The Research For Digital Image Splicing Detection

Nowadays, digital cameras are playing an important role in our daily life as well as in forensic affairs. However, the sophisticated digital image processing software such as Photoshop and Freehand have made image tampering much easier to operate but harder to detect. Today, splicing operation is widely used under various conditions. Therefore, digital images are no longer as authentic as before. Especially in resent years, numerous photos in photography competition are exposed to be synthesized. This urges us to find the ways to distinguish the authentic and tampered photos and the ways to detect image tampering.In this thesis, we focus on digital image forensic techniques for digital photographs rendered by digital cameras. As such, we start from taking a close look at the image signal processing pipeline of a digital camera, with an emphasis on the impact each processing step has on the statistic features of images. Then two related algorithms are introduced in details. Some defects of the algorithms were found when we program and test the algorithms. Last, for the defects of classical algorithms, we proposed two new detecting algorithms.The main contribution of this thesis is to develop two new algorithms for digital photograph rendered by digital cameras. The first approach works using the property that the verge of the spliced area is more sensitive than normal boundaries to the re-interpolation due to the discontinuity of CFA interpolations. So, the sensitivity matrix of the original image to re-interpolation is calculated first. Then, a localization matrix is generated by using our adaptive iterative algorithm. It shows the suspect pixels at the verge of the spliced area. At last, features of the localization matrix are extracted and fed into SVM to classify the given images. Experimental results demonstrate the effectiveness of our approach in detecting image splicing. Our algorithm makes up for the classical algorithm in the aspect that classical detecting algorithm has a low detecting ratio facing to the complex CFA interpolation algorithm in commercial cameras.In the second detecting method, we note that tampering process and the post-processing operations will inevitably disturb the statistic features of the original image. We propose a method making use of feature fusion. First, we extract the feature statistics that can represent the property of a camera from the images taken by that camera. These feature statistics are used for training a one-class classifier in order to get the feature pattern of the given camera. Then, we do sliding segmentation to testing images. Finally, feature statistics extracted from image blocks are fed into the trained one-class classifier to match the feature pattern of the given camera. The method can detect image splicing even the spliced images are re-interpolated. This method makes up for the classical algorithm in the aspect that classical detecting algorithm has a problem facing to the re-interpolation tampering operation.At the end of the thesis, the prospect of digital image forensics is discussed subsequently as well as the possible future research areas.