Efficient color image reversible data hiding based on channel-dependent payload partition and adaptive embedding

Existing color image reversible data hiding (RDH) methods mainly focus on utilizing the similarity between color channels for performance enhancement, but seldom pay attention to the discriminate properties among them. So the payload in these methods is equally partitioned and indiscriminately embedded into each channel. In fact, the prediction-error histogram (PEH) is varied for different channels, and embedding more bits into the one with a higher peak value would be beneficial to the color image RDH. In this work, to better exploit the inter-channel correlation and further enhance the embedding performance, a novel color image RDH scheme based on channel-dependent payload partition and adaptive embedding is proposed. Specifically, the total payload is advisably partitioned and then embedded into each channel according to its PEH, such that the total embedding distortion is minimized. Besides, for a single channel, the adaptive embedding that utilizes the smooth pixels preferentially is adopted by using other channels for reference. Experimental results verify that the proposed method can yield a better performance than some state-of-the-art works.