Title: Data security by means of cryptography and image processing – a deep learning based method

Authors: Pooja Sharma; Anurag Patel

Addresses: Department of Computer Science and Engineering, IINTM, Janakpuri, India ' Department of Computer Science and Engineering, IINTM, Janakpuri, India

Abstract: Traditional picture encryption methods make use of rounds of diffusion and confusion in order to create an explicit trade-off between the degree of security and the length of time it takes to decode an image. Deep learning is a method that has the potential to be used in the production of responses that are acceptable for present problems in photo encryption systems. We explore the advantages and disadvantages of each of these research approaches and develop conclusions that are relevant to our findings. In the second step of the process, the various methods are compared and analysed in terms of the cryptographic properties of the recovered photos and the quality of the cipher images that are produced. Finally, conclusions are formed from comparing and analysing the deep learning technique in end-to-end encryption and decryption systems. These findings serve as a platform for future research.

Keywords: deep learning; image encryption; cryptographic attacks; encryption keys; style transfer.

DOI: 10.1504/IJESDF.2025.149335

International Journal of Electronic Security and Digital Forensics, 2025 Vol.17 No.6, pp.748 - 761

Received: 31 Dec 2023
Accepted: 04 Mar 2024
Published online: 27 Oct 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article