A new reversible watermarking scheme using the content-adaptive block size for prediction

摘要

Reversible watermarking aims to embed more messages in a cover image with less distortion. For reversible watermarking schemes (RWSs) using the prediction error expansion and histogram shifting (PEE-HS), the key point is to predict image pixels accurately. To obtain more accurate prediction, by taking into account the fact that there exist edges and textures in natural images, this paper designs a local predictor exploiting the content-adaptive block size for each to-be-predicted pixel. Specifically, in constructing the prediction context, the original to-be-predicted pixel is replaced by its rhombus-averaged version, aiming to result in the same prediction context for the embedder and decoder. A number of candidate block sizes are then set and the same number of prediction errors between the rhombus-averaged and predicted values are generated. Considering that the watermarking performance is measured with respect to the original pixels, a mechanism is thus devised to find a preferable content-adaptive block size leading to the minimum prediction error close to the difference between the original and predicted values. After generating prediction errors, we deploy the PEE-HS to design a new RWS exploiting the content-adaptive block size. Experimental results show that the proposed scheme outperforms the-state-of-the-art using the least-square-based predictor. (C) 2019 Elsevier B.V. All rights reserved.