CN114051082B - Steganography detection feature selection method and device based on distortion degree and information gain ratio - Google Patents
Steganography detection feature selection method and device based on distortion degree and information gain ratio Download PDFInfo
- Publication number
- CN114051082B CN114051082B CN202111213537.2A CN202111213537A CN114051082B CN 114051082 B CN114051082 B CN 114051082B CN 202111213537 A CN202111213537 A CN 202111213537A CN 114051082 B CN114051082 B CN 114051082B
- Authority
- CN
- China
- Prior art keywords
- image
- feature
- carrier image
- detection
- component
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 132
- 238000010187 selection method Methods 0.000 title claims abstract description 24
- 238000012549 training Methods 0.000 claims abstract description 18
- 238000005259 measurement Methods 0.000 claims description 4
- 230000002349 favourable effect Effects 0.000 claims 2
- 238000012217 deletion Methods 0.000 claims 1
- 230000037430 deletion Effects 0.000 claims 1
- 238000000034 method Methods 0.000 abstract description 63
- 238000002474 experimental method Methods 0.000 description 14
- 230000000694 effects Effects 0.000 description 10
- 230000008569 process Effects 0.000 description 9
- 238000012360 testing method Methods 0.000 description 6
- 238000004458 analytical method Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 238000004891 communication Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000006835 compression Effects 0.000 description 1
- 238000007906 compression Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000005484 gravity Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 238000004003 stimulated Raman gain spectroscopy Methods 0.000 description 1
- 239000013598 vector Substances 0.000 description 1
Classifications
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/32—Circuits or arrangements for control or supervision between transmitter and receiver or between image input and image output device, e.g. between a still-image camera and its memory or between a still-image camera and a printer device
- H04N1/32101—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title
- H04N1/32144—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title embedded in the image data, i.e. enclosed or integrated in the image, e.g. watermark, super-imposed logo or stamp
- H04N1/32149—Methods relating to embedding, encoding, decoding, detection or retrieval operations
- H04N1/32203—Spatial or amplitude domain methods
-
- H—ELECTRICITY
- H04—ELECTRIC COMMUNICATION TECHNIQUE
- H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
- H04N1/00—Scanning, transmission or reproduction of documents or the like, e.g. facsimile transmission; Details thereof
- H04N1/32—Circuits or arrangements for control or supervision between transmitter and receiver or between image input and image output device, e.g. between a still-image camera and its memory or between a still-image camera and a printer device
- H04N1/32101—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title
- H04N1/32144—Display, printing, storage or transmission of additional information, e.g. ID code, date and time or title embedded in the image data, i.e. enclosed or integrated in the image, e.g. watermark, super-imposed logo or stamp
- H04N1/32149—Methods relating to embedding, encoding, decoding, detection or retrieval operations
- H04N1/32267—Methods relating to embedding, encoding, decoding, detection or retrieval operations combined with processing of the image
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Signal Processing (AREA)
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
- Image Analysis (AREA)
Abstract
The invention belongs to the technical field of steganography detection feature selection, and particularly relates to a steganography detection feature selection method and device based on a distortion degree and an information gain ratio; then, the distortion degree value and the information gain ratio of each steganographic detection characteristic component are arranged in descending order respectively; then, according to the two arranged column numbers, deleting the steganographic detection characteristic components with larger difference of arrangement sequences; and finally, training and detecting the reserved steganography detection characteristic component as a finally selected steganography detection characteristic. The method can effectively reduce the DCTR feature dimension while maintaining or even improving the detection precision of the carrier density image, thereby reducing the space-time complexity of detecting the carrier density image.
Description
Technical Field
The invention belongs to the technical field of steganography detection feature selection, and particularly relates to a steganography detection feature selection method and device based on distortion degree and information gain ratio.
Background
Steganography, another expression of covert communication, is a technique of hiding a message in an object that is not easily suspected and then transmitting it to an intended recipient, and has received a great deal of attention in the field of information security in recent years. Where digital media "steganography" is engaged in activities that jeopardize national security, as it may be used by illegal organizations, etc. for covert communications. The corresponding attack technology is steganalysis, which aims to extract the hidden information so as to achieve the aim of resisting the steganography technology and protecting national security.
