Image Encryption using Convolutional Neural Network

Abstract

The use of cryptography has become increasingly important in the transmission of multimedia, such as digital images, text, audio, and video, to ensure secrecy, integrity, confidentiality, and prevent unauthorized access to sensitive information. While Chaos-based cryptosystems are not yet standardized like AES, DES, RSA, they have emerged as an active area of research in recent years and can provide additional security when used with standard public key cryptosystems. This project aims to implement an effective image encryption approach using a Chaos-based cryptosystem to overcome differential attacks. The system involves dividing the original image into parts and repositioning them to form the first level of encryption. The encryption process starts with generating a one-dimensional sequence using a logistic map, which is then multiplied by the maximum pixel value and subjected to bit-by-bit operation. The result is used to encrypt the image, which can be decrypted using the same process in reverse.