Study On Intelligent And Robust Algorit Hm For Image Forgery Detection In Digital Forensics

In today's world,the development of multimedia content is moving towards to digital and intelligent direction.As an import.ant form of information storing,digital image are applied more and more widely.Image editing operation becomes drastically simpler by using modem digital image processing software,therefore has brought great convenience for people's work and life.However,there are also some negative effects,such as the forgery and tampering of the image content,which is more easily operated.Therefore,it is particularly essential and urgent to study the forensics technology for the identification of authenticity and credibility of digital image content.Copy-Move is a common image tampering operation type.The existing detection methods are not robust enough to resist geometric rotation and scaling,and the detection process lacks intelligence.This research project focus on Copy-Move forgery,combining with Polar Harmonic Transform(PHT)in the field of signal processing,Particle Swarm Optimization(PSO)algorithm from intelligent optimization domain,and Support Vector Machine(SVM)in machine learning and other techniques,developed an effective solution for detecting and locating copy-move tamper region.The proposed forensics method can achieve intelligent detection,and implement strong robustness against geometric attacks and post-processing operations.The specific work of this project is as follows:(1)An image copy-move forgery detection(CMFD)algorithm against geometric attacks based on PHT moments is proposed in this paper.Firstly,the image endures some pre-processes,including overlapping circular block division.Secondly,the PHT moment coefficients of each overlapping image block are calculated as the invariant feature,and Singular Value Decomposition(SVD)is applied to reduce the dimension of feature.All the block features are gained and following the similarity distance calculation between every two blocks,PSO is used to find the optimal similarity threshold based on the histogram of block similarity me.asures.Extensive experimental results reveal that this algorithm can resist the geometric attacks such as rotation and scaling effectively.The presented novel scheme also has good robustness against post-processing operations such as Gaussian noise and JPEG compression,and at the same time improves the forgery detection the existing methods,this technique has better performance in higher detection accuracy and lower computational complexity.(2)This paper proposes an intelligent image copy-move tamper detection algorithm based on SVM classification.PECT features are extracted from the overlapping circular blocks of images,and the Histogram of Block Similarity Measures(HB SM)of each image is obtained.Appropriate features are selected to distinguish natural images from copy-move tampered images.The experimental results show that this method has a high accuracy against geometric and post-processing attacks.(3)In this research proj ect,the existing blind image authentication technologies for copy-move forgery is compared and analyzed thoroughly,and the tamper detection mechanism is studied based on this.The manipulation model target at the geometric transformation in the tamper process is established.The unified supplements are added to the open database in order to complete the simulation experiments.The tamper type includes rotation with different degrees,scaling with various scaling factors,and post-processing operations like Gaussian noise,filtering and JPEG compression.