Enhanced model for fake image detection (EMFID) using convolutional neural networks with histogram and wavelet based feature extractions

Abstract

In recent scenario of digital world and image forensics image is forged or unforged to be checked. It is a critical process in the areas where the images are taken as crucial proof in decision making in several cases. Hence, for handling the delinquent activities on image, a novel model, which is capable of detecting or classifying the authenticated and altered images, is required. A new model called Enhanced Model for Fake Image Detection (EMFID) is used for classifying the obtained images as AUTHENTIC and FORGED. Moreover, this work also discusses about the images that are altered with the splicing methods that is performed by cropping and pasting certain image part from one to another original image. The proposed model comprises four phases such as Image Pre-processing, Histogram based Feature Extraction, Discrete Wavelet Transform based Feature Extraction, Image Classification with ConvolutionalNeural Networks (CNN). The test images that are obtained from data set are classified under Authentic Class and Forged class. The results are better with classification accuracy and model efficiency.