With the rapid development of digital media, how to improve the speed and accuracy of steganography detection is a problem to be solved. Therefore, the steganography detection algorithm based on digital image self-adaption is a direction which is very interesting to current scholars, and the steganography detection algorithm is mainly used for training and detecting by utilizing an integrated classifier through extracting steganography detection characteristics, so that a good detection effect can be obtained. At present, a series of Gao Weiyin write detection algorithms have been developed by scholars. Although Gao Weiyin writing detection features achieves higher detection accuracy for image self-adaptive steganography, the dimension of the self-adaptive steganography detection algorithm for extracting the steganography detection features is higher, so that higher space-time complexity is brought to detecting the secret-loaded image, and development of quick steganography detection is affected. Therefore, how to select the features with large contribution to detection reduces the dimension of the steganography detection features, and further reduces the space-time complexity of detecting the secret image, which becomes the center of gravity of the current steganography detection field research.
At present, a series of researches on the selection and dimension reduction of steganographic detection features are carried out by some students. These research methods can be classified into general and specific feature selection methods according to objects to which the feature selection method is applied. The general steganography detection feature selection method is suitable for measuring various detection features, measuring the contribution of feature components to the detection secret image, and selecting feature components with large contribution to the detection secret image as feature vectors for training and testing. The specific steganographic detection feature selection method is a selection method aiming at a certain steganographic detection feature. The calculation of the feature selection method is simpler than the general feature selection method, but the application range is narrower.
To date, several studies have achieved different steganalysis feature selection effects, such as CC-PEV, GFR, CC-JRM, SRM, J +SRM features. However, the existing method has unsatisfactory selection effect on DCTR features, and has the problems of excessively high selected feature dimension, excessively low detection precision and the like.
Disclosure of Invention
In order to greatly reduce the characteristic dimension of DCTR without affecting the detection precision, the invention provides a steganography detection characteristic selection method (abbreviated as S-FUND method) and a steganography detection characteristic selection device based on the distortion degree and the information gain ratio, which improve the detection precision of a secret image and reduce the characteristic dimension at the same time, thereby reducing the space complexity of the secret image; and the dependence on classification results can be avoided, so that the time complexity of detecting the loaded image is reduced.
In order to achieve the above purpose, the present invention adopts the following technical scheme:
the invention provides a steganography detection feature selection method based on distortion and information gain ratio, which comprises the following steps:
measuring a difference between the carrier image and the secret image for each steganographically detected feature component using the distortion degree and the information gain ratio;
the distortion degree value and the information gain ratio of each steganographic detection characteristic component are arranged in descending order respectively;
deleting the steganography detection characteristic components with larger difference of arrangement sequences according to the two arranged column numbers;
and training and detecting the reserved steganography detection characteristic component as a finally selected steganography detection characteristic.
Further, the distortion degree K of the carrier image before and after steganography is measured by using the formula (1) i The formula is as follows:
in the method, in the process of the invention,and->Values representing the ith steganographic detection feature component in the carrier image and the carrier image, respectively,/-, respectively>And->Values of the ith steganographic detection feature component in the jth carrier image and the carrier image are respectively represented; k (K) i The larger the value, the greater the degree of distortion that the carrier image produces after embedding the information, indicating that the greater the difference between the carrier image and the carrier image is for the steganographic detection feature component, the more advantageous the feature component is for detecting the carrier image, and the more should be preserved.
Further, the information gain ratio is defined as the ratio of the information gain value of the feature component between the carrier image and the carrier density to the partial entropy of the feature component in the carrier image with respect to the feature component in the carrier image, and is measured by using the formula (2)The formula is as follows:
in the method, in the process of the invention,information gain value representing characteristic component between carrier image and carrier density,/for>Representing partial entropy of the median of the feature component in the carrier image with respect to the median of the feature component in the carrier image; />The larger the value, the more the steganographic detection feature component isThe larger the difference between the carrier image and the secret image, the more advantageous the feature component is in detecting the secret image, the more should be preserved.
