Utilizing Sensitive Features for Image Tampering Detection

Image tampering has brought a great negative impact on society. People who do not know the truth are easy to be misled and used by people with intentions. Its impact on society has attracted extensive attention of Chinese and foreign researchers. With the development of image editing technology, image tampering detection has made remarkable achievements from the initial naked eye recognition to machine learning recognition of tampered images, and then to the excellent characteristics of deep learning algorithms in processing images in complex image tampering detection in recent years. In view of the outstanding performance of Faster RCNN network in the field of target detection technology, this paper studies an image algorithm and detection method based on Faster R-CNN network, and adds position sensitive candidate area pooling on this basis. In each rectangular cell, the pixel values of a fixed number of sampling points at a fixed position are calculated for average pooling. The input tampered image is feature extracted through the feature extraction network. After the extracted feature map passes through the RPN layer, the tampered area candidate frame is automatically generated for subsequent classification and regression. Experimental results show that this method greatly improves the effect of image tampering detection.