【摘要】 在人物新闻图像中，人眼视觉更关注于人脸高频细节部分，为实现含水印的人物新闻图像受到任何微弱攻击情况下均不能完整提取出水印信息，且具有较好的不可感知性，提出了一种Arnold置乱结合离散小波变换(DWT)、正交三角分解法(QR)的脆弱性图像水印算法。首先将水印图像进行Arnold置乱加密，再提取载体图像B通道中人眼不敏感的低频细节部分进行QR分解，最后将Arnold置乱加密后的水印图像QR分解后嵌入在载体图像矩阵R中每一个元素，使较少的水印信息均嵌入在载体图像的每一个像素中。实验结果表明，该算法在未受攻击情况下平均峰值信噪比(PSNR)为51 dB,归一化相关系数(NC)为1.0,在任何强度的攻击下均无法提取完整水印图像，具有较好的人物新闻图像认证能力。更多还原

【Abstract】 In news profile image captioning, human eye focus more on the face details with high frequency. In order to realize the inextractibility of the watermark embedded profile image under any weak attack and the good perception, this paper puts forward a Arnold scrambling combined with discrete wavelet transform(DWT), orthogonal triangle decomposition method(QR) vulnerability image watermark algorithm. First, the watermarked image is encrypted by Arnold scrambling, and then the insensitive low-frequency details of the human eye in the B channel of the carrier image is extracted for QR decomposition. Finally, the QR decomposition of the encrypted watermark image is embedded in every element of the carrier image matrix R, so that less watermark information is embedded in every pixel of the carrier image. The experimental results show that the average peak signal-to-noise ratio(PSNR) is 51 dB, and the normalized correlation coefficient(NC) is 1.0. It cannot extract the complete watermark image under any intensity attack, and has good character news image authentication ability.更多还原