Copy-Move Forgery Detection using SIFT and GLCM-based Texture Analysis

Easier access to editing tools and growing risk of image manipulation has encouraged extensive research in copy-move forgery detection. Although the current methods have been able to detect this tampering to a good extent, their accuracies drop when tested on images with different sizes of tampered regions and in the presence of similar but genuine objects in the image. In this paper, these issues are addressed by including a novel GLCM-based Texture Analysis Filter that gives information about the textural similarity of the keypoint neighbourhoods by using difference of GLCM contrasts as the similarity metric. Experimental results show that the proposed technique can address a variety of different tampering scenarios and outperforms the existing state-of-the-art Copy-Move Forgery Detection(CMFD) techniques by handling multiple forgeries, returning corresponding geometrical parameters and significantly improving the false positive rates.

I. Introduction

With better access to various image editing tools, the credibility of digital images is at stake. Images are used as a proof of reality - both in formal and informal settings and hence, its authenticity is of prime concern. In the past few years, the detection of tampered images has gathered much attention from researchers worldwide. A particular focus is given to copy-move forgery detection as it is one of the most commonly used image tampering techniques. Although there are methods dealing with this issue, they are not robust enough to handle realistic tampering or tampering where copied areas are subjected to different geometrical transformations. This calls for a need to have a technique that is not only immune to these processing steps but also differentiates well between actually copy-moved objects and Similar but Genuine Objects (SGO).