Image Reproduction - Paste Fake Detection Algorithm Based On GLCM And GGM

With the development of image processing technology, a variety of image editing software is popularized and applied rapidly. It makes images easy to modify and people can more easily achieve the image that is beautified and edited. In reality, however, some people may tamper the image wantonly, even it will be transmitted illegally via the internet to disturb the social order. If these images to be used in the military, judicial and other official sector, it will bring negative effects greatly. In order to identify these tampered images, digital image forensics technology gradually comes into people’s horizons. Now, the digital image forensics has become one of research hot spots of image processing, especially the passive digital image forensics technology become a research focus.Copy-move is a common and simple tamper in the image manipulation with the operation mode. It is to copy and paste the image of an area to another area of the image. In the actual manipulation process, in addition to simple to copy-move the image operation, also tend to have other image processing such as rotation, scaling, and at the same time there will be added various kinds of noise. For not just copy and paste to tamper with image area, this paper puts forward two kinds of detection methods.For detecting the image tempered by copy-move tampering with rotating, this paper proposes a method of detection algorithm using Gray Level Co-occurrence Matrix and radix sort. The improved algorithm is mainly based on the study proposed which is the detection algorithm using Gray Level Co-occurrence matrix. Firstly, the detected image was divided into multiple overlapping blocks with same size, and then calculate gray level co-occurrence matrix of each image block, get the feature vector of the image. Secondly, we used the radix sort method instead of the commonly used dictionary sort to sort the feature vector. Finally, we could locate the tampered region combining with the displacement vectors. Experimental results show that the algorithm can detect copy-move tampering in terms of robustness against rotate operation and efficiency, and shorten the time of detecting.For copy-move tampering accompanied by rotating and adding noise, According to the advantage of Gaussian-geometric moment, this paper proposed the detection algorithm is based on Gaussian-geometric moment, which is better than geometric moment in resistance to noise and the rotation invariance same as it. The steps of this algorithm are similar with the former. After overlap block for the image, we will extract features of Gaussian-geometric moment for each image block. Then the feature vectors are sorted by dictionary sort. Lastly, on the basis of the displacement vectors, we will find out the parts of image with tampered. By analyzing and comparing experimental results, it is concluded that the algorithm can achieve detecting of the mages tampered by copy-move with rotating and adding noise successfully.The proposed method are much better than geometric moment in noise resistance.