Adjacent Pixel Values Blocking and Prediction Error Expansion Based High Fidelity Reversible Data Hiding Scheme

Abstract

This paper presents a novel reversible data hiding method based on pixel value blocking and prediction-error expansion. A cover image is divided into non-overlapping sub-blocks of two pixels. Watermark bits are embedded in two phases. In Phase-I, for each sub-block, second pixel value is predicted by the first pixel value and depending on prediction-error within a threshold limit secret bit is embedded into second pixel. Also secret bit is embedded into first pixel just by adjusting the location map without effecting into the pixel value. Then another compressed location map value is used to indicate overflow/underflow or the sub-blocks are outer threshold limit. Again in Phase-II, for each sub-block, first pixel value is predicted by the second pixel value and by depending on prediction-error within a threshold limit secret bit is embedded into first pixel. Also secret bit is embedded into second pixel by adjusting location map without effecting into the pixel value and another compressed location map value is used to indicate overflow/underflow or the sub-blocks of outer threshold limit. All secret bits can be recovered and restored the cover image completely from watermarked image. Experimental result of comparison of this scheme with recent existing scheme using different standard images shows that the embedding capacity with visual quality and PSNR values of the proposed scheme is larger than the existing scheme.