DEEP TEXTURE FEATURES EXTRACTION USING CNN FOR IDENTIFYING FAKE IMAGES ON OSN

Abstract

These days, the availability of image processing software, such as Adobe Photoshop or GIMP have made image manipulation so common. Detecting such fake images is unavoidable for unveiling of the image-based cybercrimes. An image taken by digital camera or smartphone is usually saved in the JPEG format due to its popularity. JPEG algorithm works on image grids, compressed independently, with a size of 8x8 pixels. While unmodified images, have a similar error level. For resaving operation, each block should degrade at around same rate due to similar number of errors across the whole image. The compression ratio of this fake image is different from that of the original image and is detected using Error Level Analysis. The objective of our paper is to develop a photo forensics algorithm which can detect any photo manipulation. The error level analysis was then enhanced using vertical and horizontal histograms of error level analysis image to pinpoint the location of modification. Results show that the proposed algorithm could identify the modified image while showing the exact location of modifications.