Research And Implmentation Of Image Copy-move Forgery Detection

With the development of digital media technology,it has become easier and easier to process and modify digital images using image editing software.When a digital image is maliciously tampered with and exploited,it may cause numerous negative effects and may even affect the stability of society,so digital images for tampering detection is an issue that cannot be ignored for image information security.There are various types of image tampering,among which copy-and-move image tampering is a common image tampering method.The traditional copy-and-move algorithm requires manual design features with certain limitations.With the development of deep learning,there are more and more algorithms in the field of image copy-move tampering detection based on deep learning,because deep learning can actively learn the connection between copy region and paste region from a large amount of data to perform more accurate detection.In this paper,we use a deep learning-based algorithm to detect digital image copy-move tampering and apply the algorithm to a digital image copyright management system to combine theory and practice.This paper proposes nested U-Net networks based on a progressive supervised attention mechanism to detect image copy-move tampering,with three main contributions: using Ushaped residual blocks instead of the ordinary convolutional operations in U-Net,and the residual blocks fuse rich multi-scale information;using a similarity matching module,which matches the multi-scale features obtained from the residual blocks for feature similarity;3 using a progressive supervised attention mechanism,which allows the network to focus on learning features with higher similarity,i.e.,filtering features with low match and removing false matches.The effectiveness of the algorithm for image copy-move tamper detection is verified through the experiments in this paper.This paper also designs and implements a digital image copyright system based on the image tampering detection algorithm as the core of the tampering detection algorithm.The system combines tamper detection with copyright management for the first time,and the tamper map generated by tamper detection provides more powerful reference evidence for image copyright traceability.