Image Splice Detection Through Noise Pattern Analysis

Digital Images have become one of the most basic communication methods which makes it an essential part of human day today life. Most of the people hold mobile devices with them that are equipped with image capturing facilities. With the rise of the digital image forgery, the reliability and the confidentiality of the images were questioned. Digital image forensics was considered as an important field of research due to this reliability issue. Many of the images that are circulating through the web are being either tampered or forged. Image Splicing is one of the popular tampering techniques in used. This research has been carried out to detect the above identified technique with focusing on noise pattern variations that are visible in tampered images. The images are being segmented into non-overlapping blocks and then noise variances are estimated after a series of processes including wavelet transformations on luminance component of the image. Experimental results based on White Gaussian noise have proved the proposed detection method.