利用异常边缘进行图像锐化篡改取证

为实现图像的完整性和真实性鉴别,提出了一种针对人工锐化的取证方法.利用非抽样Contourlet变换分析图像边缘点,进行边缘点分类;通过研究锐化前后预测误差图像和相位一致性来统计正常边缘点与锐化边缘点之间的差异,提取特征,训练分类器;利用SVM标定人工锐化痕迹.实验表明,该方法能够有效地检测出图像人工锐化篡改操作,较为准确地定位图像篡改边界.

Exposing image sharpening forgeries by abnormal edges

In order to distinguish the integrity and authenticity of digital images,a method to detect sharpenings from the tamped image was proposed.Firstly,the image edges were analyzed and claassified usingNonsubsampled Contourlet transform.Then,by researching the prediction-error image and phase congruency,the differences between the normal edge and the sharpened edge as features were extracted,bywhich SVM could be trained.Finally,the artificial sharpening traces could be distinguished through classifier.Experimen...