Image Tampering Detection Using Error Level Analysis and Metadata Analysis

As the volume of digital content continues to increase, the problem of information pollution has become increasingly pressing. Information pollution refers to the dissemination of false, misleading, or biased information through various channels such as social media, news outlets, and websites. This can include fake news, propaganda, as well as tampered or edited images and videos. With the proliferation of digital image editing software and artificial intelligence (AI) techniques, it is increasingly challenging to determine the originality of a photograph. Field of digital image forensics, which aims to identify and prevent image forgery, has gained significant attention in last few years. In this study, we propose a combined approach of Error Level Analysis (ELA) and Metadata Analysis for the passive authentication of images. Utilizing deep neural networks, our proposed approach seeks to evaluate the originality of an image. System demonstrated a test accuracy of 87.24% for the ELA approach and a test accuracy of 73% for the metadata analysis approach.