Research On Blind Identification Of Regional Copy-Move Forgery In Digital Image

Nowadays, with the popularity of digital image acquisition equipment, particularly digital cameras and the rapid development of Internet information technology, it is an indisputable fact that digital image become an important carrier of transmission and storage of visual information. It not only makes modern people's lives more colorful, but also brings a lot of confusion and pain. Some powerful image processing and editing software can be easily carried on digital image tampering and forgery, to see no longer reality. If tampering and forgery image is used in some official media reports, scientific discovery, accident insurance and court exhibits and even culture, political, military, diplomatic medium, to exaggerate or distort the truth and fabricate facts in order to achieve the ulterior purpose, which will have serious negative effects on the economy development and social progress, and it can not be estimated. For example, in recent years, "The Tibetan Antelope" photo event, "The Square Pigeon" photo events in china take very serious negative effects in society."We live in the world where we no longer believe what we have seen and listened" the professor of the Dartmouth University in American ,the expert on the authenticity of photos, Hany Farid once said with such mixed feelings. Image authentication is by analysising image statistical properties to determine originality, authenticity and integrity of the image. In the image authentication, image blind authentication technology is more realistic and challenging.This paper outlines the background of image authentication, research status, the main image tampering ways and classical blind authentication algorithms. The copy-move tampering which is one of the popular tampering ways is researched and analyzed in the article. For the copy-move tampering in the same image, the paper proposes a fast and accurate blind authentication algorithm. The algorithm uses wavelet decomposition to reduce the overall image search space, makes use of region growing, list-query and some optimizations to solve the problem of large amount of calculation and long time in the authentication process. And in the experiment, it can be proved that it is a fast, accurate and robust algorithm to detect the copy-move tampering region.The process of the copy-move tampering in different images is to select a image as the background image, and extract the key parts from other images, to paste it into the proper position of the background image to generate a forged image. In fact, in order to eliminate visual or statistical aberration of the image copy-move part edge, the forger will feather, fuzzy the edge of copy-move parts of the image. It is hiding the truth of the image tampering which will bring in the other tampering truths. By analysising image edge features and the characteristics of artificially fuzzy, we design a new fuzzy edge detection algorithm based on homomorphic filtering in wavelet domain and mathematical morphology which can accurately detect and locate the copy-move tampering region.