The Research On The Copy-move Forgery Detection Based On PCET Multi-radius

With the rapid and continuous development of multimedia technology, thehigh-definition digital camera and image editing software have been widely used in practicalapplication. It’s easy for someone to obtain an image and edit it according to their ownpreferences or some ulterior purpose, which reduces the credibility of images we obtain. Atthe same time, as one of the most typical and easy to operate tampering method in the diverseDigital Image Tampering and Authentication Technology, Copy-Move Forgery(Especiallywith the same image) make it easier for the criminals to tamper with the image content. Thus,this paper proposes a new solution on tamper detection with the same image, and the mainwork of this paper is as follows.Firstly, we introduce the research background, significance and status of digital imageauthentication technology. Aiming at the tamper detection with the same image, we evaluatethe advantage and disadvantage of state-of-the-arts in this field, and summarize the commonproblem of state-of-the-arts, that is, it’s hard to detect the pasted regions which have beenresized in large scales.To solve this problem, this paper proposes a new algorithm on the copy-move forgerydetection based on PCET Multi-radius. The contribution of this algorithm is as follows:(1)the scale invariance of Multi-radius PCET is proved with an intuitive method;(2) features areextracted using the GPU acceleration technology to make up the consumption of extraction ofmultiple scales;(3) the classic sorting algorithm is improved using the big top heap to avoidthe explosive growth of similarity matching and miss match;(4) an accurate detectionalgorithm based on radius and space information is proposed with the common used the sameaffine transformation selection(SATS) method, which reduces the miss match rate and isrobust to rotation and scaling;(5) the intuitive results of experiment demonstrate that theproposed method achieves better performance than other methods on condition of JPEGcompression and operations resistant to blur and noise, and the analysis of robustness indicatethat the proposed method is better than other state-of-the-arts on detection of rotation andscaling, especially in large scale resized.Last, we summarize the main work of this paper and point out the future work of theproposed algorithm.