查询结果:   刘红梅,干彬,刘金华,赵明峰.基于奇异值分解和条件异方差的数字水印算法[J].计算机应用与软件,2018,35(5):253 - 257,268.
中文标题
基于奇异值分解和条件异方差的数字水印算法
发表栏目
图像处理与应用
摘要点击数
758
英文标题
DIGITAL WATERMARKING ALGORITHM BASED ON SINGULAR VALUE DECOMPOSITION AND CONDITIONAL HETEROSCEDASTICITY
作 者
刘红梅 干彬 刘金华 赵明峰 Liu Hongmei Gan Bin Liu Jinhua Zhao Mingfeng
作者单位
四川传媒学院 四川 成都 611745 电子科技大学电子工程学院 四川 成都 611731 中国移动通信集团设计院有限公司四川分公司 四川 成都 610045   
英文单位
Sichuan University of Media and Communications,Chengdu 611745, Sichuan, China School of Electronic Engineering, University of Electronic Science and Technology of China, Chengdu 611731,Sichuan, China Sichuan Branch, China Mobile Group Design Institute Co., Ltd., Chengdu 610045, Sichuan, China   
关键词
数字水印 奇异值分解 条件异方差模型 工作特性曲线 小波变换
Keywords
Digital watermarking Singular value decomposition Conditional heteroscedasticity model Operating characteristic curve Wavelet transform
基金项目
作者资料
刘红梅,讲师,主研领域:计算机图形学,计算机视觉。干彬,讲师。刘金华,高工。赵明峰,高工。
文章摘要
针对基于高斯分布的图像模型在水印嵌入和检测中存在的性能较差问题,提出一种基于奇异值分解和条件异方差的数字水印方法。给出一种运用小波变换和奇异值分解的加性水印嵌入方法。利用条件异方差模型较好地描述图像小波系数的“尖峰重尾”分布特性和异方差特性,来对小波系数进行建模。在此基础上,应用统计假设检验理论分析水印的盲检测过程,并推导水印检测中虚警概率和检测概率之间的工作特性关系。仿真结果表明了该方法的有效性,而且在诸如噪声、JPEG压缩、滤波、旋转以及缩放等攻击下具有较好的检测性能。
Abstract
Image model based on Gauss distribution has low performance on watermark embedding and decoding. To solve this problem, a digital watermarking method is proposed based on singular value decomposition and conditional heteroscedasticity in the wavelet domain. By using wavelet transform and singular value decomposition, an additive watermarking embedding method was given in this paper. The conditional heteroscedasticity model was utilized to model the wavelet coefficients by taking advantages of the important characteristics such as “sharp peak and heavy tailed” marginal distribution and the heteroscedasticity. On the basis of this, the blind detection process of watermark was analyzed by using the theory of statistical hypothesis testing. And the relationship between false alarm probability and detection probability was derived. Simulation results prove the effectiveness of the additive watermark embedding method, and it has better detection performance under attacks such as noise, JPEG compression, filtering, rotation and scaling.
下载PDF全文