基于整数小波变换与矩阵分解的感兴趣区域可逆水印算法研究

目的针对传统感兴趣区域水印算法抵抗几何攻击能力较弱的缺陷,提出一种基于IWT-Schur的感兴趣区域可逆水印算法。方法首先对载体图像做小波变换,筛选出各子带ROI系数,接着采用Arnold变换加密水印图像,加密水印图像做整数小波变换得到一系列分量。最后结合Schur分解,将水印各分量对应加至载体各子带的ROI。结果采用整数小波变换IWT与矩阵Schur分解的嵌入方式,使得含水印图像的视觉质量良好,算法实现容易。含水印图像没有受到干扰时检测到的水印与原水印一致,含水印图像受到攻击时,也表现出良好的性能,检测到的水印品质较好。结论实验证明,此方案水印提取正确,且感兴趣区域无损恢复。

Algorithms of Digital Watermarking in Region of Interest Based on Integer Wavelet Transform and Matrix Decomposition

The work aims to propose a reversible watermarking algorithm of region of interest based on IWT-Schur for the shortcomings of the watermarking algorithm in traditional region of interest against geometric attack. Firstly, the carrier image was wavelet transformed, and the sub-band ROI coefficients were filtered out. Then, the Arnold transform was used to encrypt the watermark image, and the encrypted watermark image was used as an integer wavelet transform to obtain a series of components. Finally, combined with the Schur decomposition, the components of the watermark were correspondingly added to the ROI of each sub-band of the carrier. The embedded wavelet transform and the Schur decomposition of the matrix made the water-printed image have good visual quality and the algorithm was easy to implement. The watermark detected when the watermarked image was not disturbed was consistent with the original watermark. When the watermarked image was attacked, it also showed good performance, and the quality of detected watermark was good. Experiments show that the watermark extraction is correct and the region of interest is not restored.