Sustainable Image-based Encryption Using Cryptography and Steganography with Autoencoder

Data security is a major concern in today’s rapidly evolving landscape of communication and information exchange. With the growing importance of digital transactions and communication channels in our daily lives, where traditional cryptographic methods face escalating challenges from sophisticated cyber threats, this research introduces a novel encryption paradigm that integrates the autoencoder neural network model as powerful tools for data representation and image-based encryption to encrypt the plaintext, making a cryptographic steganography. The innovative integration of autoencoder neural networks aims to enhance data transmission security by harnessing the unique strengths of the proposed cryptosystem. A comprehensive analysis that covers key space exploration, key sensitivity testing, entropy tests, histogram analysis, and correlation analysis is conducted to evaluate the efficacy of the approach. As cyber threats continue to evolve, the need for advanced encryption techniques becomes paramount, and this work aspires to address this imperative by offering a novel perspective on cryptographic steganography empowered by autoencoder-generated keys.