File(s) not publicly available
Selective encryption of JPEG images with chaotic based novel S-box
journal contributionposted on 2022-09-29, 09:22 authored by N A Khan, M Altaf, F A Khan
Increased demand of multimedia data over heterogeneous networks has led to the requirement for increased compression and suitable security. Both of these demands require heavy computation, and are contradictory to each other, as encryption may severely affect the compression efficiency. On the other hand, handheld devices have limited energy and computational resources, making compression friendly encryption a challenging issue. One of the solutions is to encrypt a subset of data based on perceptual importance. But, that too may not be compression efficient, format compliant and secure. To address the security issue in such applications the substitution box (S-box), which is one of the most important and the only non-linear operation of the block ciphers is redesigned using chaotic equations. These equations have the characteristics that easily meet the requirements of the block ciphers. In this research work, the S-box based on the chaotic equation was designed and tested for security strength and then used for selective encryption of the multimedia (image) data. To increase the security and generate a key dependant S-box, Mackey Glass equation is used due to its inherent properties like sensitivity, pseudo random characteristics, non-uniform behaviour, ergodicity, and high confusion and diffusion. The security of the proposed S-box was tested in terms of non-linearity, Bit Independence and Strict Avalanche Criteria etc. This S-box is then used to selectively encrypt a subset of image data during the process of compression, giving an encryption-compression strategy. The subset of image data was selected such that the statistical and structural dependencies were not violated due to encryption. This resulted in compression efficient and format friendly encryption. In this regard the quantization table is selected for encryption along with a subset of bits that have already encoded and their statistical dependencies are exploited for compression. This approach not only reduced the computational burden due to the selective nature of encryption but also kept the security at the required level due to introduction of key dependent S-Box. The results of encrypted images were compared with that of the AES in terms of compression ratio, correlation and Peak Signal to Noise Ratio (PSNR), giving better results for the proposed algorithm.