Reversible Information Hiding Algorithm Based on Multi-MSB Histogram Shift

Among the reversible data hiding (RDH) algorithms, histogram shift (HS) is simple and relatively high security, due to its advantage of low algorithmic complexity is suitable for lightweight systems, based on this this paper proposes a reversible information hiding method based on the improved histogram shift algorithm. The method divides the complete reversible information hiding process into three phases including predictor selection, pixel selection, and embedding algorithm. Predictor period selection refers to selecting the most applicable prediction method based on the texture structure of the image considering the accuracy and reversibility of the prediction. Pixel selection refers to prioritizing the pixels that are in the smoothing region, and the PSNR of the image can still be kept high after embedding the information. The embedding algorithm of the method can adaptively use the multi-MSBs of the original pixel value and the predicted pixel value to obtain the prediction error, and realize the embedding of pixel values by the multi-expanded bin histogram shifting algorithm. The advantages of this method are (1) Unlike the complete pixel values, the predicted values of the multi-MSBs are more accurate, and a smaller prediction error can be obtained even in rough regions. (2) The use of multi-extended bin histogram shifting algorithm can realize the increase of embedding capacity without increasing the algorithm complexity. (3) The predictor can be selected adaptively according to the image texture. After the experiment, it can be proved that the method can realize larger capacity of information embedding under the guarantee of higher PSNR.