Further, the formula of the information gain value of the feature component between the carrier image and the carrier density is as follows:
in the method, in the process of the invention,and->The information entropy values of the characteristic components in the carrier image and the secret image are respectively represented, conditional entropy value of feature component in carrier image median under the condition of representing feature component in carrier image median, +.>Representing the joint entropy of the feature components between the carrier image and the secret image.
Further, the distortion degree value and the information gain ratio value of each steganographically detected feature component are arranged in descending order, respectively, including:
according to K i The steganographic detection feature components are arranged in descending order of values;
then according toValue descending order hiddenThe detection feature component is written.
Further, the deleting the steganographic detection feature component with a large difference in arrangement order according to the two arranged column numbers includes:
according to the two sorting results, calculating the difference of the ith characteristic component according to two criteria;
the feature components whose absolute values of the arrangement order differences are larger than the threshold T are deleted.
The invention also provides a steganography detection characteristic selection device based on the distortion degree and the information gain ratio, which comprises the following steps:
a measurement module for measuring a difference between the carrier image and the secret image for each steganographically detected feature component using the distortion degree and the information gain ratio;
the descending order arrangement module is used for descending order of the distortion value and the information gain ratio of each steganography detection characteristic component;
the deleting module is used for deleting the steganography detection characteristic components with larger difference of arrangement sequences according to the two arranged column numbers;
and the training module is used for training and detecting the reserved steganography detection characteristic component serving as a finally selected steganography detection characteristic.
Compared with the prior art, the invention has the following advantages:
according to the steganography detection feature selection method based on the distortion degree and the information gain ratio, firstly, the difference of each steganography detection feature component between a carrier image and a carrier image is measured by using the distortion degree and the information gain ratio; then, the distortion degree value and the information gain ratio of each steganographic detection characteristic component are arranged in descending order respectively; then, according to the two arranged column numbers, deleting the steganographic detection characteristic components with larger difference of arrangement sequences; finally, training and detecting the reserved steganography detection feature component as a finally selected steganography detection feature; the method can effectively reduce the DCTR feature dimension while maintaining or even improving the detection precision of the carrier density image, thereby reducing the space complexity of detecting the carrier density image, and can greatly improve the operation efficiency, thereby reducing the time complexity of detecting the carrier density image by the classifier and reducing the detection cost.
Drawings
In order to more clearly illustrate the embodiments of the invention or the technical solutions in the prior art, the drawings that are required in the embodiments or the description of the prior art will be briefly described, it being obvious that the drawings in the following description are only some embodiments of the invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
FIG. 1 is a flow chart of a steganography detection feature selection method based on a distortion degree and an information gain ratio according to an embodiment of the present invention;
FIG. 2 is a process diagram of a steganographic detection feature selection method based on a distortion degree and an information gain ratio according to an embodiment of the present invention;
FIG. 3 is a graph comparing the S-FUND method of the invention before and after selecting DCTR features.
Detailed Description
The following description of the embodiments of the present invention will be made clearly and completely with reference to the accompanying drawings, in which it is apparent that the embodiments described are only some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
The larger the difference of the steganographic detection characteristic component between the carrier image and the secret image is, the more the carrier image and the secret image are favorably distinguished, so that the more the secret image is favorably detected; on the other hand, a feature component having little or no difference between the carrier image and the secret image has too little effect on detecting the secret image, and is regarded as a useless feature. These features can lead to an increase in feature dimensions, leading to unnecessary detection space-time overhead, which is unfavorable for the application of steganography detection and hinders the development thereof. Therefore, a feature component that contributes as much as possible to distinguishing the carrier image from the secret image is selected, and in order to measure the difference between the carrier image and the secret image of the steganographically detected feature component, a distortion degree and an information gain ratio are introduced herein.
As shown in fig. 1 and 2, the present embodiment provides a steganography detection feature selection method based on a distortion degree and an information gain ratio, which includes the following steps:
step S11, measuring the difference of each steganographically-detected characteristic component between the carrier image and the secret image by using the distortion degree and the information gain ratio; thus, the effects of the distortion degree and the information gain ratio are regarded as the same, and the reliability of the selected characteristics is increased.
