An efficient image encryption algorithm using a discrete memory-based logistic map with deep neural network

Abstract

Abstract In the last few years, multimedia technology has made tremendous strides. These days, the Web is frequently used to transfer multimedia content, including audio, video, and photos. However, the Internet is a very vulnerable medium with many security holes. To ensure that multimedia content carried across unprotected channels, like the Internet, is secure and private, several encryption techniques have been proposed. New encryption strategies must be developed because multimedia data streams cannot be encrypted using traditional methods. Therefore, the main goal of the recommended system is to present an analytical research approach for introducing a sophisticated framework wherein the suggested encryption technologies' efficacy is increased through the use of deep neural networks (DNNs). The robustness of the DNN principle is coupled with a discrete memristor-based logistic chaotic map notion for enhanced security performance. In this paper, three distinct encryption algorithms—Arnie cat with an artificial neural network (ANN), Henon map with an ANN, and logistic map with a DNN—are compared for security and performance with the suggested algorithm. Correlation coefficients, information entropy, number of pixels changing rate (NPCR), encryption quality, and encryption duration are the cryptographic analysis parameters examined here. The results show that the recommended implementation enhances security performance without degrading image quality. The proposed algorithm achieves 35.9% of UACI, 99.95% of NPCR, and 7.997231 of entropy.

Keywords

• Image encryption
• Image decryption
• Arnold map
• He map
• Logistic map
• Artificial neural network