The Study, Based On The Statistical Properties Of The Digital Image Tamper Detection Algorithm

During the past decades, powerful computers, high-resolution digital cameras and sophisticated photo-editing software packages have become affordable and available to a large number of people. It has become fairly straightforward to create digital forgeries that are hard to distinguish from authentic photographs. These forgeries, if used in the mass media or courts of law, will have a great impact on our society. So the detecting of digital tampering has become a crucial problem.Two algorithms of detecting the digital forgeries are proposed in this paper. They are based on the study of some statistical properties of digital images. The first one is called detection of forgery using inherent attributes of the digital camera. It detects the forgeries on the basis of two inherent attributes of the digital camera, the sensor pattern noise's invariability and the special correlation caused by color filter array interpolation algorithms. We detect the tampered regions with the sensor pattern noise, and then verify the result with the other inherent attribute.In the making of digital image frauds, there would be many manipulations. But applying blurring is inevitable to get vivid forgeries. As a result, the tampered regions are included in the blurred regions of the forgeries. Therefore, in this paper, we propose another algorithm to figure out the forgeries by detecting the blurred regions of images.Finally we implement the proposed algorithms and experiment on them. The result shows that the first algorithm can detect the tampered regions accurately and the other one can figure out the original blurred and tampered regions of images. Both of them are demonstrated on several forged images and have been proved to have high efficiency.