The Image Forgery Detection Based On Discrete Wavelet Transform

Produce powerful image processing software, digital rapid change, making it easier for the average user a simple digital image without a trace visible, if the user mess with fictional images, it will cause bad influence on society, such as social sedition interference to undermine the law of evidence extracted. This type of problem and other issues related to data protection, will take place one after another. Therefore, in the field of digital image forensics, digital image tampering forensics technology research it has become essential.Now, in the digital image detection area of research, detection techniques can be roughly divided into two types:one is similar to a digital signature or digital watermark detection technology, which we call proactive detection technology; another contrast, we call passive detection technology, which does not need to any additional information embedded in the image. The most representative is known to everyone passive blind detection technology. Such methods include copy-paste inspection, open print double compression test, resampling test and fuzzy estimation inspection. In this paper, a special type of digital image tampering blind test, the image region copy-paste tamper detection for major research. In the image tampering technology, frequently used means is to copy and paste an image region tampering, the technology can be copied to an area in the original image, and then paste the copied part of the image does not intersect with one another area, thus can remove important information in the original image.In this paper, the existing algorithm is improved, they also got a new and effective image copy and paste tamper detection algorithm. The algorithm works as follows:The original image is first treated detected discrete wavelet transform, image size reduction, followed by "cross Shadows" chunking method will reduce the image after the image is divided into nine pieces, and then calculate twenty-two image sub-block phase correlation image and determining the phase correlation pulse, pulse according to the calculated correlation coefficients copy the image-the transition between the value of the paste area. Then, according to the desired amount of displacement corresponding to the image pixel, by copying and pasting the image area match is found. Finally, the use of morphological filtering techniques to increase the area of the positioning image tamper, the algorithm is easy to understand, easy to implement. Can be obtained by experiment found that the algorithm can not only achieve a small range of image positioning, multiple images can be tampered with test area to ensure the accuracy of the algorithm, but also extend the application of the algorithm.