The significant essential of secret image sharing approaches is that the revealed content of the secret image must be lossless. Moreover, the distorted stego images can be reverted to the original cover image. In order to achieve these purposes, we first transform the secret pixels into the m-ary notational system. Then, the information data used to reconstruct original pixels from camouflaged pixels are calculated. The information data and transformed secret data are shared using the (t, n)-threshold sharing scheme. In this way, we can retrieve the lossless secret image and reverse the stego image to the original image. According to the experiments, the shadows can be successfully camouflaged in the cover image, and the stego images have satisfactory quality. Moreover, our scheme allows for a large capacity of embedded secret data.
Proceedings of the 2009 International Conference on Computational Intelligence and Natural Computing, CINC 2009 :253-256