Information Hiding Based on Morphological Component

Abstract

Steganography in sparse domain has drawn more and more attention in the past few years due to its high security. In this paper, we propose a sparse domain steganography based on morphological component for grayscale images. Images are composed of two morphological components—piecewise smooth (cartoon-like) parts and textures. Complex contents of images are harder to be modeled, such as textures, thus cannot easily be detected when we embed secret data in them. By properly select dictionaries, content-adaptive steganography in sparse domain can have rather large payloads and low statistical detectability. We combine two dictionaries to obtain sparse coefficients of morphological components of an image, separately. When embedding in sparse domain, we give top priority to coefficients of textures. We present two ways to construct these two kinds of dictionaries in our work, dictionaries using mathematical models as well as dictionaries wisely learned by K-SVD algorithm. Experiments show better visual quality of stego-images and undetectability of secret messages in comparison with other methods in sparse domain.