Visual image encryption based on compressed sensing and Cycle-GAN

At present, most image encryption schemes directly change plaintext images into ciphertext images without visual significance, and such ciphertext images can be detected by hackers during transmission, and therefore subject to various attacks. To protect the content security and visual safety of images, a learning visual image encryption scheme based on compressed sensing (CS) and cycle generative adversarial network is proposed. First, the secret image is sparse by discrete wavelet transform and compressed by CS. Secondly, the compressed image is permuted and diffused by an improved Henon map to obtain the ciphertext image. Finally, the images are migrated from the ciphertext domain to the plaintext domain by generating an adversarial network to obtain visually meaningful images. We constrain and guide the image generation process by introducing a feature loss function to guarantee the quality of the reconstructed images. Experimental results and security analysis show that the image encryption scheme has sufficient key space, strong key sensitivity, and high reconstruction quality.