A sequential convolutional neural network for image forgery detection

In this digital era, images are the major information carriers of contemporary society. Several multimedia manipulation tools like CorelDRAW, GIMP, Freehand, Adobe Photoshop, etc. are being used to forge the visual media for malicious reasons. It is becoming increasingly difficult to distinguish forged images from pristine images as a result of new manipulation techniques that have emerged over the past time. The most intriguing area of multimedia forensics research is image forgery detection. In the field of forensic image analysis, the most important task is to verify the authenticity of digital media. A novel passive approach for detecting digital image forgery is proffered in this manuscript. It is a sequential framework that uses a deep convolutional neural network to differentiate between original and altered images. On the COVERAGE dataset, numerous experiments have been evaluated in order to construct an effective and robust model, achieveing AUC value of 0.85 and F-measure of 0.70. The comparative results have been represented in summarized form and the results perform better than the state-of-the-art techniques.