Deep multi-scale network-based secondary JPEG compressed image forensics method

Abstract

The invention provides a deep multi-scale network-based secondary JPEG compressed image forensics method. The invention aims to improve the accuracy of image forensics. The implementation process of the method include the following steps that: N DCT coefficient histogram features of a JPEG image to be subjected to forensics are extracted; four deep neural networks are trained; the preliminary tampering detection result of the corresponding data block of one DCT coefficient histogram feature in the JPEG image to be subjected to forensics is obtained; the final tampering detection result of thecorresponding data block of the one DCT coefficient histogram feature in the JPEG image to be subjected to forensics is obtained; the final tampering detection results of the corresponding data blocksof the other N-1 DCT coefficient histogram features in the JPEG image to be subjected to forensics are obtained; and the forensics result image of the JPEG image to be subjected to forensics is obtained. The method of the invention can be used for fields such as news photography authentication, judicial authentication, insurance authentication and bank electronic bill authentication.

Description

technical field [0001] The invention belongs to the technical field of image processing, and relates to a second JPEG compressed image forensics method, in particular to a second JPEG compressed image tampering detection method based on a deep multi-scale network, which can be used in the field of second JPEG compressed image forensics. Background technique [0002] With the rapid development of image acquisition tools and the popularity of social media, digital images are widely used and become the mainstream information carrier. With various image processing tools, people can easily modify the image to any desired content. In many fields such as journalism, law, business, medical applications, and academic research, the credibility of visual images has been compromised by digital technology. Therefore, digital image forensics, which aims to identify the original source of an image or determine whether the image content has been modified, has become particularly important....

AI Technical Summary

Problems solved by technology

However, because this method does not fully consider some statistical characteristics of the secondary JPEG compression, it simply extracts the features by using the convolutional neural network, and the extracted information is less, and the consideration of the situation is not comprehensive enough. When the JPEG image is the first When the secondary compression quality factor is greater than the second compression quality factor, there is no effective solution, and the accuracy of image forensics is low

Images