Directional lifting wavelet transform domain image steganography with deep-based compressive sensing

For image steganography, it is necessary to improve the quality of the reconstructed image and stego image as high as possible while maintaining the security of the system. To achieve this goal, we propose a novelty image steganography via deep-based compressive sensing (CS) for the reconstructed image and directional lifting wavelet transform (DLWT) for the stego image. The plain image is first randomly under-sampled and diffused by the measurement matrix and simulated noise to generate the secret image. And the above two matrices were created using a logistic map with two initial values. Then, we embed the secret image into the DLWT domain of the carrier image by singular value decomposition (SVD), resulting in the meaningful stego image. Finally, for enhancing the quality of the reconstructed image from the extracted secret image, we present the deep-based CS reconstruction algorithm. Experimental results verify the effectiveness that the proposed scheme can achieve visual quality, robustness, and security.