Distortion-based metrics: after secret information is embedded by using a steganography algorithm, part of characteristic components of the carrier image are changed, so that differences exist between the carrier image and the secret image, and the characteristic components with larger differences between the carrier image and the secret image are more beneficial to distinguishing the carrier image from the secret image in consideration of the fact that the degree of change of all the characteristic components is not the same. The carrier image is distorted to different degrees due to the embedding of the information, and the distortion degree K of the carrier image before and after steganography is measured by using the formula (1) i The formula is as follows:
in the method, in the process of the invention,and->Values representing the ith steganographic detection feature component in the carrier image and the carrier image, respectively,/-, respectively>And->Representing the ith steganographic detection feature component at the jth carrier map, respectivelyValues of the image and the carried image; k (K) i The larger the value, the greater the degree of distortion that the carrier image produces after embedding the information, thus indicating that the greater the difference between the carrier image and the carrier image is for the steganographic detection feature component, the more advantageous the feature component is for detecting the carrier image, the more should be preserved.
A metric based on an information gain ratio:
the information gain is currently used to measure the difference between the feature component and the carrier image, the formula is as follows:
in the method, in the process of the invention,information gain value representing characteristic component between carrier image and carrier density,/for>Andthe information entropy values of the characteristic components in the carrier image and the secret image are respectively represented, conditional entropy value of feature component in carrier image median under the condition of representing feature component in carrier image median, +.>Representing the joint entropy of the feature components between the carrier image and the secret image. />The larger the value, the larger the information gain of the steganographic detection feature component between the carrier image and the carrier image, thereby indicating that the larger the difference between them, and thus the more advantageous the detection of the carrier image.
However, when the value of the feature is more, a more definite subset is easily obtained according to the feature division, namelyThe value is lower, again because +.>Since the information gain is larger because of the fixed value, if the information gain is used as the basis for selecting the features, the selection of the features with more values is favored. To solve this problem, the difference between the carrier image and the carrier image of the feature component is steganographically detected using an information gain ratio defined as the ratio of the information gain value of the feature component between the carrier image and the carrier density to the entropy of the feature component in the carrier image median with respect to the feature component in the carrier image median, using equation (2)>The formula is as follows:
in the method, in the process of the invention,representing partial entropy of a median value of a feature component in a carrier image with respect to a median value of the feature component in a carrier image；/>The larger the value, the larger the difference between the carrier image and the secret image, the more advantageous the feature component is in detecting the secret image, the more should be preserved.
Step S12, the distortion value and the information gain ratio of each steganographic detection feature component are arranged in descending order.
Step S13, deleting the steganographic detection characteristic components with larger difference of arrangement sequence according to the two arranged column numbers.
Specifically, according to K i The steganographic detection feature components are arranged in descending order of values and then are based onThe steganographic detection feature components are arranged in descending order of value.
And step S14, training and detecting the reserved steganographic detection characteristic component as a finally selected steganographic detection characteristic.
Specifically, according to the two sorting results, calculating the difference of the ith characteristic component according to the two criteria; the feature components whose absolute values of the arrangement order differences are larger than the threshold T are deleted.
Thus, the difference of the steganographically detected feature component between the carrier image and the secret image is measured herein using the distortion degree and the information gain ratio; and, regarding the difference between the carrier image and the secret image of the two criterion measurement feature components as the same, selecting the feature component with smaller information gain ratio and distortion degree arrangement difference. The method greatly reduces the feature dimension, thereby reducing the space-time complexity of detecting the carrier image.
The time complexity of the main steps in the S-FUND method provided herein is analyzed one by one, and compared with the time complexity of the classification result of the Fisher linear discrimination integrated classifier, so that the performance of the method can be better known.
The S-FUND method provided herein mainly comprises the following steps: calculating the distortion value and the information gain ratio, arranging the characteristic components in descending order according to the distortion value and the information gain ratio, deleting the characteristic components with larger difference in arrangement sequence, and the like, and analyzing the time complexity of different steps respectively, as shown in table 1.
