High capacity reversible data hiding in encrypted images using prediction error encoding

Abstract

Reversible data hiding in encrypted images (RDHEI) is a popular way for embedding the information in cover images without knowing their original contents. The RDHEI techniques are quite useful for annotating management data to encrypted images in cloud storage for their effective handling. In this paper, a high capacity RDHEI technique is proposed using prediction error encoding. The proposed RDHEI technique applies gradient edge detection predictor to predict the pixel values of nearby pixels of original image and calculates the prediction error. The host image is then encrypted using stream cipher to encode its contents. Next, the prediction errors are encoded using Huffman encoding to condense their size which helps in embedding the additional amount of secret message bits. Finally, the vacated room is exploited to embed the secret message bits by replacing the MSBs. Experimental results show that the proposed technique maximizes the embedding rate and also ensure security of the contents while comparing with the state-of-the-art techniques. More specifically, the proposed technique provides 1.384 and 4.202 bits per pixel embedding capacity, provide for Baboon and Airplane images, respectively.