High capacity reversible data hiding algorithm based on central prediction

Abstract

This paper proposes a reversible data hiding algorithm based on central prediction and histogram shift, aiming to increase the embedding capacity at a lower distortion rate. The cover image is scanned and divided into blocks firstly and the value of central pixel of each blocks is took out to predict other values in the same block to obtain the prediction error histogram. Data embedding can be implemented by modifying the prediction error histogram according to the secret data and the prediction error values. Since the value of central pixel is close to other values in the same block, the prediction error histogram in our scheme is steeper and the embedding capacity becomes larger. Experiments show that this scheme has a lower distortion rate and better performance under the same embedding capacity.