A visually secure image encryption method based on semi-tensor product compressed sensing and IWT-HD-SVD embedding

The conventional approach for images encryption entails transforming a regular image into an encrypted image that resembles noise. However, this noise-like encrypted image is susceptible to drawing the attention of an attacker when transmitted through a public channel. Hence, there has been a recent surge in the interest of academics towards visually secure image encryption techniques. In a broad sense, encryption methods that include visual significance should prioritize four key elements: the resemblance between the cypher picture and the carrier image, the capacity for embedding, the attainment of good recovery quality, and resilience against many forms of attacks. To address the issues pertaining to inadequate visual security, limited resistance against attacks, and subpar quality of reconstructed images observed in contemporary image encryption and compression methodologies. This paper proposes a visually secure image encryption method based on improved semi-tensor product compressed sensing, two-way cross zigzag obfuscation, and IWT-HD-SVD embedding. Firstly, the plain image is sparsely represented in the Discrete Wavelet Transform (DWT) domain, and a two-way cross zigzag mismatch strategy is proposed to disarrange the coefficient vectors. Then the plain image is encrypted as a secret image by the improved semi-tensor product compression sensing technique. After that, IWT-HD-SVD embedding technique is proposed to embed the secret image into the carrier image to generate the final meaningful cryptographic image. This dramatically improves the visual security of the cryptographic image. Simulation results show that the quality of the decrypted image is approximately 36 dB and up to 44 dB. In addition, the cryptographic image is highly robust against common noise attacks of 0.05 %.