Image encryption/decryption accelerator based on Fast Cosine Number Transform

Abstract

The Cosine Number Transform (CNT) and its Fast Cosine Number Transform (FCNT) are widely used in image encryption due to their modular arithmetic, which enhances computational accuracy. However, existing hardware architectures for FCNT suffer from long computation cycles and high resource consumption, making it challenging to meet the demands for fast image encryption. This paper proposes an eight-point FCNT hardware architecture with multiplier-less multiplication (MM), employing pipeline and time-division multiplexing methods. Based on this architecture, an image encryption hardware accelerator was implemented. Experimental results show that compared to existing methods, the proposed FCNT architecture reduces computational delay by 18.2%and decreases LUTs usage by 18.5% and FFs usage by 21.7%. Furthermore, compared to the current state-of-the-art, the image encryption hardware accelerator based on our FCNT architecture achieves the fastest processing speed, requiring only 0.505 ms to encrypt a single 256 × 256 grayscale image, with a throughput of 1038 Mbps.