利用颜色滤波矩阵估计原图的拼接定位盲取证方法

图像拼接是伪造数字图像最基本的操作，对其的被动盲检测和定位也是数字图像取证亟待解决的关键问题之一。该文通过对图像进行颜色滤波矩阵插值，对图像进行了原图估计，重建其像素邻域一致性。利用待取证图像的边缘点在像素邻域一致性上偏离估计的原始图像的程度，对图像的拼接进行准确定位。实验表明，该方法能够有效地检测图像拼接伪造区域的位置、大小和形状。

A Blind Forensics Method for Image Splicing Based on Original Image Estimation Using Color Filter Array Interpolation

Image splicing is the most fundamental operation of creating an image forgery, thus the detection and localization of image splicing attract increasing attention recently. In this paper, a blind and passive approach of splicing localization is proposed based on original image estimation. By color filter array interpolation, the original image of the test image is estimated and the coherence of the pixels’ neighbor is re-constructed. Using a comparison between the test image and the estimated original one in edge pixels, the localization is implemented. Experimental results show the good performance of the proposed method.