Real-Time Reversible Image Steganography with Lightweight Vision Transformers for Embedded Systems

Authors

Abstract

This paper presents a real-time, resource-efficient framework for reversible image steganography that utilizes lightweight Vision Transformers (ViTs), specifically designed for edge computing devices. Building upon the foundational StegoTransformer model, the proposed architecture incorporates MobileViT and TinyViT for embedding and extracting hidden image data. The system is optimized to function effectively under constrained computational resources, enabling secure and reversible data hiding on platforms such as Jetson Nano, Raspberry Pi, and mobile devices. Experimental results indicate competitive performance in terms of payload capacity, visual fidelity, and message recovery accuracy, while achieving low latency and memory consumption suitable for real-world deployment.

Additional Files

Published

2026-02-17

Issue

Section

Cryptography and Cybersecurity