Research Of Image Steganographic Method Based On Noise Visibility Function And Substitution Table

With the development of computer technology and communication network, how to solve the problem of secure transmission have been paid more and more attention. Steganography takes an important role in secure transmission and convert communication, especially in the domain of military affairs and national security which have urgent requirements on the confidentiality and security of information.Steganography based on the characteristic of HVS considered the human visual system sufficiently when embedding the secret data. Human visual system has some masking effects, that is the reason why steganography based on the characteristic of HVS has good visual imperceptibility and large payload. Therefore, the methods based on HVS have been popular in the research of steganography.In this thesis, the theory of information security and information hiding were reviewed firstly. Taking these works as beginning deeply analyzed the characteristic of HVS and studied some traditional steganographic methods based on HVS. Based on the analysis of the existing algorithms, we proposed a novel image steganographic method based on noise visibility function and substitution table.In order to improve the security of steganography, the preprocessing of secret data has been researched and a steganography strategy based on the cryptology was employed in this method. The traditional method of cryptology protected information by encryption itself, while information hiding technology has hidden its transmission process. Therefore, the combination of the traditional cryptology and the steganography, may better guarantee the security of information itself and its transmission process. Based on the researching on the periodicity and the performance of Arnold scrambling, the secret image was scrambled by Arnold transform.Based on the analysis of the characteristic of HVS, we proposed to classify the pixels of host image into some groups according to the noise visibility function values. For each group of host pixels, different strength of secret data was embedded by the method of substitution table associated the modulus function. More secret bits were embedded into the regions which more sensitive to human visual system and few secret bits were embedded into the regions which have some masking effects. We simulated the algorithm and tested the performance of imperceptibility and payload. As experimental results proved, this algorithm can provide better stego image quality with large payloads and the quality of extracted secret image was good. The method proposed in this thesis was feasible.