Reversible Data Hiding in Encrypted Images Based on Adaptive Prediction-error Label Map

In the field of reversible data hiding in encrypted images (RDH-EI), predict an image effectively and embed a message into with lower distortion are two crucial aspects. However, due to linear regression prediction being sensitive outliers, it is challenge improve accuracy predictions. To address this problem, paper proposes RDH-EI scheme based on adaptive prediction-error label map. stage, threshold estimation algorithm local complexity proposed. Then, pixels selection method gradient designed train parameters model. embedding reserve enough space auxiliary information secret by flipping least significant bits (LSBs) encrypt original image. receiver, map, error-free recovery can be achieved. Extensive experimental results have shown that proposed achieve effective pixel get higher rate (ER).