Image Forgery Localization via CFA Based Feature Extraction and Poisson Matting

In this era of digital computing, peoples are used to represent the information that they want to convey in visual forms. Representing in visual forms rather than in pure text form make them more understandable. Digital camera images are used in several important applications in the recent decades. But with widespread availability of image processing and editing software make the integrity of digital camera images at risk. So verifying the authenticity and integrity of digital images, and detecting the traces of forgery are important. In order to develop techniques for detecting the forgeries and to improve them, a separate science has been formed, called the Image Forensic Science. A lot of methods have been developed to solve the issue This paper describes a novel passive fine grained approach to this problem. we use an in-camera processing method to detect the forgery rather than focusing on the statistical differences between the images textures. We recognize that digital camera images contain a CFA interpolation relationship between the pixels as a result of using a color filter array with demosaicing algorithms. The proposed method detects the forgery by estimating a feature value that indicates presence of demosaicing artifacts (interpolation relationship). After detecting the forgery, poisson matting algorithm is used enhance result by cutting the forged region from the image. This method can localize forged region efficiently.