| With the development of technologies,people use the Internet for easy transmission.However,in important areas such as business,politics and the military,there is sensitive information that must not be exposed in public channels.To solve this problem,academics often use cryptography with information hiding.However,traditional cryptography,which converts plain text into meaningless garbled code,is vulnerable to strong pointing surveillance.In contrast,information hiding solves this problem by exploiting steganography.In the field of information hiding based on digital images,the need to resolve the conflict between maintaining imperceptibility and image quality while embedding in large volumes.New methods need to be introduced.In addition,in modern user social interaction,the simultaneous embedding of secret information in multiple carriers for transmission is more in line with demand.However,existing algorithms ignore the problem of stealthiest of a batch of carrier secret images in the transmission channel.Specifically,the main objective of this paper is to solve the following two problems: 1)the contradiction between hiding capacity and carrier image quality in the single-carrier information hiding domain;2)The statistical difference between the carrier image set and the original image set is ignored when embedding in the multi-carrier information hiding domain.The main work and contributions of this paper are as follows:(1)Many algorithms have been developed for single carrier data hiding algorithms,but there is still room for improvement in the two metrics of hiding capacity and stegoimage quality.In this thesis,a new multi-pixel value adjustment scheme is proposed.The method uses a strategy with coding functions to further improve the comprehensive performance of digital image data hiding.The specific implementation idea is to divide the adjacent n pixels into two subgroups and perform multi-bit based modulo operations in each subgroup.The MPA algorithm generates a higher quality of payload than other algorithms for the same hiding capacity;and the MPA has the highest hiding capacity for the same quality of payload image comparison.The effectiveness of the scheme was verified and evaluated using the embedding payload,peak signal-to-noise ratio,structural similarity index and quality index.The results show that the PSNR value of 30.01 d B for the steganographic image with a maximum payload of 5 bpp still meets the requirements,while also having a high structural similarity index and a high quality index.In addition,the steganalysis test did not successfully attack MPA.Therefore,in practical application scenarios,the MPA algorithm can be chosen for single carrier-based information embedding to increase the diversity of steganography algorithms.(2)The developing cloud storage technology provides a platform for the transmission and sharing of large volumes of digital images.The multi-carrier data hiding algorithm is not only easier to confuse attackers,but also more in line with practical application scenarios and users’ social behavior.However,existing data hiding algorithms usually treat the selection of carriers and the distribution of payload as two separate problems.The increased statistical variability between the encrypted image set and the natural image set after embedding is not taken into account.Such negligence of one’s own behavioral characteristics would still be exposed under the attacking behavior of the steganography analyst.At the same time,the development of deep learning-based steganography analysis techniques means that scholars have higher security requirements for multi-carrier steganography.Therefore,this thesis proposes a joint batch steganography framework,i.e.joint image carrier selection and embedding payload distribution.The scheme is designed to better adapt to practical application scenarios and user social behavior,and to meet the security requirements of multi-carrier steganography more effectively.The framework combines the analysis of digital image texture features with the use of distance constraint metrics to reselect image carriers and dynamically embed secret data.Unlike existing multi-carrier steganography techniques,the framework uses carrier selection and embedding volume assignment as joint objectives to better account for the behavioral characteristics of the steganographic and to avoid the problem of increasing statistical distance between the set of carrier images and the set of natural images after embedding secret data.The scheme was also examined using several evaluation criteria,including anti-single carrier steganography analysis and anti-batch steganography analysis.The experimental results show that the algorithm can effectively resist traditional steganalysis and improve the steganographic security of the algorithm. |
