Lossless Data Hiding for Color Images

In this paper, a novel lossless data hiding scheme is presented to exhibit high data-hiding capacity for color palette images. In the proposed method, each index value is predicted by using the average of its four precedent neighbors. If the difference between the index value and its predictive value is larger than a threshold, the pixel will be skipped. Otherwise the four precedent neighbors are used to predict the current index during data embedding and retrieving. Moreover, the difference expansion algorithm is used for lossless hiding secret data into prediction error. The prediction error of the indexed image will also be reduced. The algorithm allows the data embedding process to be reversed and recovers the exact original indexed image without any distortion at all. The experimental results show that the proposed method not only can improve the payload capacity but also can maintain good image quality.