Information Hiding Model Based on Channel Construction of Orthogonal Basis

Introduction

With the development of cryptography, computer and other technologies, information hiding has become a hot issue in information security research. Especially, pure ciphertext communication tends to attract the attention of non-cooperative recipients. The main forms of expression are as follows: To take some multimedia information as the carrier, such as image, video, voice and so on. To superimpose the secret information on the multimedia information in some way, so as to reduce the sensitivity of the non-cooperative recipient to the secret information and improve the performance of anti-steganography analysis without prior knowledge. In this kind of research, multimedia carrier does not have the significance of information transmission, but only the hidden function of secret information (Wang, 2011; Sun, 2016). According to the change method of secret information to multimedia carrier, this research direction is mainly divided into two categories:

• 1)

  The secret information is superimposed directly on the multimedia carrier, and most of the carriers of this kind of methods are images. Usually, the secret information is realized by changing the characteristics of the image according to a certain algorithm or law. (Mehmet, Gaurav, Ahmet & Eli, 2005) use the hiding information to change the gray distribution of the whole image. (Qin, Chang & Chen, 2013) use the hiding information to change the smoothing conditions of the edge region. (Wu & Huang, 2012) use similar method that use the hiding information to change the gray of image. It may be common to use secret information absolute or difference to change the gray distribution of the whole or redundant region of the image, and so on . (Wang, Sun & Li,2019) is a recently method to complete the transmission of secret information by using interpolation algorithm with the change of image gray value. This method expands the information capacity to a certain extent.

• 2)

  The carrier of this kind of methods is more diverse, which superimpose the secret information indirectly on the multimedia carrier. For example, the multimedia carrier is changed in the transform domain by changing the coefficient of the transform domain with the secret information, such as Fourier, wavelet and other transform domain methods (Vaishnavi & Subashini, 2015; Islam & Laskar,2017). (Wu & Wu, 2016) hided the secret information in areas where the human ear is less concerned and the degree of redundancy is higher, using different domains corresponding to different scales of wavelet bases in voice.

Among them, method 1) often need secret key to explain the algorithm and law of changing carrier with secret information. Although the methods and algorithms to realize covert communication are relatively simple, and the secret information can be solved completely, the transmission problem of secret key (including the design of secret key, transmission mode, etc.) must be considered in the construction of covert communication. At the same time, the secret information should be reasonably designed to change the threshold control of the multimedia carrier, otherwise it will make a great change to the carrier, and the ability of anti-steganography analysis is weak. The central idea of method 2) is to disperse the secret information into the carrier. And the security of the secret information is improved compared with 1). But the mapping method of the transform domain makes it more difficult for the receiver to recover secret information. Therefore, the research of covert communication basically focuses on two aspects:

• 1)

  No key or special key;

• 2)

  Seeking the balance between less change of multimedia carrier and undistorted recovery of secret information.

With two aspects, the information hiding model is constructed by orthogonal basis. And the orthogonal basis is get by media carrier’s orthogonal decomposition basis. The orthogonal decomposition basis constructs two independent channels for carrier and secret information. Because the basis decided by carrier, it matches 1). And the secret information is scattered over the carrier, without changing special characters of carrier. So the secret information cannot be easily detect. By orthogonal basis get from carrier’s self-correlation, the secret information is not confined to the same type with carrier. For example, if carrier is image, the secret information’s type is image too. The capacity of the secret information is evaluated by type and encode. So the model get better flexibility。