基于ASIFT-Kmeans-九域卷积小波变换的抗旋转攻击图像水印算法  

Watermarking algorithm based on ASIFT-Kmeans-Nine-Region convolutional wavelet transform for rotation-resilient images

作　　者：白健平 杜庆治[1] 龙华[1] 邵玉斌[1] BAI Jian-Ping;DU Qing-Zhi;LONG Hua;SHAO Yu-Bin(College of Information Engineering and Automation,Kunming University of Science and Technology,Kunming 650500,China)

机构地区：[1]昆明理工大学信息工程与自动化学院,昆明650500

出　　处：《四川大学学报(自然科学版)》2025年第2期412-424,共13页Journal of Sichuan University(Natural Science Edition)

基　　金：云南省媒体融合重点实验室开放项目(220235205)。

摘　　要：针对大多数鲁棒性数字图像水印算法抵抗旋转攻击能力不佳的问题,本文提出了一种新的ASIFT-Kmeans-九域卷积小波变换的鲁棒水印算法.首先用ASIFT方法采集的图像中仿射不变性良好的特征点,并将这些特征点使用Kmeans算法聚类为3簇,构造每个簇中稳定且不重叠的特征区域作为水印嵌入区域;然后使用本文定义的九域卷积小波变换将图像变换到9个不同的频率域,尽可能全面地获取图像的细节信息,从而找出最适合的水印嵌入域;最后做DCT-SVD嵌入水印信息.仿真实验表明,本文算法对旋转攻击和旋转组合攻击均具备很好的鲁棒性,对常规攻击也具有较好的鲁棒性.在峰值信噪比高于37 dB的情况下,提取的盲水印归一化系数均在0.87以上.In response to the problem of poor resistance to rotation attacks in most robust digital image watermarking algorithms,this paper proposes a novel robust watermarking algorithm based on ASIFT-Kmeans-Nine-Region Convolutional Wavelet Transform.Firstly,the ASIFT method is employed to robustly extract image keypoints with well-maintained affine invariance.These feature points are then clustered into three clusters using the Kmeans algorithm,constructing stable and non-overlapping feature regions in each cluster as watermark embedding regions.Subsequently,the Nine-Region Convolutional Wavelet Transform defined in this paper is utilized to transform the image into nine different frequency domains,comprehensively capturing detailed information of the image to identify the most suitable watermark embedding domain.Finally,the watermark information is embedded using the DCT-SVD technique.Simulation experiments demonstrate that the proposed algorithm exhibits excellent robustness against rotation attacks and rotation combination attacks,as well as strong robustness against common attacks.With a Peak Signal-to-Noise Ratio(PSNR)higher than 37 dB,the normalized coefficients of the extracted blind watermark are all above 0.87.

关 键 词：特征区域 盲水印 九域卷积小波变换 随机密钥 K-Means 

分 类 号：TN911.7[电子电信—通信与信息系统]