Research On Image Copy-Move Passive Forensics Algorithm Based On Feature Points And Localization Algorithm

With the popularization of computer technology and the development of information technology,digital images are widely used,while the probability of being tampered is also greatly increased.Once image tampering is used by lawbreakers with ulterior motives,it will greatly affect the credibility of digital images and bring many negative effects to people's social life.Among many tampering methods,copy-move tampering is a very common tampering method,which can achieve the purpose of emphasizing or hiding by copying and pasting part of the same image.Digital image forensics technology is to judge whether the image has been tampered and locate the tampered areas by computer technology.This paper mainly studies the traditional passive forensics algorithm copy-move image forgery passive forensics algorithm.First of all,this paper summarizes the research background and significance of digital image forensics.It focuses on the classification and development of copy-move image forgery with passive forensics.Then,several classic passive forensics algorithms are introduced.Finally,in view of the shortcomings of the existing copy-move forgery passive forensics algorithm,such as complex features,the large time cost of calculation,and the inability to accurately locate tamper areas based on feature points,two novel improved algorithms are proposed.In order to solve the problems of existing algorithms,this paper proposes a copy-move forgery detection method based on binary descriptors and density clustering algorithm.Firstly,the method uses the binary descriptor AKAZE to detect the feature points and describe the features,then uses the density based clustering algorithm DBSCAN to remove false matches,finally,the final tampered area is located by comparing the similarity between the original image and the image after affine transformation.This method can effectively reduce the time complexity by using binary features,and has good detection effect for various tampering methods.In order to solve the problems of large computation of high-dimensional complex features and inaccuracy of tamper location to pixel level,this paper proposes a passive forensics algorithm based on feature points clustering and similar neighborhood search location for digital image copy-move forgery.This method uses SIFT to extract feature points,and clusters the feature points based on scale and color.Then feature matching is carried out in each cluster to reduce the number of feature comparisons.Finally,a similar neighborhood search algorithm is proposed to search and mark the matching neighborhoods,and locate the final tampered area at the pixel level.This method reduces the calculation cost of high-dimensional complex features in the matching process,can effectively detect geometric transformation and post-processing tampering,and has good performance under various tampering attacks.