摘要
提出一种支持向量回归机SVR的数字图像水印算法.选择载体图像平滑区域中心像素及其7×7邻域做为SVR的训练样本,利用SVR的学习和泛化能力,建立图像空域像素邻域之间的SVR模型,通过调整中心像素值实现水印嵌入.实验结果表明,该算法具有良好的不可感知性,对JPEG压缩、椒盐噪声、高斯低通滤波、模糊、锐化、剪切、尺寸变化等攻击有较强的稳健性.同时提取水印过程不需要原始载体图像,可实现水印的盲检测.
An digital image watermarking algorithm based on support vector regression(SVR)is proposed in this paper.Firstly,7×7 neighbors of center pixel in smoothing areas is chosen as training samples.Then,the SVR model between the neighbors pixels and center pixel in spatial domain is constructed utilizing the good learning ability and generalization ability of SVR.Finally,the watermark is embedded by adjusting the value of center pixel.Experimental results show that this algorithm has the good unperceptibility and the strong robustness for the JPEG compression,the spiced salt noise,Gauss lowpass filter,blur,sharpen,cropping and size transformation.The algorithm is a blind watermarking because of the process of extracting watermark without the original digital image.
出处
《大庆石油学院学报》
CAS
北大核心
2008年第6期106-108,114,共4页
Journal of Daqing Petroleum Institute
基金
黑龙江省科技厅科技攻关项目(GZ06A102)
黑龙江省教育厅科学技术研究项目(11513007)
