MI-DCC: A Content-Aware Dynamic Cipher Composition Framework for Adaptive Multi-Image Encryption

This paper presents MI-DCC (Multi-Image Dynamic Cipher Composition), a content-aware framework for secure multi-image encryption. Unlike conventional schemes applying uniform cryptographic processes, MI-DCC combines structured cross-image region permutation with adaptive dynamic cipher selection. The framework establishes deterministic inter-image dependencies through key-driven region exchange, then analyzes regional statistical complexity using entropy, variance, intensity, and gradient descriptors. An adaptive Dynamic Cipher Composition mechanism assigns appropriate primitives from a heterogeneous cryptographic library according to computed complexity scores, improving diversity while maintaining competitive performance. A global diffusion stage propagates local modifications across the entire image batch, strengthening resistance against differential and statistical attacks. Experimental results on benchmark grayscale and color images demonstrate near-ideal entropy (7.9993 bits/pixel), high NPCR (99.612%), UACI (33.465%), and negligible pixel correlation. These results indicate that MI-DCC provides effective adaptive behavior, strong statistical diffusion, and exact reversibility under the evaluated experimental configuration.