一种鲁棒性强的视频图像在线识别算法  

An Online Identification Algorithm for Robust Video Image Watermarking

作　　者：杨琼[1] 王家全[1] YANG Qiong WANG Jia-quan(Information Technology Department, Qiongtai Teachers College ,Haikou 5 71100, China)

机构地区：[1]琼台师范高等专科学校信息技术系,海南海口571100

出　　处：《内蒙古师范大学学报（自然科学汉文版）》2016年第5期739-743,共5页Journal of Inner Mongolia Normal University(Natural Science Edition)

基　　金：海南省高校教育教学改革研究项目(Hnjg2015-81);琼台师范高等专科学校课题(qtjg201312)

摘　　要：针对当前视频图像水印在线识别正确率低的问题,提出一种鲁棒强的视频水印在线识别算法.首先利用支持向量机对邻域像素之间的相关性进行估计,并根据估计结果进行水印的嵌入,然后采用支持向量机对邻域像素值之间的关系进行估计,并根据估计结果与水印嵌入位置的像素关系进行视频图像水印的识别.仿真实验结果表明,该算法嵌入的水印不仅具有不可见性,而且可以实现视频图像水印的在线检测,具有较高的实际应用价值.Video image watermarking is a kind of technology to keep the security of video images.It is a hot spot in computer security field.On the problem of low accuracy of online recognition of video image watermarking,a robust video watermarking algorithm is proposed in this paper.Firstly,this paper analyzes the current status of video watermarking,and points out the shortcomings of the current video image watermarking algorithm.Then we use SVM to estimate the correlation between neighboring pixels.Based on the estimation results,we use support vector machine to estimate the relationship between neighborhood pixels.The experimental results on the Matlab 2012 platform show that the proposed algorithm is not only invisible,but also can be used in video image watermarking.

关 键 词：视频图像 视频水印 支持向量机 鲁棒性 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]