Copy-move Image Forensics Algorithm Based On Block Feature Vectors Matching

With the rapid development of computer image processing software,image tampering is becoming increasingly common.In major fake image events,some false images can be found by copying an area of the image and pasting it to another area of the image to hide the original objects in the image.Some false images can be found by adding an object that is not the part of the image to enhance the emotions which the image itself wants to convey.However,people look through the copy-move tampering image with the naked eyes and do not see any obvious tampering marks.Therefore,it is becoming gradually important to give people a truth by identifying the authenticity of images through scientific means.However,any obvious tampering marks of the copy-move tampering image can not be seen with the naked eyes.This paper analyzes digital image passive forensics algorithms and summaries the passive image forgery algorithms based on feature points and overlapping blocks based on digital image copy-move tampering.Meanwhile,the advantages and disadvantages of the existing algorithms are analyzed.For the current image copy-move tampering passive forensics algorithm based on overlapping blocks,a large amount of redundant matching will be brought in the step of generating the feature description of the block,resulting in a problem of higher algorithm time complexity and similarity in itself.There is a certain problem of misdetection in the tampered image with similar areas and small tampered areas.A kind of copy-move tampering passive forensics algorithm based on feature block and R-HOG is designed in this paper.Firstly,the dense feature blocks are extracted to reduce the matching redundancy.Then the feature points corresponding to the preliminarily matched feature blocks are filtered by combining the color information and finding the main affine transformation.Finally,the regional correlation is used to quickly locate the copy-paste tampering area with high accuracy.Experimental results show that the proposed algorithm has high detection accuracy and low detection complexity for copy-move tampered images with similar areas and small tampered areas.However,flip-type copy-move tampered images of the tampered areas cannot be detected.In order to more effectively locate the tampering region of the flip-type copy-move tampered images,a kind of copy-move tampering passive forensics algorithm based on the "rice" font of GLCM feature vector is designed in this paper.Based on the color imagepreprocessing,the trait of each block is described by using the "rice" font of GLCM feature vector with certain angular rotation invariance and multiple flip invariances.Meanwhile,the way of counting the number of other marker points around the detection marker in the image is used to remove the error marker in the non-tampered region.The experimental results show that the algorithm has higher detection accuracy for certain angular rotation and multiple flip-type copy-move tampered images.Finally,a set of digital image copy-move tamper passive forensics simulation system is designed by using MATLAB.The algorithms of this paper and the algorithm of the comparative literature is presented.