Improved Method of Prediction Error-based Reversible Watermarking for Copyright Protection
Guangyong Gao, Shimao Yao, Zhijun Chong · 2014
The large prediction error of original prediction-error expansion (OPE) reversible watermarking algorithm causes that the embedded image is distorted to some extent. Aiming to the situation, based on extreme learning machine (ELM) and optimized-ELM (OELM), the improved methods are presented, which are called as ELM-PE (EPE) method and OELM-PE (OEPE) method, respectively. Compared with OPE, EPE and OEPE can maintain a better transparency with a more embedding capacity. In addition, due to that OELM is an optimized version of ELM, hence, OEPE has more superior performance than EPE. Experimental results demonstrate the validity of the proposed methods.