Identification method of coping digital image

Abstract

The invention belongs to the multimedia information safety field and particularly relates to an identification method of a coping digital image. The method comprises the following steps that: (1), an image classification model is established; more particularly, an image library is established and a copying image and a natural image are stored in the image library; discrete cosine transform (DCT) coefficient matrixes of all images in the image library are extracted and DCT coefficient first effective digit distribution characteristics of all the images are extracted; and a classifier is trained to obtain an image classification model; and (2), the image classification model is utilized to treat a detected image to carry out coping image identification. According to the invention, on the basis of a DCT coefficient first effective digit distribution change during a secondary image obtaining process, statistics of the DCT coefficient first effective digit distribution characteristics is carried out and the characteristics are input into the classifier to carry out study, so that the image classification model is obtained; and the classification model is utilized to analyze a digital image, so that it can be effectively identified whether the digital image is a copying image or not.

Description

technical field [0001] The invention belongs to the field of multimedia information security, and in particular relates to an identification method of a reproduced digital image. Background technique [0002] At present, digital cameras are becoming increasingly cheap and high-resolution, and image processing software is becoming more and more easy to operate, which creates conditions for digital image fraud, and the authenticity of images seen in daily life is greatly reduced. In view of the increasing phenomenon of digital image forgery, the research of digital image forensics technology is particularly important. In the prior art, digital cameras, cameras and other shooting equipment are used to remake books, diagrams, paintings, photos, etc., and the remake images are of high quality, and the images themselves are imaged by digital cameras or cameras, which may cause disputes with the original owners of the images, and this Similar images are often able to confuse peopl...

AI Technical Summary

Problems solved by technology

In the prior art, digital cameras, cameras and other shooting equipment are used to remake books, diagrams, paintings, photos, etc., and the remake images are of high quality, and the images themselves are imaged by digital cameras or cameras, which may cause disputes with the original owners of the images, and this Similar images are often able to confuse people's eyes, and some existing forensics systems are not easy to identify remake images

In addition, some existing object recognition systems, such as face recognition systems, image forensics systems, etc., may be attacked by remake images. How to detect and exclude remake images in these systems is extremely important to ensure the security of object recognition systems

Images