Prediction error adjustment technique for high quality reversible data hiding

The original modification of prediction error (MPE) method made use of the calculated difference errors and histogram shifting for reversible data hiding which significantly increased image quality and payload in comparison to other similar methods. However, MPE did not fully exploit the merit of prediction since modified pixels were used for predicting errors. In this paper, we proposed a new backward prediction scheme that significantly increased the precision in predictions of the original MPE. Results showed that payload was increased by 15%, significant high PSNR and significant decline in overhead for maintaining the location map.