ImageCloak: An CNN Based Stegenography System

Abstract

ImageCloak is an advanced deep learning-based image steganography system that enables secure embedding and extraction of hidden data within digital images using Convolutional Neural Networks (CNNs). The system achieves high performance, with 91.48% testing accuracy in text encoding and image encoding results of SSIM: 0.8898 and PSNR: 31.12 dB, ensuring that the concealed data remains imperceptible while preserving the visual quality of the cover image. To enhance scalability and usability, ImageCloak integrates MongoDB for efficient storage and retrieval of stego-images, allowing users to manage hidden data securely over time. The system also employs robust security measures to prevent unauthorized access, ensuring that sensitive information remains protected. Additionally, it supports seamless sharing of encoded images among authorized users, making it an effective solution for private communication, secure data exchange, and long-term confidential storage. By leveraging deep learning techniques and advanced embedding methods, ImageCloak provides a reliable and scalable approach to modern steganographic applications, enhancing data privacy and security in digital communication.