A Coverless Text Steganography by Encoding the Chinese Characters' Component Structures

Introduction

Images, audios, and videos have been widely employed as carriers for steganography (Fridrich, 2009; Lazic & Aarabi, 2006; Mastafa et al., 2017). Compared with these media, a text usually has few redundancies to be a carrier, and this leads to the difficulty of achieving text steganography and small steganographic capacity. But a text is widely used on the Internet, which makes it not prone to be suspected of a carrier and this is an indispensable feature of best steganography carriers (Zielińska et al., 2014). Therefore, how to accomplish text steganography and how to achieve a high capacity have become a research hotspot nowadays.

The traditional text stego-systems (Bennett, 2004; Richard, 2004) are mainly implemented by modifying a text by leveraging its redundancy in text formats, the characters’ appearances, text syntax or semantics, and so on. Due to the aforementioned issues, the emerging trends in text steganography is to generate a new text or select a text to be a carrier, which is usually called “coverless text information hiding” (Chen et al., 2015; Ali, 2018; Wang, 2019), instead of modifying a text. Herein, “coverless” doesn’t mean no carrier to convey a secret, but actually, it performs no modification to a carrier. Compared with steganography by modification, the coverless steganography is free of comparison attacks, what’s more, the steganography by selection doesn’t suffer from semantic attacks, which will be further studied in the following.

This paper proposes a new coverless text steganography method by encoding Chinese characters’ component structures (CCCSs). A Chinese character usually consists of single or multiple components, which are assembled to form different CCCSs. These structures can be grouped according to their usage frequencies and every group will be assigned to express a different Binary Digital String (BDS). The characters in a text are organized in a square matrix which is reshaped to get the Minimal Square Matrix (MSM) who should include all the structure groups. The MSM will be further transformed into a Code Square Matrix(CSM). A secret message is converted into a secret BDS which will be split into binary digital slices (BiDSs) that are elements in the CSM; the row and column numbers of these slices in the CSM are organized to construct two systems of linear remainder equations, which can be solved according to the Chinese remainder theorem, and their solutions are taken as the keys shared with the receiver. The receiver figures out the corresponding row and column numbers by calculating the remainder using the shared key, and gets the slices, then reconstructs the complete BDS, finally converts it to the original secret.