Reversible data hiding in multispectral images for satellite communications

Abstract

The Landsat multispectral images are taken by satellites and then transmitted to the monitor center on the ground. Due to that governments and companies sometimes transmit secret messages through the satellite channel, the transmitting process of multispectral images can be a good cover for the secret messages by applying reversible data hiding (RDH) technology. This work proposed a novel RDH method for multispectral images. Before data embedding, each band is creatively matched with a reference band to perform the proposed inter-band complexity computation and inter-band prediction. For the pixels of the to-be-embedded band, we inventively combine the proposed inter-band complexity and the inter-space complexity to classify them into four categories and then process them by the inter-space predictor, inter-band predictor, multi-predictor and second-time prediction, respectively. Moreover, a general complexity is used to control the embedding priority to achieve lower distortion. Compared with the generally used reversible data hiding methods, our proposed method can reach a larger embedding rate and lower distortion for multispectral images.