Research And Implementation Of Copy-move Forgery Image Passive Detection Algorithm

With the rapid development of technology,digital images have been integrated into people's lives in every way,helping people to record all kinds of beautiful and unforgettable moments in life.At the same time,with the increasing development of image processing technology,image editing softwires have been easier-to-use and more powerful,so people with image processing skill can easily modify the image content by the image editing softwires.For ulterior purposes,however,people with evil intentions will use the image editing softwires to modify the image content,making the experts not to distinguish between true and false.This will have a serious influence on the long-term stable development of the society and country.How to ensure the credibility and authenticity of digital images has become a top priority.To solve this problem,digital image forgery detection technology is proposed in the field of multimedia information security.The digital image forgery detection technology applies the inside or outside image information to detect the image and estimate whether the image is forged or not.Among the image forgery methods,copy-move forgery is one of the forgery methods,which is the simplest and most difficult to be discerned.Therefore,a solution for copy-move forgery detection(CMFD)is proposed in this paper.This paper contains four parts as shown in follows.1.The research background and significance of image forgery detection are introduced,and the state of current research is discussed in the introduction of this paper.The current image forgery detection technology can be divided into two types: active detection and passive detection.CMFD method is one of passive detection methods.The advantages and disadvantages of the CMFD method are analyzed,from which the existing problems of most methods are obtained.2.The module of copy-move forgery image is introduced and analyzed its characteristic.Furthermore,the copy-move forgery image detection method is divided into two types: the block-based copy-move forgery image detection algorithm and the keypoint-based copy-move forgery image detection algorithm.The two types of classic CMFD algorithms are introduced and analyzed to obtain the current problems.3.To solve the current problem,the low robustness against rotation and scale in CMFD method,a CMFD algorithm against rotation and scale is proposed in this paper.The algorithm content contains four parts: a.the improved Analytical Fourier-Mellin Transform(AFMT)as an image block feature has rotation and scale invariance in theory;b.the Local Sensitive Hashing(LSH)method is applied to match the similar image block features,which can be used to improve the matching efficiency;c.a post-processing operation combining Euclidean distance with block variance is proposed to filter out most of the false detected image region.In the qualitative analysis,the detection results of the proposed algorithm have satisfactory performance,even if the image suffers from rotation,scale,JPEG compression or additional Gaussian noise attack.And in the quantitative analysis,the proposed method outperforms the other block-based and keypoint-based methods.4.The research contents of this paper are systematically analyzed and made conclusions.Besides,this paper also points out the reformative aspect in the future.