Reversible adaptive chaotic multi-image privacy protection scheme for efficient batch encryption of various images

Addressing the limitations of traditional encryption algorithms with fixed structures, this paper proposes a multi-image encryption scheme based on chaotic map. This scheme allows users to input multiple images of varying sizes and types, which are then encrypted into a single noise-like image stored on a cloud platform. First, a three-dimensional hyperchaotic memristor map with excellent randomness is constructed. Through the proposed key-initial value generation mechanism based on the SHA-384 hash algorithm, the user’s provided key is converted into an initial value to generate the chaotic sequence used for encryption. Subsequently, multiple input images are pre-processed and concatenated into a large plaintext image, which undergoes bit-level adaptive domain diffusion and chaotic index-Zigzag composite scrambling to produce the final ciphertext image. The algorithm introduces adaptive encryption parameter that control the size of the traversal matrix and the rule matrix based on the input images. By transforming large matrix operations into multiple smaller matrix operations, the encryption process is dynamically adjusted and computational resources are optimized, thereby enhancing both the flexibility and efficiency of the algorithm. Finally, security analyses with multiple groups of test images show excellent performance across all metrics. Notably, the scheme can encrypt 50 images in 25.8 seconds, demonstrating remarkable efficiency.