Analysis of zig-zag scan based modified feedback convolution algorithm against differential attacks and its application to image encryption

• Published in: Applied intelligence (Dordrecht, Netherlands) Vol. 50; no. 10; pp. 3101 - 3124
• Main Authors: Vidhya, R., Brindha, M., Gounden, N. Ammasai
• Format: Journal Article
• Language: English
• Published: New York Springer US 01.10.2020; Springer Nature B.V
• Subjects: Algorithms; Artificial Intelligence; Computer Science; Computer simulation; Convolution; Encryption; Feedback; Machines; Manufacturing; Mechanical Engineering; Pixels; Processes; Security; Differential cryptanalysis; Image encryption; Convolution; Security
• ISSN: 0924-669X; 1573-7497
• DOI: 10.1007/s10489-020-01697-1

In this paper, a novel zig-zag scan-based feedback convolution algorithm for image encryption against differential attacks is proposed. The two measures Number of Pixel Change Rate (NPCR) and Unified Average Changed Intensity (UACI) are commonly utilized for analyzing the differential attacks. From the study of the existing papers, even though high Number of Pixel Change Rate and Unified Average Changed Intensity values are obtained, a few values lie in the critical range of α -level significance which in turn increase the possibility of differential attacks. To overcome differential attacks, two aspects of scanning with different test cases are analyzed and from these analyses, it is concluded that zig-zag scan based feedback convolution in forward and reverse direction achieves good Number of Pixel Change Rate and Unified Average Changed Intensity without critical values. Zig-zag scan based feedback convolution in forward and reverse direction is enforced for key sequence generation and applied in diffusion process to achieve high level of security. Moreover, plain image related initial seed is also generated to overcome the chosen/known plain text attacks. Both numerical and theoretical analyses are performed to prove that the proposed encryption method is resistant to differential attacks. General security measures are carried out for the proposed method to validate its security level. From the simulations, it is shown that the proposed methodology has good keyspace, high key sensitivity, good randomness, and uniform distribution of cipher image pixels.