Image Tampering Forgery Detection Using Convolutional Neural Network with Blockchain

Abstract

AbstractFake news contains misinformation communicated through various channels including social media. Fake news is spreading in large numbers in this internet era as social media provide an easy way to share it without any authentication. Fake news may have multiple forms like text and images, from which we can extract features to identify the patterns and operations applied on the images. Objective of the system is to build the model for “Detection of fake news presented as image” problem. Deep learning algorithms can be used to develop approaches which can detect tampered images retrieved from any source. However, it is challenging to determine the tampered images as it requires training of models to differentiate between the actual images and tampered images in order to classify it as fake. Automatic separating of fake news requires thorough investigation of images data to get explicit features from the images used in the fake news. With the use of CNN, the images features including the hidden patterns are extracted using the multiple convolutional layers. People upload and spread images over the internet and there is no control on the authenticity and ownership on their own photos. Images stored can be audited to trace the transactions applied on the images using the smart contracts deployed on blockchain. Image forgery detection using convolutional neural network with blockchain (CNNB) is proposed in this work where we combine blockchain technology with image processing techniques using deep learning to build a new hybrid model for mining the tampered images used in the news.KeywordsBlockchainConvolutional neural networkFake newsImage forensicSmart contractsImage tampering