An asymmetric multi-level image privacy protection scheme based on 2-D compressive sensing and chaotic system

In the current digital and network era, digital images play a crucial role across various domains, such as social media, healthcare and security surveillance. However, these images typically contain sensitive personal information, and if appropriate security measures are not taken during collection, transmission, or processing, there may be a serious risk of privacy breaches. To solve these problems, we propose an asymmetric multi-level image privacy protection scheme based on 2-D compressive sensing and chaotic systems. The proposed solution exhibits significant advantages compared with the existing methods in several aspects. Firstly, the image data is compressed and sampled using the public-key sampling matrix instead of the private-key sampling matrix, and asymmetric encryption is applied, to ensure the broad applicability of the proposed solution in various scenarios. Secondly, by combining 2-D CS with the iterative gradient projection reconstruction algorithm accompanied by sensitive region decryption (IGPRA-ASRD), it effectively addresses the single privacy protection needs in digital images while demonstrating excellent scalability, thus making it applicable for solving the challenges of multi-tiered privacy protection. Lastly, the introduced shared key mechanism effectively addresses key management issues, ensuring the secure distribution of keys. Experimental results and comparative analyses demonstrate that the proposed scheme exhibits excellent effectiveness, compressibility and security. The approach not only protects privacy at a single level but also provides a robust solution for hierarchical protection of multiple privacies in the context of digital image security.