Enhanced Cast-128 with Adaptive S-Box Optimization via Neural Networks for Image Protection
摘要
The CAST-128 happens to be an exceptional encryption algorithm improved through the inclusion of key-dependent dynamic substitution boxes (S-boxes), which are created by a hybrid chaos system composed of a logistic-Sine Map (LSM). The technique overcomes flaws of fixed S-boxes used in conventional cryptographic approaches that are susceptible to linear and differential attacks. The mechanism has five primary steps, which are pre-processing the input image, generating adaptive S-boxes based on the LSM, incorporating these S-boxes into CAST-128, and locating the encrypted image blocks, and the analysis of the security. The produced S-boxes are highly non-linear, have an avalanche effect, and are bijective, making them appropriate to come up with robust encryption. Results given by benchmark grayscale images (e.g., Lena, Baboon) confirm these postulations, manifesting superior values in entropy, NPCR, UACI, and histogram uniformity and effective running time. The lightweight, chaos-based model increases robustness to statistical and brute-force attacks, presenting a secure and practical solution to real-time protection of images in various fields like medical imaging, surveillance, and digital communication.