The Analysis And Forensic Research Of Heterogeneous Image

With the widespread use of digital image, the authenticity and safety of the image has become an important research topic. Image processing software become more and more mature, making more and more people can tamper images easily. Digital image forensics technology is one of the branch of information security technology, in recent years have made good progress in the laboratory environment. Under the condition of actual network, however, a given unknown image is difficult to obtain ideal detection results since it come from a mixture of various sources and various categories of heterogeneous mixed image. In order to solve this problem, this paper analyzes the features of the heterogeneous multiple images, combining with the image compression history and digital image tampering analysis, and achieved the following results:First of all, the situation of heterogeneous mixed image was studied and introduced, including the research background and research status, etc. Analyzes the application in field of digital image forensics, including detecting tampering and steganalysis. And according to the properties of the mixed heterogeneous image, put forward a heterogeneous image hierarchical forensics system framework.Secondly, this paper put forward a algorithm for detecting lossless image compression history based on the image information compression amount. By measuring the information loss between the compressed image and the original uncompressed image and setting a threshold, image compression history can be detected. The experimental results show that the algorithm is little affected by image content and size.Thirdly, this paper put forward a algorithm for detecting lossy image double compression based on Markov model for DCT coefficients. By further analyzing the compression block artifacts, we proposed a tamper region localization algorithm. The experimental results show that the algorithm can effectively detect double compression operation, and has a great improvement than previous work, and can effectively locate the tampering region, and robust with the subsequent processing such as scaling, vague operations.