Hybrid Image Encryption Using Compressive Sensing and Substitution Permutation Networks

In an era of rapid digital transformation, ensuring the confidentiality and integrity of image data is paramount, particularly in applications such as secure communication, cloud storage, and medical imaging. Traditional encryption schemes often introduce computational overhead and storage inefficiencies, making them unsuitable for real-time and bandwidth-constrained environments. This paper presents an efficient and secure image encryption framework that combines Compressive Sensing (CS) and Substitution-Permutation Networks (SPNs) to address the challenges of real-time, storage-constrained applications. The process begins with image preprocessing and compression using Discrete Cosine Transform (DCT)-based CS, where high-magnitude frequency coefficients are preserved while reducing data redundancy. The compressed image is then encrypted using an SPN structure, which applies a Substitution layer (S-box) and a Permutation layer (P-box) for confusion and diffusion, ensuring strong resistance to cryptographic attacks. Decryption follows the reverse order with inverse SPN and IDCT to reconstruct the image with high fidelity. The proposed system achieves a PSNR of 57.43 dB, SSIM of 0.9986, and a low MSE of 0.1175, indicating near-lossless reconstruction. It demonstrates strong security performance with NPCR of 99.92% and UACI of 44.49%, alongside fast encryption/decryption times of 0.0013 sec and 0.0082 sec, respectively. With a compression ratio of 0.85 and memory usage of 3.11 MB, the framework effectively balances speed, quality, and security making it suitable for secure image transfer in IoT, cloud storage, and medical imaging.