The Design of an Adaptive Enhanced AMP-Based Image Block Compressed Sensing and Its Application to Image Encryption

Compressed sensing (CS) has become a widely employed technique in the field of image encryption. Despite achieving a high level of data encryption security, the resulting decrypted images often lack satisfactory quality. In this paper, we develop an adaptive enhanced approximate message passing (AMP) block CS (AE-AMP-BCS) algorithm for image BCS and its application to image encryption. Initially, we present an adaptive energy-based anti-aliasing filtering strategy (AAFS) for preprocessing the input image, mitigating the noise effect during reconstruction. The filtered image is then transformed into the Haar domain for enhanced security, followed by a twofold perturbation operation and a piece-wise linear chaotic system-based measurement generation approach for ratio allocation and security consideration. Finally, the cipher measurements are decrypted using an existing AMP-based algorithm. It is noteworthy that the keys for perturbation operations and the chaotic system are derived using the SHA512 hash. Comprehensive experimental results demonstrate that the proposed AE-AMP-BCS significantly outperforms state-of-the-art BCS methods in terms of image reconstruction quality. Its application in image encryption showcases competitive encryption capabilities, strong robustness, and outstanding image decryption quality compared to other CS-based image encryption algorithms.