High performance reversible data hiding for block truncation coding compressed images

Reversible data hiding has been a hot research topic because it can recover both the host media and hidden data without distortion. Because most digital images are stored and transmitted in compressed forms, such as JPEG, vector quantization, and block truncation coding (BTC), the reversible data hiding schemes in compressed domains have been paid more and more attention. Compared with transform coding, BTC has a significantly low complexity and less memory requirement, it therefore becomes an ideal data hiding domain. Traditional data hiding schemes in the BTC domain modify the BTC encoding stage or BTC-compressed data according to the secret bits, and they have a relatively low efficiency and meanwhile may reduce the image quality. This paper presents a novel reversible data hiding scheme based on the joint neighbor coding technique for BTC-compressed images by further losslessly encoding the BTC-compressed data according to the secret bits. First, BTC is performed on the original image to obtain the BTC-compressed data that can be represented by a high mean table, a low mean table, and a bitplane sequence. Then, the secret data are losslessly embedded in both the high mean and low mean tables. Our hiding scheme is a lossless method based on the relation among the current value and the neighboring ones in mean tables. In addition, it can averagely embed 2 bits in each mean value, which increases the capacity and efficiency. Experimental results show that our scheme outperforms three existing BTC-based data hiding works, in terms of the bit rate, capacity, and efficiency.