TABLE 1 time complexity analysis of the main steps
There is no nesting relationship of steps in table 1, so the time complexity of the S-FUND method presented herein is equal to the maximum of all steps. When O (Nlog) 2 N). Ltoreq.O (Nn), i.e. log 2 When N is less than or equal to N, the time complexity of the S-FUND method is O (Nn); when log 2 N>n, the time complexity of the S-FUND method is O (Nlog 2 N). However, most of the existing feature selection methods rely on the classification result of the Fisher linear discriminant integrated classifier, and the time complexity of the Fisher linear discriminant integrated classifier is as follows:
wherein L represents the number of individual learners, N trn Represents the number of training sets of each type, d sub Representing subspace dimensions. So the time complexity O (FLD) of this class of selection method depend ) Necessarily greater than or equal to O (FLD), i.eFrom this, the time complexity of the selection method depending on Fisher linear discriminant integrated classifier results is far greater than that of O (Nn) or O (Nlog) 2 N). Because the DCTR feature dimension is 8000, n<N and log 2 N<N. The S-FUND method has a time complexity less than that of PCA-D, steganalysis-alpha, fisher-G and SRGS methods, similar to the CGSM method. Therefore, the S-FUND method greatly reduces the time complexity of operation and improves the efficiency of detecting the loaded image.
To examine the performance of the S-FUND method presented herein, we performed a series of selection and comparison experiments using the 8,000-dimensional DCTR steganography detection feature. All experiments utilize images in a BOSSBase1.01 image library, the types of the images are gray images, the formats of the images are JPEG formats, and the images are operated in MATLAB R2018a of an Inteli7-8550UCPU and 8G RAM computer, so that different methods can be guaranteed to be compared fairly. The experimental results plot was processed and generated in Originp Pro8.5.
1. Subject setting
Computer software, hardware, an image library and a steganography detection feature used in all experiments are the same, so that different methods can be compared fairly, and the experiments are more reliable.
Break Our Steganographic System (BOSS) is the first implementation of image steganography and steganography analysis image library from theory to practical application. A series of operations were performed on the BOSSBase1.01 image library of the website (the BOSSBase1.01 image library was derived from the website: http:// dde. Binghamton. Edu/download /), and the following steps were performed to prepare for the next experiment:
(1) The 10,000 PGM format images in the BOSSbase1.01 image library were converted to JPEG images with a compression quality factor of 95.
(2) 10,000 JPEG carrier images were generated using the SI-UNWARD steganography algorithm [6] to generate 10,000X5=50,000 dense images with an embedding rate of payload= 0.1,0.2,0.3,0.4,0.5 (bpAC), respectively.
(3) The carrier image and the secret image were extracted with 8,000-dimensional steganographic features using the DCTR extraction algorithm, resulting in a steganographic feature set of 10,000× (1+5) =60,000 images in total. The specific subject settings are shown in table 2.
Table 2 experimental subject setup
The Fisher linear discriminant integrated classifier is used for training and testing the carrier image features and the selected steganographic detection features, and is widely used for training and detecting steganographic analysis feature selection. Firstly, randomly selecting one half of carrier image features and carrier image features corresponding to different embedding rates from each group of feature image sets as a training set; and then taking the rest carrier image characteristics and the carrier image characteristics corresponding to different embedding rates as a test set. The error rate in the integrated classifier is:
wherein P is FA Representing the false alarm rate, P MD Indicating the omission factor, N TS The number of test sets is represented because the test set contains a carrier image set and a dense image set, i.e., N TS =2. The error rate represents the proportion of the number of classification errors to the total test feature component. The lower the detection error rate is, the better the effect of the selected characteristic on detecting the secret image is. In order to more intuitively represent the quality of the comparison experiment result, we use the formula:and converting the detection error rate obtained by the classifier into detection accuracy. />Representing the average detection accuracy,/->The larger the feature, the better the effect of the selected feature on detecting the dense image.
