Enhanced Copy-Move Forgery Detection in Images Using Hybrid Deep Learning Approach

¹ Research Scholar, SOCSE, Sandip University, Nashik, India
² Assistant Professor, SOCSE, Sandip University, Nashik, India

^* Corresponding author: This email address is being protected from spambots. You need JavaScript enabled to view it.

Published online: 20 November 2025

Abstract

Copy-move forgery (CMF) is one of the most common and challenging forms of digital image tampering, where a part of an image is copied and pasted within the same image to conceal information. To address the limitations of traditional CMF detection techniques, this paper proposes a novel hybrid deep learning framework combining Vision Transformers (ViTs) and Long Short-Term Memory (LSTM) networks. The proposed approach begins with comprehensive preprocessing, including resizing, normalization, and patch division, followed by patch embedding. The embedded patches are processed through a Transformer encoder that captures long-range dependencies and spatial relationships via multi-head self-attention and multilayer perceptrons. The output feature sequences are further analyzed using an LSTM network to capture temporal dynamics and fine-grained contextual patterns. Finally, a classification layer distinguishes between authentic and forged regions. Publicly available datasets such as CASIA and MICC-F220 are utilized for training and evaluation. The hybrid ViT-LSTM architecture is expected to enhance detection accuracy, reduce false positives, and improve robustness against various post-processing attacks. Preliminary results from the preprocessing stage are promising, and the complete implementation aims to demonstrate superior performance over conventional methods.