This study presents a robust image encryption method that integrates structured permutation, chaotic scrambling, and key-dependent diffusion, ensuring high security and computational efficiency. The input RGB image has been partitioned into \(64 \times 64\) blocks, which are then subdivided into \(4 \times 4\) micro-blocks. A novel H-pattern zigzag traversal reorders pixel positions to improve confusion, followed by ten iterations of the Arnold Cat Map for spatial scrambling, which leverages its ergodic and sensitive properties. A pseudo-random key-stream generated from a secure seed is applied via the XOR operation to achieve diffusion at the bit level, providing strong resistance to statistical and differential attacks. Experimental analysis of the Baboon image demonstrates highly accurate reconstruction, with an MSE of 0.425, a PSNR of 51.87 dB, an MAE of 0.0000, and an SSIM of 0.9969, confirming near-lossless decryption. The proposed method has achieved entropy values of 7.999394, 7.999260, and 7.999328 for the R, G, and B channels, respectively. The UACI score is 33.3359%, the NPCR is 99.9312%, and the achieved pixel-level avalanche effect score is 99.6063%. This validates its strength against differential attacks. This proposed method achieved encryption and decryption times of 0.015747 s and 0.000352 s. The unique combination of H-zigzag traversal, Arnold Cat Map, and key-dependent diffusion establishes an unprecedented level of security.
