High Capacity Reversible Data Hiding Based on Multiple Histograms Modification

Among various techniques of reversible data hiding (RDH), histogram modification has been mostly investigated and used in practice. Recently, a new histogram modification technique, namely, multiple histograms modification (MHM) has been proposed and received the increasing attentions. By MHM, multiple histograms can be modified differently to achieve an excellent embedding performance. However, as the maximum modification on each pixel value is one, only one pair of bins in the prediction-error histogram (PEH) is allowed to be expanded for data embedding. This drawback leads to a low capacity and limits its extension for high capacity scenario. To remedy this, we extend the MHM scheme and propose an efficient solution for the high-capacity embedding. Specifically, instead of only selecting one pair of expansion bins in each PEH, in the proposed method, multiple pairs of bins are utilized for expansion in each PEH. Moreover, the embedding performance is optimized by an advisable expansion bin selection strategy. Consequently, the multiple PEHs are allowed to embed with different capacities according to the sharpness of distribution, i.e., the low-index PEH that consists of more smooth pixels is embedded with a larger capacity than the high-index one. Besides, a simplified parameter determination is developed to seek the optimal solution with a good computational efficiency. The experimental results show that the proposed method can achieve the high capacity and provide a better performance than some state-of-the-art RDH methods.