The Passive Blind Forensics Technology Research Of Forgery Image

Along with the development of digital technology represented by computerinformation, people’s life has entered the digital age; especially the advent of digitalcamera products make that digital image has been widely spread in our daily life. In orderto meet the general needs of consumers, some software such as Adobe’s Photoshopsoftware has made many ordinary non-professional user can do any changes with yourimages to achieve their purpose. All pros and cons, because the properties of digital imageseasy to be modified, the increasingly rampant of forged image in the social life seriouslyaffected the social morality, fairness, justice and so on, caused serious adverse effects inthe society. Under such a background, the passive blind forensics technology about digitalimage has become a very important research subject.With passive blind forensics technology about digital image, some researchers haveproposed many detection algorithms about the technology. In this paper, we proposed twoalgorithms, the first kind is to respond to that the camera resolution is more and more highdemand, put forward a kind of based on principal component analysis (PCA) detectionalgorithm of the forgery image for JPEG image. The second algorithm is a preliminarylocalization algorithm based on the copy-and-paste forged image.The article first detection algorithm for JPEG images forged features, because any aJPEG image after forging generally is compressed again as a different way from theprevious way, and it makes an image be compressed twice, for this situation, this paperproposed double JPEG compression detection based on principal component analysis(PCA) on the basis of predecessors. Algorithm firstly use linear smoothing filter for imagenoise reduction, and then extract the low-frequency components of three-leveldecomposition through wavelet multi-scale transform, principal component analysis (PCA)dimensionally reduce on low frequency component, and extracts the matrix of discretecosine transform after dimension reduction, drawing the transform coefficient histogram,whether has jitter mode through histogram, testing whether it be forged. The algorithmimproves the detection efficiency on the basis of predecessors, and reduces thecomputational complexity, largely adapting to the characteristics in today’s increasinglylarge amount of image data, is very practical. The second algorithm for copy-and-paste tamper with the rough location, on the basisof predecessors, the algorithm firstly do wavelet domain threshold shrinkage algorithm forimage noise reduction, and then extracted low frequency component of the image by usingthe secondary wavelet decomposition, which effectively reduces the image noise andunnecessary data. Then the image is partitioned in low-frequency component and do phasecorrelation analysis and determination, through the derived displacement of imagetranslation which determine image forgery area. The experimental results show that thealgorithm is able to achieve accurate positioning forge area, has good robustness withnoise.