Reversible data hiding in encrypted images using histogram modification and MSBs integration

Through cloud computing development, image encryption is employed to protect content-owner privacy. By the way, the need for embedding secret data in an encrypted image is the reason introducing reversible data hiding in an encrypted image. We may vacate the room before, by or after encryption to embed data. On the receiver side, data extraction can be done separately or jointly with the original image reconstruction. Meanwhile, presenting a separable scheme by vacating the room after encryption is more practical and more challenging than others which we do in this paper. In this way, we employ some notions of reversible data hiding (RDH) in a plain image, such as histogram modification and prediction-error computation. The most significant bits (MSBs) of the encrypted pixels are integrated to vacate the room for embedding data bits. The integrated MSBs would be more robust against failure of lossless reconstruction if modified for data embedding. Meanwhile, a data hider who embeds data is entirely blind to the original content. On the recipient side, data may be extracted error-freely and the original image may be reconstructed losslessly. We employ the chessboard predictor for the lossless reconstruction of the original image through prediction-error analysis. Integration of the encrypted pixels is an alternative method of compressing the encrypted pixels. Experimental results confirm that the proposed integration method outperforms the prior arts.