2. Selection experiment
8000-dimensional DCTR image steganography detection features are used herein. The feature is the first order statistic of quantization noise residuals obtained from decompressed JPEG images using 64 discrete cosine transform kernels. DCTR features have lower dimensionality, computational complexity, and higher detection performance than other rich models.
In the S-FUND method, the steganographic detection characteristic component which is measured by two criteria and is different from the threshold T is deleted, and in order to enable the S-FUND method to have a better selection effect, the threshold T is analyzed. Firstly, in order to effectively reduce the feature dimension, we consider that the difference measured by two criteria is larger than 15% of the original feature dimension, and delete the steganographically detected feature components with the difference larger than 8,000x15% =1,200 dimension for the DCTR feature of 8,000 dimension; then, the value of the threshold T is gradually reduced, so that the threshold T epsilon {0.15,0.14, …,0.02,0.01}. Experiments were performed separately, comparing the feature dimensions and detection accuracy of each group, and the experimental results are shown in table 3:
TABLE 3 comparison experiment results before and after selecting DCTR characteristics based on S-FUND method
In table 3, dim represents the feature dimension,indicating the accuracy of the detection. As can be seen from Table 3, the S-FUND method can greatly reduce the feature dimension while maintaining or even improving the detection accuracy at different embedding rates. For example: when payload=0.1, the detection precision of the selected feature based on the S-FUND method on the loaded image can reach 0.5270, which is improved by 0.31% compared with the original detection precision, and the selected feature dimension is reduced by 3462 dimension compared with the original feature dimension; and, when t=0.04, the feature selected based on the S-FUND method is only 30.44% of the original feature dimension while maintaining the detection accuracy of the dense image. When payload= 0.2,0.3, the features selected based on the S-FUND method can reduce the DCTR feature dimension to different degrees, and the detection precision is improved by 0.49% and 0.16% compared with the original detection precision; and the feature selected based on the S-FUND method keeps the detection precision of the dense image, and the feature dimension is only 29.79% and 51.79% of the original feature dimension, so that the space-time overhead of classifier training is reduced.
In order to more intuitively compare the selection of DCTR steganography detection features by the S-FUND method, feature dimensions and detection accuracy before and after the selection are shown in fig. 3.
In fig. 3, the horizontal axis represents the threshold value, the vertical axis represents the corresponding feature dimension and detection accuracy, and five lines from top to bottom represent the effects of the DCTR features selected under five different embedding rates in sequence, and the point of optimal performance under each embedding rate is processed and marked with a numerical value. The S-FUND method provided herein can maintain or even improve the detection precision of DCTR features on the dense image, and simultaneously greatly reduce the feature dimension, thereby proving the effectiveness of the S-FUND method.
A large number of experiments are carried out, and the DCTR features selected by the S-FUND method are proved to greatly reduce feature dimensions while the detection precision of the dense image is maintained or even improved. Comparing experiments with a Random-D method, a CGSM method and a PCA-D method shows that the detection precision of the features selected based on the S-FUND method on the loaded image is higher. For example: in a comparison experiment with the Random-D method, when payload=0.5, the detection accuracy of the features selected by the S-FUND method on the loaded image is up to 1.81% compared with the features selected by the Random-D method; in comparison experiments with the CGSM method, the detection precision of the features selected by the S-FUND method on the loaded image is up to 2.25% higher than that of the CGSM method; in the contrast experiment with the PCA-D method, the detection precision of the features selected by the S-FUND method on the loaded image is up to 4.25% higher than that of the PCA-D method.
The embodiment also provides a steganography detection feature selection device based on the distortion degree and the information gain ratio, which comprises the following steps:
a measurement module for measuring a difference between the carrier image and the secret image for each steganographically detected feature component using the distortion degree and the information gain ratio;
the descending order arrangement module is used for descending order of the distortion value and the information gain ratio of each steganography detection characteristic component;
the deleting module is used for deleting the steganography detection characteristic components with larger difference of arrangement sequences according to the two arranged column numbers;
and the training module is used for training and detecting the reserved steganography detection characteristic component serving as a finally selected steganography detection characteristic.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus.
