An Extended Reversible Data Hiding Method for HDR Images Using Edge Estimation

We extend a reversible data hiding method for high dynamic range (HDR) images with RGBE format in this paper. In RGBE format, the edge distribution in each channel is similar to those in the other channels. The proposed method focuses on this feature and improves the edge estimation algorithm of the conventional method to detect more accurate prediction values. Optimal prediction values are first obtained for target pixels based on the edge estimation, and then the payload is embedded using a histogram of the prediction errors. Our experimental results show that the prediction values, which are close to the target pixel values, can be obtained, and thus the hiding capacity of the proposed method is higher than that of the conventional method.