الفهرس 
CHAPTER 1: INTRODUCTIONOnly 14 pages are availabe for public view
 |  | from | 97 |  |  |
| | | from | 97 | | |
Abstract
Image forgery can be classified into three types: Image enhancement; such as blurring, contrast or brightness alteration, etc. Image compositing; that mixes between two or more different images and Copy-move forgery; where some parts of an image are copied and pasted to another place of the same image. The techniques of image forgery detection can be categorized into two methods: passive and active which depend on prior information about the original image that in many cases is not available.
In this thesis, We focus on passive or blind methods for digital image forgery detection. Most researchers agree that passive detection system consists of three phases: Image Preparing, Feature Extraction and Matching. In the proposed approaches, we have followed the previously mentioned phases.
Image preparing is the first phase in our approaches. Our proposed methods are based on block matching searching strategy. As the main features considered in our system, all test images are converted to gray scale in this phase.
The second phase in our proposed approaches is the feature extraction phase. Features are extracted from each block by applying DCT in PBM and FS approaches. While in CRMS, the average intensity of nested frames for each block are used as the discriminative feature vector. In CMP, Fourier transform decomposition for each column in polar blocks is performed to extract the representative features.
Matching is the final phase in blind image forensics system. Absolute difference and correlation measurements are utilized between feature vectors. If the difference is less than some threshold or the correlation is more than some threshold, then the two blocks are considered to be similar. Spatial distance between these blocks is calculated to reduce false detection too.