In this paper, the authors propose an accurate prediction scheme to improve the histogram modification based reversible information hiding (HMRIH) algorithm. The prediction is based on both local gradients and context pixels. The proposed scheme detects not only the existence of an edge (vertical or horizontal), but also its strength (sharp, normal, or weak). Since such an approach can generate more accurate prediction values, the prediction errors tend to form a much sharper histogram distribution. The experimental results reveal that, without degrading the image quality, the proposed scheme can improve the embedding capacity by 7.9% in average.
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Computer Science Applications
- Electrical and Electronic Engineering