Reversible Color-to-Grayscale Conversion Using Hierarchical Watermarking Based on Multiple Manifolds

Reversible color-to-grayscale conversion hides the chromatic information of a color image as a watermark into the grayscale version. Later, whenever necessary, the original color image can be recovered based on the embedded data. The main objective of reversible color-to-grayscale conversion is to obtain the best quality possible for both the grayscale and the reconstructed color images. The color information should be embedded as much as possible for color recovery. However, it increases the payload of watermarking in luminance plane in return. In order to overcome this conflict, an innovative highcapacity watermarking is proposed in this paper, which hides color data adaptively and hierarchically according to multiple manifolds in image. Following the orthogonal transform to get the chromatic and luminance planes, the proposed scheme first divides the grayscale image into manifolds, which reflect pixel complexity. Then different payload of watermarking is embedded into manifolds adaptively. In this way, the watermarking capacity can be doubled on average, compared with the LSB method. Experimental results demonstrate the effectiveness of the scheme.