Research On Digital Image Forensics Algorithm

Image forensics technology is the use of the statistical properties of the image to determine the integrity of the digital image content,original and authenticity to determine the digital camera's digital image technology has been tampered with.Image forensics technology has been studied,the main work is as follows:(1)Research on the blind detection based on the noise correlation method.The paper presents a digital Image forensics technology based on the noise correlation between the digital image and digital camera.Firstly,to extract the digital image noise margin by wavelet filtering,then using multiple images average noise as reference model of the camera;Second,we calculate the child window correlation coefficient of the target image noise margin and camera noise reference model noise,according to the correlation coefficient we can determine whether the target image has been tampered with.(2)Research on the copy-paste image blind detection method.A mixed feature blind detection algorithm based on texture feature and invariant moment is presented for copy-paste image forensics.The algorithm extractes three texture features,as well as seven moment invariants to form a rotation scaling invariant feature vector for detection feature vector within the image sub-region to determine whether there is copy-paste tampering.(3)Research on the blind detection methods for people tampered image.A fast blind detection algorithm for people tampered image is presented.The face region is obtained by the face detection method.Appropriate expansion of the detected face region as a region to be detected,the people image is determined whether it has been tampered by the aforementioned two the image blind detection methods.The experimental results show that this method can reduce the search area,effective to detect the the copy-paste tampered image within the face region,the method also have some effect for the face-lift type tampering.