Stego Secure: A CNN-Assisted Web Platform for Secure Image Steganography with AES-256 Encryption, LSB Embedding, and ML-Based Steganalysis Detection

Abstract

This paper proposes Stego Secure, a full-stack web-based steganography system incorporating Least Significant Bit (LSB) image steganography, AES-256 encryption, and a machine learning-based steganalysis detection mechanism in a single production-grade steganography system. Stego Secure has a three-tier architecture consisting of a Next.js/React/TypeScript frontend, a Python Flask-based RESTful API backend, and a Chrome browser extension. AES-256-CBC encryption is implemented as a preprocessing mechanism before the image steganography embedding mechanism. A custom CNN-based model for steganalysis has been implemented, which has a detection accuracy of 96.4% at maximum embedding density. The PSNR results are always above 51 dB. Stego Secure supports JPEG, PNG, and BMP image file formats up to 5000 characters and is protected using OAuth 2.0 protocol through Google and GitHub OAuth 2.0-based NextAuth.js. This paper bridges the gap between theoretical research in steganography and implementation.