A method for authenticity identification of images

Abstract

The invention relates to a method for authenticating an image, comprising the following steps: the first step, image salience processing: the original image is segmented by the SLIC algorithm to obtain superpixels, and the contrast between each superpixel and the edge superpixel is calculated , to obtain a salient image based on the background; segment the original image according to the pixel threshold to select the foreground, and obtain a salient image based on the foreground by calculating the similarity between the original image and the foreground; fuse the salient image based on the foreground and background, and use Gaussian filtering The smoothing process is carried out by the filter to obtain the salient image of local features; the salient image at the pixel level is obtained; the texture enhanced image is obtained; the real and fake face image identification network is trained.

Description

technical field [0001] The invention relates to the field of image forensics, in particular to a method for identifying deep fake images. Background technique [0002] In recent years, with the development of generative confrontation network (GANs) technology, the birth of GANs-based Deepfake in 2017 opened a precedent for face replacement. Later, Deepnude and Zao software were also born for face replacement. These software can replace the selected face with any other person's face, and use it to create various videos such as pornographic clips that do not exist, and spread them on the Internet, polluting the network environment and seriously endangering personal information security. . What's more, rumormongers use deep forgery technology to publish false speech videos of leaders of countries and public organizations, creating hidden dangers for the world's public safety. Therefore, it is imminent to establish an effective identification system for fake images. [0003]...

AI Technical Summary

Problems solved by technology

[0003] At present, the research on the identification of true and false faces at home and abroad has shifted from the traditional digital forensics method to the research of artificial intelligence methods, and among the many patents currently published, few of them involve research on the identification of true and false faces in the direction of intelligence

Images