摘要
利用数码相机拍摄自然图像时,往往存在彩色滤波阵列(CFA)插值效应,CFA插值使得图像三颜色分量之间具有很大的相关性。提出一种利用自然图像颜色通道之间的相关性来检测图像篡改操作的方法。首先,利用后验概率的方法对各颜色通道进行建模,从而准确估计出CFA插值产生的噪声;然后,利用高斯滤波器滤波并提取CFA插值特征;最后,根据提取出的特征进行分类并定位篡改区域。实验结果表明:该算法能够有效地对篡改的图像进行检测和定位。
When using digital camera to get natural image,there exists CFA interpolation effect,it makes great correlation between three color channels. Propose a novel method to detect tampering operation using correlation between nature image color channels. Firstly,use method of posterior probability to build model for each color channel to accurately estimate noise generated by CFA interpolation;then,use Gaussian filter for filtering and extract CFA interpolation features;finally,classify according to extracted features and locate tampering area. Experimental results show that the proposed algorithm can effectively detect tampering image and locate.
出处
《传感器与微系统》
CSCD
2015年第6期141-144,共4页
Transducer and Microsystem Technologies
基金
国家自然科学基金资助项目(61172160)
湖南省自然科学常德联合基金资助项目(12JJ9019)
关键词
彩色滤波阵列
插值
后验概率
定位篡改区域
color filtering array （CFA）
interpolation
posterior probability
localization tampering area
