Conventional image steganalysis mainly focus on presence detection rather than the recovery of the original secret messages that were embedded in the host image. To address this issue, we propose an image steganalysis method featured in the compressive sensing (CS) domain, where block CS measurement matrix senses the transform coefficients of stego-image to reflect the statistical differences between the cover and stego- images. With multi-hypothesis prediction in the CS domain, the reconstruction of hidden signals is achieved efficiently. Extensive experiments have been carried out on five diverse image databases and benchmarked with four typical stegographic algorithms.

Compressive sensing based secret signals recovery for effective image Steganalysis in secure communications

Conventional image steganalysis mainly focus on presence detection rather than the recovery of the original secret messages that were embedded in the host image. To address this issue, we propose an image steganalysis method featured in the compressive sensing (CS) domain, where block CS measurement matrix senses the transform coefficients of stego-image to reflect the statistical differences between the cover and stego- images. With multi-hypothesis prediction in the CS domain, the reconstruction of hidden signals is achieved efficiently. Extensive experiments have been carried out on five diverse image databases and benchmarked with four typical stegographic algorithms.