2-D Compressive Sensing-Based Visually Secure Multilevel Image Encryption Scheme

Xiaofei He,Lixiang Li,Haipeng Peng,Fenghua Tong
DOI: https://doi.org/10.1109/jsen.2023.3341428
IF: 4.3
2024-01-01
IEEE Sensors Journal
Abstract:Aiming at the problem of how to effectively protect the sensitive information of multiple images and ensure the security of their transmission, we propose a visually meaningful multilevel image encryption solution that combines 3-D chaotic mapping, discrete cosine transform (DCT) and 2-D compressed sensing. First, the association relationship between related grayscale images is established by using different reversible relationship matrices. We then undersample multiple grayscale images with established relationships via 2-D compressive sensing (CS). Second, chaotic scrambling and index scrambling are performed on the obtained multiple measurement results. Finally, the visually meaningful steganographic image is obtained by embedding different measurement results into the selected color image via the DCT. Compared with the existing methods, this approach solves the problem of multiprivacy protection by realizing multilevel encryption of multiple images. More importantly, it improves the encryption ability and transmission efficiency by establishing the correlation relationship between different images. Realize the secure transmission of ciphertext data by information hiding and chaotic encryption mechanisms. Provide multilevel image recovery quality to authorized users by providing different authorization keys. In short, the proposed scheme has the advantages of high encryption capability, high security and controllable access. It is especially suitable for compressive sampling, privacy protection and secure transmission of images in sensor networks with limited computing power and transmission bandwidth.
What problem does this paper attempt to address?