一种参数自适应的鲁棒性数字图像水印算法  

A Parameter Adaptive Robust Digital Image Watermarking Algorithm

作　　者：夏嵬 赵小厦 XIA Wei;ZHAO Xiaosha(Nanjing Vocational College of Information Technology,Nanjing Jiangsu 210023,China;Nanjing Technician College,Nanjing Jiangsu 210023,China)

机构地区：[1]南京信息职业技术学院,江苏南京210023 [2]南京技师学院,江苏南京210023

出　　处：《信息与电脑》2021年第24期70-72,共3页Information & Computer

基　　金：南京信息职业技术学院自然科研基金项目“一种平衡各指标的抗几何攻击数字图像水印算法研究”(项目编号:YK20181501)。

摘　　要：数字水印技术作为保护版权的重要手段,在信息安全领域有着重要的地位。传统的数字水印算法的参数大多凭经验设定,达不到鲁棒性最佳的效果。本文根据图像和水印的容量自适应地在图像的低频区嵌入水印,并利用机器学习的梯度下降法对水印算法的参数进行训练,使得含水印图像在不同攻击下恢复出的水印总体误码率相对最低,让参数能最大限度地接近最优值。实验证明,该算法不但对JPEG压缩、裁剪、噪声等常见攻击有较强的免疫力,而且对平移、旋转、缩放等几何攻击也具有较强的免疫力。As an important means to protect copyright,digital watermarking technology plays an important role in the field of information security.The parameters of traditional digital watermarking algorithms are mostly set by experience,which can not achieve the best robustness.In this paper,the watermark is adaptively embedded in the low-frequency region of the image according to the capacity of the image and the watermark,and the gradient descent method of machine learning is used to train the parameters of the watermark algorithm,so that the overall bit error rate of the watermark recovered from the watermarked image under different attacks is relatively low,so that the parameters can be close to the optimal value to the greatest extent.Experiments show that the algorithm has strong immunity not only to common attacks such as JPEG compression,clipping and noise,but also to geometric attacks such as translation,rotation and scaling.

关 键 词：自适应 梯度下降 训练 误码率 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]