Blind, SVD-based Scheme for Information Hiding in Digital Images

In this paper, we propose a scheme based on the singular value decomposition for information hiding in digital images. The hidden information is embedded into a transparent, color, host image by modifying some singular values of a matrix that represents one image channel with the lowest pixel intensity. In order to blindly extract the hidden information, we develop new embedding and extraction rules, as well as the image segmentation algorithm. We evaluate the proposed scheme regarding three primary requirements: robustness, imperceptibility, and blindness. Experimental results show that the proposed scheme can blindly extract the hidden information from stego images. Also, it is robust against the white Gaussian noise addition when the signal-to-noise ratio is more than 60 dB. Besides, the proposed scheme distorts host image imperceptibly.