A Reversible Data Hiding Scheme Based on Frequency Domain with Pseudo-Block-Quantization and Prediction-Error-Expansion

Abstract

In this paper, we propose a prediction-error-expansion (PEE) algorithm based on pseudo-block quantization and two prediction-generation methods (PGMs) in the discrete wavelet transform (DWT) domain. Using pseudo-block quantization (PBQ) to construct a pseudo block for generating a prediction value yields a higher probability that a message can be embedded, and using two PGMs makes possible the use of both smooth and rough areas in the frequency domain. In this article, we apply the tools of context-value ordering and context-weight averaging to identify areas that are similar. In the experimental results, we achieve an overall optimization on Embedding efficiency (EE). The results show that our proposed scheme clearly demonstrates better performance than previous scheme.