One of the main aim of the research on reversible watermarking is to achieve higher data-hiding capacity with as low distortion as possible. In this study, an effective reversible watermarking framework is proposed. First a host image is down sampled into two sub-images. Then the Difference Expansion (DE) technique is applied to one sub-image and output its location map and watermarked image. Next the suitable pixel pairs in the other sub-image are chosen by a simple predictor to embed more data. Finally the watermarked host image is reconstructed by combination of two watermarked sub-images. The experimental results show the size of compressed location map decreases dramatically by more than half of the traditional one. The schemes based on the framework are effective on different test images and can achieve a high data-hiding capacity.
展开▼