基于压缩传感和LDPC码的图像水印算法研究

Research of Image Watermark Algorithm Based on Compressive Sensing and LDPC

现有的数字水印算法在鲁棒性和篡改检测方面存在不足,提出一种新的水印算法.该算法通过对原始图像进行线性随机投影,得到的压缩测量值经过LDPC(Low Density Parity Check,低密度奇偶校验码)编码,以水印方式嵌入到原始图像中;篡改检测时,同样对篡改图像进行线性随机投影,得到的测量值作为边信息,对提取的水印进行LDPC译码.通过估计原始图像和篡改图像的压缩测量值,并求解一个l1范数问题,实现图像的篡改检测.仿真实验证明,该算法不仅具有较强的抗攻击能力,还具有较好的篡改检测能力.

As the robustness and tamper detection for existing digital watermarking algorithms are poor, this paper proposes a new watermarking algorithm. It conducts liner random projection on the original image. The compressive measurements are encoded with LDPC （ Low Density Parity Check） and embedded into the original image as watermark. While tampering detection, it also conducts liner random projection on the tampered image. The measurements act as side information when the extracted watermark is decoded. By estimating the compressive measurements between original image and tampered image, it can detect the tamper to image. Simulation results show that the proposed algorithm not only has strong ability against attacks, but also has good ability for tamper detection.