High-capacity reversible data hiding based on star-shaped PVO method

Abstract

In recent years, many reversible data hiding (RDH) methods have been proposed. Among them, the method of pixel value ordering (PVO) can generate high-fidelity camouflage images with good embedding capacity. The original PVO method used the block-by-block approach to generate prediction error, where only two pixels can be predicted per block. In this paper, we proposed the “star-shaped” PVO method. Furthermore, to achieve the effect of increasing the embedding capacity, we can maximize all the pixel values in the 3 × 3 block to embed secret data and take advantage of the correlation between adjacent pixels in the natural image. The advantage of this method is that a moderate image quality can be maintained even when multi-layer embedding is performed. Results indicate that the cumulative quantity in smooth images can reach up to 110000 bits and that the embedding capacity of complexity images is also superior to previous PVO series methods. Therefore, any kind of image is suitable for this method.