扫描图像倾斜角度。结果 以嵌入强度 a为0.4嵌入打印扫描后， 含水印图像的PSNR值达到39.53， 提取水印SSIM达到0.783。嵌入水印有很强的不可见性， 提取水印清晰， 嵌入量较大， 有一定的鲁棒性。结论 该算法能够快速、 准确地从受到旋转、 滤波、 噪声、 裁切和几何形变等攻击的宿主图像中提取水印。 |
| 关键词: Radon变换 余弦变换 全息水印 角度校正 SSIM算法 |
| DOI： |
| 分类号:TS805.4 |
| 基金项目:上海市教委科研创新重点项目 （13ZZ111） |
|
| Anti-geometric Distortion Holographic Counterfeiting Watermark Technology |
|
BAO Guan-xiao, SUN Liu-jie, ZHOU Zhong-yuan
|
|
University of Shanghai for Science and Technology， Shanghai 200093， China
|
| Abstract: |
| Objective In order to improve the watermark against geometrical attacks, holographic watermark was produced by DCT transform (Cosine Transform) and embedded in the IF region of the host image by DCT technology, in combination with Randon transform technology, the watermark was granted with high resistance to cutting and geometric distortion. Methods Firstly, the imaging characteristics of holographic watermark was introduced according to the FFT (Fast Fourier transformation) and DCT holographic features. Then hologram optical interference was generated by the computer system with reference to the original digital information, and then the holographic image underwent DCT transform, and was embedded into the Y channel of the host image by DCT. Finally, Randon transform algorithm was used to detect the binding inclination angle of the scanned image. Results Experimental results showed that when the embedding intensity was a=0.4 for the embedding of print scan, the PSNR of the watermarked image reached 39.53, and the watermark SSIM reached 0.783. The embedded watermark had strong invisibility, the extracted watermark was clear, the embedding capacity was high, and there was certain level of robustness. Conclusion Therefore, this algorithm could quickly and accurately extract watermark image from the host image which suffered from rotation, filter, noise, cropping, or other geometric distortion attacks. |
| Key words: Radon transform DCT holographic watermark angle correction SSIM algorithm |
