A reversible data hiding in encrypted image based on prediction-error estimation and location map

Abstract

In recent decades, visual privacy of digital data has gained much attention from the researchers, especially in cloud-based services. The efficacy of Reversible Data Hiding in Encrypted Images (RDHEI) cannot be ignored as it meets the requirements of visual privacy and data security. Generally, it comprises three different stakeholders namely content owner, data hider, and receiver. The original image is encrypted, by the content owner, using encryption function. After encryption, there is still a possibility for the data hider/cloud-owner to embed the additional data in it. At the receiver end, the embedded data and the original image are recovered losslessly. In our proposed RDHEI scheme, the room is reserved for the data hider before image encryption, by the content owner. Firstly, the image is preprocessed using the prediction-error estimation method to create spare space. Next, the location map is created to capture the information whether a particular location can be used for embedding or not. After this, image encryption is done through standard stream cipher and the compressed location map is embedded. Furthermore, the data hider (without having any knowledge of the original image and the encryption key) embeds the additional data into the Most Significant Bits (MSBs) of the assigned locations using the data hiding key. Finally, at the receiver’s side, the additional data is recovered flawlessly using the data hiding key and the location map. And the original image is reconstructed with the help of the decryption key and the location map. Besides, the approximate image is recovered using the decryption key without the use of location map. Experimental results validate that our proposed scheme outperforms most of the existing schemes in terms of embedding capacity and reconstructed image quality. Additionally, the quality of recovered image without using data hiding key, that is the image recovered prior to losslessly recovered image, is relatively better than most of the methods used in recent works.