Authors: Professor Shivani Karhale, Mr. Rohit Pawar, Ms. Sanskruti Marawade, Ms. Nandini Jadhav, Ms. Vaishnavi Pawaskar
Abstract: The rapid advancement of artificial intelligence, image editing tools, and generative models has made visual manipulation easier than ever. Altered images can influence legal investigations, journalism, social media, and political narratives, creating a critical need for automated authenticity verification systems. This research introduces an AI-Powered Forensic Image Suite integrating shadow analysis, image consistency detection, and metadata verification to identify tampered digital images. The system preprocesses images through resizing, normalization, and noise reduction, followed by shadow recognition using gradient-based and geometric estimation techniques. Image consistency is evaluated using structural similarity, lighting coherence, and texture uniformity checks. Metadata analysis extracts EXIF information to verify timestamps, camera signatures, and editing traces. Experiments conducted on a dataset of 500 real and manipulated images demonstrate high accuracy, with shadow detection (94%), consistency check (92%), and metadata validation (98%). The suite serves as a reliable tool for investigators, journalists, and forensic professionals, and provides a scalable foundation for advanced features such as deepfake detection, reverse image search, and error-level analysis.
