基于压缩感知的图像水印信息生成算法  

Image watermark information generation and reconstruction algorithm based on compressive sensing

作　　者：佟德宇 任娜[1,2] 朱长青[1,2] Tong Deyu Ren Na Zhu Changqing(Key Laboratory of Virtual Geographic Environment of Ministry of Education Jiangsu Center for Collaborative Innovation in Geographical Information Resource Development and Application, Nanjing Normal University, Nanjing 210023, China)

机构地区：[1]南京师范大学虚拟地理环境教育部重点实验室,江苏南京210023 [2]南京师范大学江苏省地理信息资源开发与利用协同创新中心,江苏南京210023

出　　处：《华中科技大学学报（自然科学版）》2017年第8期38-41,63,共5页Journal of Huazhong University of Science and Technology(Natural Science Edition)

基　　金：国家自然科学基金资助项目(41301413);公益性行业科研专项基金资助项目(201512021;201512019);江苏省高校优势学科建设工程资助项目

摘　　要：针对图像水印信息冗余性高、安全性和鲁棒性低的问题,提出了一种基于压缩感知的水印信息生成方法.首先对二值图像进行提升小波变换并进行分块处理,各分块使用部分哈达玛测量矩阵进行非线性测量,对测量结果进行二值化,作为水印信息进行水印的嵌入;当水印提取时,使用正交匹配追踪法进行重构,恢复出原始水印图像.对提出的算法进行了实验研究,结果表明:提出的算法能够保证水印信息的安全性,提升水印信息容量,并有效去除图像水印的冗余部分,提高了水印算法对裁剪、拼接、噪声等攻击的鲁棒性.Considering about the problem of high redundancy,low security and robustness of image as the watermark information,a watermark information generation algorithm was proposed based on compressive sensing.Firstly,original binary image was transformed into lifting wavelet domain and divided into blocks.Then each block was measured by part Hadamard measurement matrix in nonlinear way.The measuring result was binaried and embedded as watermark information.Watermark information was reconstructed by using orthogonal matching pursuit method to extract watermark from media data.Experiments have been conducted to verify the effectiveness of algorithm,which show that the proposed algorithm can reduce the redundancy,enhance the watermark information security and improve the watermarking algorithm robustness to attacking such as deleting,mosaic and adding noise.

关 键 词：水印信息 压缩感知 测量 提升小波 分块 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]