Contrast Enhancement Forensics based on Convolutional Neural Networks(CNN)

This work is about a CNN-based contrast enhancement image forensic method, which can verify autheticity of digital images, i.e., it can distinguish the contrast enhancement forgeries from unaltered images. <br>The dataset consists of grey level co-occurrence matrix(GLCM) of three types of images: unaltered images, contrast-enhanced images processed by common contrast enhancement techniques (histogram stretching, gamma correction, and S-curve mapping), and the images manipulated by using three current counter-forensic attacks. <br><br>README.pdf has descriptions of the other .tar.gz format files. <br><br>- Train dataset.tar.gz: train data<br>- Test dataset.tar.gz: test data<br>- CEForensicsCNN.tar.gz: source codes &amp; trained model<br>