Research And Application Of Blind Forensic Technology On Digital Image Splicing Tamper

In recent age, digital image has been becoming one of the most important digital media to carry and communicate information. The flourishing digital image processing technology brought up a mass of simple and powerful image processing software. Image processing technique is used to beautify people's life. Meanwhile, a series of image forgery which is in purpose of distorting truth happen frequently, which have bad effects on society. In view of these, the development of digital image forensics technology is increasingly imperative. And the blind forensics that based on the content of image is the hot research area nowadays.On the basis of the research of digital image blind forensics technology,a deep study has been done on blind forensics of image splicing forgery, which is in a different image, in this paper. The main work includes three aspects as follow:Firstly, the flow of blind forensics on image splicing tamper is proposed by analyzing the method of this field. And the main approaches, used in the each courses on the flow, are studied. Firstly, the extract methods of researching feature, especially the ones of Markov and gray level co-occurrence matrix feature extract, which are popular, are researched. After this research, the details of two feature extraction process and the implementation of the note parameter settings are understand clearly and coding feature extraction part independently. Then the classified algorithm of Support Vector Machines is studied by realizing the coursed of arithmetic realization and the LIBSVM arithmetic realization and learning SVM training, cross-validation and grid search.Secondly, through the study of exisiting image splicing detection algorithm found there are single section and high computational complexity problems on method to detect. And the study found that in all parameter optimization method, the convergence speed characteristics of particle swarm optimization can reduce computation time. Therefore, an image splicing detection algorithm based on particle swarm optimization is proposed. And it is demonstrated from both theoretical and experimental ways.Finally, a blind forensics system is designed and implemented. The system is able to describe the implement process of the flow of blind forensics on image splicing tamper. Firstly, two courses of feature extraction and a process of image splicing detection are showed. Secondly, identified cause of SVM and fit curve of particle swarm algorithm are showed.