Reversible data hiding and authentication scheme for encrypted image based on prediction error compression.

In the most existing reversible data hiding schemes for encrypted images, cover images can be reversibly recovered, but the integrity of image content cannot be guaranteed. This paper proposes a reversible data hiding and authentication scheme for encrypted images to implement reversible recovery and content authentication of cover images and secret data. The content owner employs a new predictor ISGAP to generate more accurate predictions and smaller errors. The errors are then compressed with adaptive Huffman coding to enlarge the embedding space, and the plaintext authentication information is embedded in it. The data hider embeds secret data into the encrypted image along with ciphertext authentication information. The receiver performs ciphertext authentication first and then implements cover recovery and plaintext authentication according to different keys. Experiments were carried out with 100 images selected from each dataset of BOSSbase and BOWS-2, and the results show that the scheme has higher embedding capacity and can effectively implement image content authentication while ensuring high security and reversible recovery.