Image encryption-authentication scheme based on pixel adaptive diffusion

Traditional image encryption typically achieves the purpose of image protection by transforming ordinary images into noise-like patterns. However, in practice, only image encryption technique cannot meet the security requirement, since the source of image information is always important. To trace the origin of the image, an image encryption-authentication scheme based on pixel adaptive diffusion is proposed. The plaintext image is preliminarily encrypted by a confusion-diffusion operation. Then, a pixel adaptive diffusion mechanism is designed, in which a random value drawn from a pre-shared cipherbook is incorporated as an external parameter to significantly enhance the diffusion randomness and strengthen the resistance against the differential attack. This diffusion mechanism operates independently of plaintext-related keys, thus reducing key consumption while ensuring that even repeated encryption of the same plaintext image can produce distinct ciphertext results. Subsequently, the hidden image is secured by the singular value decomposition ghost imaging (SVDGI) technique, and the resulting ciphered hidden image is embedded into the encrypted plaintext image. Finally, the embedded image undergoes a secondary encryption process to generate the final ciphertext image. During the decryption process, the recipient can reconstruct a high-quality hidden image through the SVDGI to verify the origin of the ciphertext image reliably. Experimental results demonstrate that the proposed image encryption-authentication scheme based on pixel adaptive diffusion can not only verify the image source, but also counteract various common attacks.