基于背景噪声的图像盲篡改检测  

Image Forgery Detection Using Characteristics of Background Noise

作　　者：刘丽娟[1,2] 林小竹[1] 

机构地区：[1]北京石油化工学院信息工程学院,北京102617 [2]北京化工大学信息科学与技术学院,北京100026

出　　处：《计算机科学》2014年第B11期136-138,149,共4页Computer Science

基　　金：北京市属高等学校人才强教计划资助项目(PHR201107146)资助

摘　　要：数字图像在成像过程中会产生特定的背景噪声,如果两幅不同噪声的图像拼接在一起,篡改区域和其他区的噪声会有差异。提出一种基于偏度统计特性的背景噪声估计算法,其通过对图像分块计算每块的噪声标准差,从而检测出噪声异常部分以达到篡改检测的目的。算法利用DCT变换去除原图细节部分,利用偏度统计特性估计噪声,利用条件最小值法求出噪声的标准差。算法改进了迭代求条件最小值法,利用微分方法求取最小值,避免了初始值设定问题,提高了算法的准确率。实验结果表明,提出的噪声估计算法正确率高,且对拼接篡改图像篡改检测有明显效果。There is a part of background noise in digital image which comes from imaging process.The noise characteristics between image forgery area and the other area are different if images with different noise levels are spliced together.This paper proposed a background noise estimation algorithm based on the statistical properties of skewness.We detected the forgery parts by dividing image into some sub-blocks and computing the noise variance of each ones.This algorithm removes the original image details with DCT transform,estimates noise with the statistical properties of skewness,and estimates the standard deviation of noise with condition of the minimum method.This algorithm improves iterative conditional minimum value method using differential method to calculate the minimum value.This algorithm avoids the problem of setting the initial value,and improves the accuracy of the algorithm.The experimental results show that the proposed noise estimation algorithm has high accuracy and effectiveness in detecting forgery part in spliced images.

关 键 词：图像取证 背景噪声 偏度 标准差 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]