Coverless Information Hiding Based On Graphic Semantic Association

With the rapid development of Internet technology,people can use the network to realize real-time interaction of data,which contains confidential information such as personal privacy.In order to ensure the safety of information transmission,information hiding technology has been widely researched and applied.However,the traditional embedded information hiding leaves traces of modification on the carrier,which is difficult to resist steganography analysis and detection,and its own development is therefore limited.Based on the shortcomings of embedded information hiding,scholars have proposed a coverless information hiding framework:Without modifying the carrier directly,the secret information sequence is used to drive the generation of carrier.According to whether the carrier image originates from nature,it can be divided into:coding-mapping formula and structuring-generating formula.Among them,it is difficult to expand the database of coding-mapping coverless that uses image features,resulting in small hiding capacity and a low degree of practicality;and using natural images as a carrier,there might be risks of being traced in the context of Big Data.The fully structured coverless algorithm requires high computing power,and it is difficult to directly generate carrier images with reasonable content.In order to solve the above problems,this paper combines the characteristics of semi-structured and behavioral structured coverless,and proposes a coverless information hiding algorithm based on the semantic association of graphics and text:a preset template constructs a carrier image with reasonable content.The text label realizes the coverless mapping;and builds a practical covert communication system based on this,the specific work content is as follows:First,a set of semi-structural template construction rules was designed.By analyzing the user's social habits on the network platform,a puzzle template of"seeing pictures" is designed,and a structured carrier is generated based on the template rules to realize coverless information hiding.No longer takes the natural image in the coded mapping as the carrier,the entire carrier image is a reflection of virtual information,there is no comparison of the original image,and the construction form is logically reasonable,the content is easy to share and spread,reducing the possibility of third party doubt.Second,the carrier generation is realized based on web crawlers and convolutional neural networks.The upper part of the semi-structured carrier is the image part,and the content is real-time hotspot.It is implemented based on web crawler technology:requesting access to the homepage URL of the specified news website,parsing the response HTML file through Xpath language and matching the news title,using natural language The word segmentation system extracts the keywords of the title and generates template images and topics with single or multiple keywords to meet the construction rules of the preset template.The lower part of the carrier is used for coverless information hiding,which is composed of small-size text labels,and the convolutional neural network is used to realize the construction of the coverless mapping relationship.Third,a complete tag library containing interference samples is constructed to improve the performance of coverless hiding algorithms.This paper conducts transfer learning training on the Alex-Net network through a new database to realize the classification and recognition of text labels;and introduces the confrontation training to improve the recognition accuracy.This mapping method also solves the problem that the features of coding-mapping coverless hiding are difficult to match.After testing,the algorithm proposed in this paper retains the advantages of coverless——anti-detection;the text label has the characteristics of easy identification and small size,which reduces the redundancy in the image content.Comparing horizontally with the existing coverless hiding algorithm,the algorithm proposed in this paper further improves the hiding capacity,has more robust robustness,and has superior comprehensive performance,which broadens the application background of coverless information hiding.Finally,based on this algorithm,a practical covert communication system was built under the environment of Matlab2018a and Python 3.8.