Blind Identification Algorithm Of Copy-Move Tampering Area Based On ORB And Clustering

Digital image acquisition equipment has been around every corner of our lives,small to mobile phones,digital cameras,large to city monitoring system,remote sensing satellites in space.Digital images make the traditional image acquisition equipments gradually fade out of the public view by fast,economical,maintaining a good balance between quality and price.However,compared with the traditional image,the digital image has the drawback that is more easily being tampered.In some cases require serious evidence,such as news reports,forensic evidence,etc.,in the face of ulterior motives of criminals malicious tampering,this drawback is infinitely magnified.Especially in recent years,digital image tempering events occurred one after another,tempering means more and more hidden.The credibility of digital images has been greatly affected.In this large context,image forensics technology came into being.Digital image forensics technology is a judge whether the digital image is maliciously tampered with,and further speaking is to detect how to tamper,and even by technical means,we can determine which areas have been tampered with.Digital image forensics techniques are summarized as two broad categories: Active forensics technology in the digital image generation process,adding the watermark or signature generated by the technical means and other a priori information,if the image after tampering,we test the integrity of these a priori information,you can simply judge the authenticity of digital images.The passive forensics technology,is a kind of image information itself,through the analysis of graphics information to determine the authenticity of digital image technology,also known as blind forensics technology.Which often involves the graphics,computer vision,pattern recognition,machine learning and other disciplines,is a comprehensive forensic technology.This paper begins with the research significance and present situation,and introduces the development of digital image forensics technology.Then,by introducing the generation mechanism of copy-move tampering image and the difference with natural image,the common blind recognition algorithm of copy-move tampering image is summarized.And the core algorithms used in this paper,the classic local feature description algorithm for a brief summary.Finally,based on the existing Copy-Move blind authentication algorithm,the paper presents its own improved algorithm and the evaluation model of tampering degree.And on the basis of the evaluation model,the use of regional growth method to complete the tampered area of accurate detection.Through the algorithm of this paper,we can make a more accurate assessment and measurement of the tampering area and tampering degree of Copy-Move tampering.In this paper,we propose a blind identification algorithm based on ORB and clustering for the Copy-Move region detecting.The existing blind recognition algorithm for the Copy-Move tampering image blind authentication algorithm has the problem of slow computation and inaccuracy.Firstly,use the Gaussian pyramid to resample the image,and the FAST corner detecting is measured on different scales of the image,so that the ORB algorithm has scale invariance.The feature point detection of the source image is performed using the ORB algorithm containing the scale information.Then,the ORB feature descriptor obtained by the detection is matched with Hamming distance,and the pair of points below the threshold is matched successfully.And set the minimum matching distance threshold,the initial results of the screening.Finally,according to the distance between the matching results and the clustering information,the core point and clustered points are determined,and the scale and angle information are determined by the spatial relationship between the core point and clustered points.And through the scale and angle information to further remove the wrong match point.Results show that the algorithm can achieve a better detection rate and effectively reduce the computation time.Aiming at the common feature-based Copy-Move tampering image blind identification algorithm can not accurately locate the tampered area or mark the edge of the tampered area.In this paper,an accurate detection algorithm of Copy-Move tampering region based on regional growth and the model of tamper evaluation is proposed.After obtaining the core point of the tampering area and the source area and the scale and angle information,we adopt a recursive method of the regional growth method to recursively match the points of the original region and the target area.After the recursive matching process for the tampering region and the source region,we can get the exact result of the tampering region and the source region.Results show that the algorithm can effectually locate the tampered area and determine the edge of the tampered area.