The Research On Region Copy-Paste Distort-Detection Of Digital Image

Nowadays, with the rapid development of powerful image processing software,it’s easier to distort digital image, which results in the transmission of informationfacing a great challenge of the authenticity. Digital image passive forensics as anemerging detection image authenticity technology, gradually attract the public’sattention. It needn’t add any information to the image in advance, only use thefeatures of the image itself, that makes it widely used. Regional copy and pastetampering is one of the most simple and effective digital image tampering. How tochoose the appropriate algorithm to solve the existing problems in the study, itbecomes extremely important.In the process of region copy-move forging, the forgers will do a series ofpost-processing to conceal the evidence of distorting, which cause the efficiency ofmatching the image area detection reduced. This paper focuses on this commonmeans of distorting, and bring up the copy and paste distorting detection algorithmbased on the moment invariant features. The moment invariant features transformwill not change the features after the rotation operation, so it can handle the rotationoperation effectively. This algorithm in this paper has a high robustness evenconfront the processed image after the rotation. It overcomes the deficiency whichPCA can’t detect the rotating operation.To conceal the evidence of distorting, the forgers will not only use the rotatingoperation to process the image, but also add Gaussian noise. To solve these twopost-processing problems, this paper brings up the copy-move distorting detectionalgorithm based on the shape of the edge direction histogram features. Thisalgorithm use the improved Canny method to extract the edge of the image, whichcan distinguish the edge points from the noise pionts effictively. In addition, becausethe histogram has translation and rotation invariance, the histogram featuresextracted based on edge direction also meet the translation and rotation invariance,and has a strong anti-noise ability. The experiment results show that the algorithm is robust against the rotation and the noise operation. It overcomes the shortcomingsthat the classic PCA algorithm can not detect the images after the rotating and thenoise operations.