A novel approach of data hiding in video using region selection and PCA

With the ubiquitous progress of information technology it is now possible to transfer multimedia information rapidly over the Internet. Significant growth of video data on the Internet insists the users towards video steganography as a popular choice for data hiding. Steganography algorithm must emphasis to improve the embedding efficiency, payload and robustness against the intruders. In this paper, we have addressed those issues and present a new approach of steganography. Our segmentation process is based on video frames. We apply a region selection method followed by the dimensionality reduction process, called principal component analysis (PCA), to compress the regions and embed secret data on those compact regions. This PCA is used as a best-fitted vector that minimizes the average square distance from the pixel values to that vector. Our results show higher embedding capacity along with better visual quality. Moreover, the proposed method improves the robustness in the sense that the secret message can be retrieved by the receiver even after some known attacks on the channel.