图像内容感知缩放的检测方法研究

For the content-aware scaling technology that can be used for image tampering, a content-aware scaling detection algorithm based on the statistical features of the probability Map is proposed. This algorithm uses the probability Map to reflect whether an image has undergone a content-aware scaling operation and uses the newly proposed integral projection and local statistical features to detect the tampered image. Then, a classifier is used for classification training, thus effectively identifying image tampering based on content-aware scaling operations. The experimental results show that the proposed algorithm can distinguish between the original image and the tampered image and has a relatively high correct detection rate.

针对可用于图像篡改的内容感知缩放技术,提出了一种基于概率Map图统计特征的内容感知缩放检测算法.该算法利用概率Map图以反映图像是否经过内容感知缩放操作,并利用新提出的积分投影与局部统计特征来检测篡改图像.而后利用分类器进行分类训练,从而有效识别基于内容感知缩放操作的图像篡改.实验结果显示,所提算法能够区分出原始图像与篡改图像,并具有较高的正确检测率.