Enhancment of Image Tampering Detection Using JPEG's Quantization and Re-interpolation Processes

Recently, image forgery has become one of the critical issue in today information security. With easy-to-find and easy-to-use photo editing software, the integrity of an image can be no longer trusted. In this paper, we proposed a novel technique using 2-dimensional cross product in combining two existing image forgery detection techniques: JPEG quantization noise and interpolation based techniques. By combining two existing methods, efficiency of detection method and accuracy of locating of the tampered areas within the suspect image are increased. As results, it is shown that, in our experiments on real tampered images, our proposed algorithm give the better results in detection with lesser amount of noise and higher contrast between tampered region and the rest of the image. Considering the use of automatic image forgery detection system, with our proposed mechanism, the accuracy of the automatic forgery detection system will be increased significantly.