A low bit-rate SOC-based reversible data hiding algorithm by using new encoding strategies

Abstract

Search order coding (SOC) benefits a lot from the correlation of neighboring blocks for vector quantization (VQ)-compressed images. SOC selects a number of different indices in its search path as candidate search order codes. In this work, we present a low bit-rate SOC-based reversible data hiding algorithm benefiting from novel encoding strategies which exploit the information of the placeholders. The blocks are classified into two categories by the placeholders, where different encoding strategies are designed, respectively. Firstly, for the block in smooth region, the placeholders in its neighborhood are employed to compress the VQ index. Secondly, for the block in complex region, SOC is employed to compress the VQ index. Finally, for the blocks that cannot be processed by its placeholders and SOC, an effective prediction method named accurate gradient selective prediction (AGSP) and Huffman coding are introduced. After encoding phase, the size of the output bit stream is reduced so that space is saved for data embedding. Experiment results show that our proposed method outperforms other state-of-the-art SOC-based algorithms in bit rate and embedding capacity.