RHFM结合形态学滤波的数字图像Copy-Move篡改取证方法

摘要

For the image is easy to attack by copy⁃move forgery,a forensics method of image copy⁃move forgery based on radial⁃harmonic Fourier moment and morphological filtering is proposed. The detection image is divided into multiple overlapping circular area blocks,after that the invariant feature of the circular area blocks is extracted by means of radial⁃harmonic Fourier moment,and sorted in the form of dictionary. And then,the Euclidean distances among the feature vectors are calculated and compared to match the area blocks and obtain the detection graph. Finally,the window filtering and morphologic operation are used to remove the false matching results in detection graph to get the final detection graph. The experimental results show that the proposed method can effectively detect the copy⁃move forgery attack,and has good robustness for the attacts of forgery image affine transform and signal processing.%针对图像易受到Copy⁃Move篡改攻击问题，提出一种圆谐⁃傅里叶矩结合形态学滤波的图像Copy⁃Move篡改取证方法。首先，将检测图像分成相互重叠的多个圆形区域块；然后，利用圆谐⁃傅里叶矩（RHFM）提取出圆形区域块中的不变性特征，并对特征进行字典排序；接着，通过计算和比较特征向量间的欧式距离匹配区域块，获得检测图；最后，利用窗口滤波和形态学操作去除检测图中错误的匹配结果，获取最终检测图。实验结果表明，该方法能够有效检测Copy⁃Move篡改攻击，且对篡改图像的仿射变换和信号处理攻击具有很好的鲁棒性。