A Prediction-Based Adaptive Reversible Data Embedding Scheme

A reversible data embedding algorithm to adaptively hide a secret message is proposed. The embeddable location is determined by a variance threshold without a location map or overhead information. The proposed algorithm comprises three phases. In the preprocessing phase, the variance thresholds are generated to control the quantity of secret data bits and the image fidelity. In the embedding phase, each block composed of four cover pixels and is manipulated to overlap with blocks to determine a prediction error that arises from the difference between a cover pixel and the average value of its upper and left neighbors. Secret data then embedded in a cover pixel by adding them to the pixel prediction value plus the prediction error. In the extraction and recovering phase, the secret data are extracted in a manner similar to that in which they are embedded, and pixels can be restored to their original values. Experimental results show that the embedding capacity can be adjusted in the proposed scheme. Moreover, the embedding capacity exceeds that of other schemes and the image quality remains satisfactory.