Research On Passive Forensics Of Digital Image Copy Move Forgery

With the popularization of multimedia technology,digital image forgery is becoming more and more frequent,which brings great negative influence to the society.As one of the important technologies of digital images content authenticity protection,passive digital image forensics technology received a great deal of attention.In many means of image forgery,copy-move poisoning is one of the common ways.To detect this kind of forgery,researchers have proposed many methods,but the detection capability and effciency of these methods for the images with several copy-move forgery need to be improved.In order to solve this problem,this paper puts forward two new detect methods.First of all,the paper proposes the multiregional copy-mzove forgery detection method based on SURF algorithm.Considering the high robustness of SURF algorithm in anti-rotation,scaling,etc,firstly,the images to be detected are converted from RGB to LAB color space and the SURF features are extracted;Then a feature matching method named k-g2 NN is designed to match the obtained feature points.This method improves the ranking method of the existing g2 NN method.This method improves the threshold setting and ranking method of the existing g2 NN method and reduces the complexity of the algorithm.Finally,a geometric constraint-based clustering method is used to eliminate the mismatch to locate the tampered area.The experimental results show that this algorithm not only can locate multiple copy and paste tampered image regions accurately,but also has robustness on anti geometric transformation attacks and JPEG compression,and the detection speed has also been greatly improved.Then,to further improve the detection speed of the algorithm and reduce the storage consumption,a D-ORB-based copy-move forgery detection algorithm is proposed.The algorithm combines DWT algorithm with ORB algorithm,and then extracts the feature points from the LL component of the image processed by DWT.This process reduces the image dimension and improves the speed of feature extraction;then uses k-g2 NN algorithm to do the feature matching;Finally,the RANSAC algorithm is used to remove the key points of the mismatch.Experiments showed that this method can further improve the detection efficiency,save the memory consumption,has excellent detection effect on multiple-area copy-move forgery images,and has strong robustness to disturbing factors such as rotation and noise.