Image Encryption Using Neural Network Based Chaotic Systems

This paper introduces a unique image encryption methodology using the chaotic signals generated by a neural network for enhanced security. The proposed system employs a master-slave configuration where a Multi-Layer Perceptron (MLP) neural network acts as the master system for encryption, while the Hopfield neural network serves as the slave system generating the chaotic signal for decryption. The process of encryption involves the XOR operation between the input image and the chaotic signal generated by the master system. To ensure accurate synchronisation between the master-slave systems for decrypting the image, an Active Sliding Mode Controller (ASMC) is used which ensures robust convergence and minimal synchronisation error. Performance evaluation through histogram analysis shows that the proposed encryption method is more secure and efficient compared to the Advanced Encryption Standard (AES) method.