The Research On Copy-move Forgery Of Digital Image Forensics

Digital image as a favorable vector for the transmission of information ,because ofits large amount of information and image and other realistic features, widely exist invarious network environments and in real life. A variety of advanced image processingalgorithms and image processing hardware and software, making it easy fornon-professionals to handle images, and makes the naked eye is difficult if notimpossible to identify, but the image potential of the statistical characteristics havebeen modified. This has brought a lot of trouble to all walks of life, causing a varietyof crisis of confidence. Therefore, the image forensics technology becomes veryimportant, has become a very popular research direction.By analyzing the passive digital image forensics technology principles andcharacteristics of the theoretical background, algorithm focuses on the basic copy andpaste in the digital image tampering with the operation. For digital images, copy andpaste to tamper with the operation of the two detection algorithms.(1) Through the analysis of several commonly used moment features values totamper with the type of evidence collection algorithm based on SIFT and Hu moments,copy and paste. The experimental results show that the algorithm in the image afterrotation, scaling and processing of the case was still able to accurately detect thetampered region. Finally, the structure and processes through the analysis of thesealgorithms to replicate the basic framework of the paste tampered with type forensics.(2) Dealing with fuzzy operation may be the most commonly used in a real lifeimage after copy and paste tampering. Proposed in this paper, copy and paste the basicframework of tampering with evidence collection, the tampering with evidencecollection algorithm, based on fuzzy copy and paste the same moment. First image isdivided into multiple overlapping blocks, each block features with the calculated fuzzyby fuzzy invariant moments invariants. Then in accordance with a predeterminedsimilarity criteria to determine the image tampering area. The experimental results show that the anti-fuzzy has obvious advantages.