Digital image forensics using statistical features and neural network classifier

Digital image forensics is a new topic in recent years, which deals with the authenticity and credibility of digital images. How to recognize fake images is still a problem. This paper presents a fake image classification scheme using higher order image statistics and RBF neural networks. The features constructed on the higher order statistics reveal the intrinsic statistical features between fake images and real images. Then a classifier based on RBF neural networks is used to classify the fake and real images using these features. Experimental results demonstrated the effectiveness of the proposed scheme.