Universal reversible data hiding for arbitrary RGB-CFA captured images
Wei-Jen Yang, Kuo‐Liang Chung, Le-Chung Lin, Chien-Hsiung Lin · 2013
In this paper, we propose a universal reversible data hiding method for arbitrary red-green-blue (RGB) color filter array (CFA) captured mosaic images. Based on spectral-spatial correlation, a new universal prediction scheme, which automatically calculates the predicted values in the color difference domain according to the corresponding mosaic structure, is first developed to achieve small predicted errors. Then, the proposed embedding scheme utilizes predicted errors in the color difference domain to embed hidden data. The results of experiments on four common types of RGB-CFA mosaic images demonstrate that the proposed universal method can deliver high embedding capacity and good quality of marked images.