Reversible data hiding in encrypted image with separable capability and high embedding capacity

In this paper, a high-capacity reversible data hiding scheme in encrypted image with separable capability is proposed. In order to mask image contents, original image is first encrypted in a blockwise manner based on stream cipher with the encryption key. By the pre-processing of run-length coding (RLC) incorporating with Huffman coding, all encrypted blocks are categorized into two sets according to their encoded lengths. Then, the data hider conducts data embedding for the two types of blocks through different mechanisms and the data-hiding key. For the blocks with shorter encoded lengths, their pixel differences are compressed by RLC; for the blocks with longer encoded lengths, matrix compression is conducted on LSB layers. Thus, a large amount of spare room is vacated to embed length bits, compressed bits, hash bits, location-map bits and additional bits. On the receiver side, through a progressive decryption process, a directly-decrypted image similar with the original image can be obtained with the encryption key; through bit parsing, the embedded additional bits can be correctly extracted with the data-hiding key. With both the two keys, besides direct decryption and data extraction, the blocks of the first type can be recovered by RLC decoding, and the blocks of the second type can be recovered with the assist of coding length judgment and hash authentication. Experimental results demonstrate that, our scheme can generally achieve better performances, including higher embedding rate and better visual quality of the direct-decrypted image and the final recovered image, than some of the state-of-the-art schemes.