Detection of Geometric Transformations in Copy-Move Forgery of Digital Images

Abstract

Digital Forensics is a branch of forensic science which is related to cyber crime. It basically involves the detection, recovery and investigation of material found in digital devices. Digital images and videos plays most important role in digital forensics. They are the prime evidences of any crime scene. So the fidelity of the image is important. Digital images can be easily manipulated and edited with the help of image processing tools. Copy-move Forgery is the most primitive form of cyber attack on digital images. In Copy-move forgery a part of image (region) itself is copied and pasted into another part of the same image. The intension behind this type of attack is to “add” or “disappear” some objects from the image. Hence to break the fidelity of the image and fool the viewer. Copy-move attack is more prevalent in images having uniform texture or patterns, for e.g. sand, grass, water etc. In this thesis exact block matching is used as a detection technique. This technique is based on block matching, for these the whole image is divided into number of block and then the matching process is applied. Sometimes the copied region is processed before pasted i.e. some geometric transformations is applied on the pasted region. The transformations like scaling, rotation etc. It is not possible for human eyes to detect such kind of forgeries. Whenever forgery is done in this manner the common techniques like block matching, exhaustive search, auto-correlation and robust match etc. are not able to detect the forgery having geometric transformations. So that for identification of forged region we need some technique which are based on local features and also invariant to transformations. In this thesis SIFT is used for forgery detection. SIFT stands for Scale Invariant Feature Transform, this gives local feature points which are invariant to scales. The key points helps to find the duplicated region with different matching algorithms