Research On Digital Image Splicing Detection Based On Image Transformation And Support Vector Machine

With the development of science and technology,more and more powerful image editing and processing softwares are popularized in our lives,which can easily make digital images being tampered and edited without leaving any obvious traces to achieve visual fidelity.The digital image has no value when it is applied to the medicare reimbursement,police system and photography,but the false images will do great harm to the country and society,and cause incalculable loss.So the digital image forensics technology is a very important research topic nowadays.Image splicing is one of the most common and basic methods of image tampering and forgery.In this paper,the passive forensics technology is mainly discussed.By analyzing a large number of images that have been known to be true or false,the variation of image tampering sensitive feature operation in the process of stitching is found out,then the support vector machine as a classifier for splicing detection is used.The main contents and results are as follows:1.The background and significance of digital image forensics technology are introduced.The current research status of digital image tampering,forgery and digital image forensics technology of China and foreign countries are discussed2.The basic theory and related technology of digital image splicing forensics are introduced,including the process of image splicing,the basic frame of the passive forensics system,the image transformation,the support vector machine theory and some specific methods of typical passive forensics technology;3.An image splicing detection method based on DTT transform,DCT transform and least squares support vector machine is proposed.Firstly,DTT and DCT are used for image respectively,and Markov state transition probabilities matrix is extracted from DTT and DCT domain as splicing feature.The least squares support vector machine is established to predict whether the image is splicing or not.Experiments are conducted in the CASIA Tampered Image Detection Evaluation Database and Columbia Image Splicing Detection Evaluation Dataset respectively.Compared with the traditional algorithm,the experimental results show that the proposed algorithm has a better detection accuracy.4.An algorithm of image splicing detection based on dual tree complex wavelet transform and least squares support vector machine is proposed.Firstly the testing images are decomposed by using dual tree complex wavelet transform to get different sub-band images.Then Markov state transition probabilities matrix is extracted from those sub-band images as splicing feature.At the last step the least squares twin support vector machine is established to determine whether the image has been spliced or not.The experimental results in the Columbia image splicing detection evaluation data sets show that the proposed algorithm has a higher accuracy on splicing detection compared with the traditional algorithm.