We propose a lightweight, one-pass, format-preserving image encryption scheme based on vectorized secret sharing (VSS), designed for high-throughput operation with minimal computational overhead. The scheme applies cryptographic primitives only to a limited subset of masked blocks, while remaining data is processed through efficient algebraic transformations over \(\mathbb{G}\mathbb{F}(2^8)\) , enabling substantial runtime gains without altering the original image format. It achieves provable IND-CCA security with a dual-IV mechanism that ensures ciphertext authentication and robustness against tampering at no additional cost. An optimized implementation using precomputed lookup tables and batch processing attains throughput exceeding 230 MB/s on desktop systems and competitive performance on IoT architectures (ARMv7, AARCH64, RISC-V64). Comparative benchmarking shows that the proposed scheme matches or outperforms several established authenticated encryption standards. Security evaluations on standard RGB images confirm strong diffusion, high key sensitivity, and excellent NPCR, UACI, and entropy metrics. Overall, the scheme offers an efficient and secure solution for real-time image encryption in resource-constrained environments.