Finally, it should be noted that: the foregoing description is only illustrative of the preferred embodiments of the present invention, and is not intended to limit the scope of the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention are included in the protection scope of the present invention.
Claims (4)
1. A steganography detection feature selection method based on distortion and information gain ratio is characterized by comprising the following steps:
step 1, measuring the difference of each steganography detection characteristic component between a carrier image and a secret image by using distortion degree and information gain ratio;
using the distortion degree to measure the difference of each steganography detection characteristic component between the carrier image and the carrier image, wherein the carrier image generates distortion with different degrees due to the embedding of information, and using the formula (1) to measure the distortion degree K of the carrier image before and after steganography i The formula is as follows:
wherein f i C And f i S Representing the values of the ith steganographic detection feature component in the carrier image and the carrier image respectively,and->Respectively represent the ith steganography detection bitsThe value of the signature component in the j-th carrier image and the secret image; k (K) i The larger the value is, the larger the distortion degree generated by the carrier image after the information is embedded is, and the larger the difference between the carrier image and the carrier image is, the more the feature component is favorable for detecting the carrier image, and the more the feature component is reserved;
the information gain ratio is defined as the ratio of the information gain value of the feature component between the carrier image and the carrier image to the partial entropy of the median value of the feature component in the carrier image with respect to the median value of the feature component in the carrier image, and the information gain ratio g is measured using the formula (2) R (f i S ,f i C ) The formula is as follows:
wherein g (f) i S ,f i C ) An information gain value representing the characteristic component between the carrier image and the secret image,representing partial entropy of the median of the feature component in the carrier image with respect to the median of the feature component in the carrier image; g R (f i S ,f i C ) The larger the value, the larger the difference between the carrier image and the secret image is, the more favorable the feature component is for detecting the secret image, and the more should be preserved;
the formula of the information gain value of the characteristic component between the carrier image and the secret image is as follows:
g(f i S ,f i C )＝H(f i S )-H(f i S |f i C ) (4)
H(f i S |f i C )＝H(f i S ,f i C )-H(f i S ) (5)
wherein H (f) i C ) And H (f) i S ) The information entropy values of the characteristic components in the carrier image and the secret image are respectively represented,H(f i S |f i C ) Conditional entropy value of feature component in carrier image median under the condition of representing feature component in carrier image median, H (f) i S ,f i C ) Representing the joint entropy of the feature components between the carrier image and the secret image;
step 2, respectively arranging distortion degree values and information gain ratio values of each steganographic detection characteristic component in a descending order;
step 3, deleting the steganographic detection characteristic components with larger difference of arrangement sequences according to the two arranged column numbers;
and step 4, training and detecting the reserved steganography detection characteristic component as a finally selected steganography detection characteristic.
2. The steganographic detection feature selection method based on the distortion degree and the information gain ratio according to claim 1, wherein the distortion degree value and the information gain ratio for each steganographic detection feature component are arranged in descending order, respectively, and includes:
according to K i The steganographic detection feature components are arranged in descending order of values;
then according to g R (f i S ,f i C ) The steganographic detection feature components are arranged in descending order of value.
3. The steganographic detection feature selection method based on the distortion degree and the information gain ratio according to claim 2, wherein the deletion of steganographic detection feature components with large differences in arrangement order according to two arranged column numbers includes:
according to the two sorting results, calculating the difference of the ith characteristic component according to two criteria;
the feature components whose absolute values of the arrangement order differences are larger than the threshold T are deleted.
4. An apparatus for implementing a steganographic detection feature selection method of a distortion degree and an information gain ratio as set forth in any one of claims 1 to 3, comprising:
a measurement module for measuring a difference between the carrier image and the secret image for each steganographically detected feature component using the distortion degree and the information gain ratio;
the descending order arrangement module is used for descending order of the distortion value and the information gain ratio of each steganography detection characteristic component;
the deleting module is used for deleting the steganography detection characteristic components with larger difference of arrangement sequences according to the two arranged column numbers;
and the training module is used for training and detecting the reserved steganography detection characteristic component serving as a finally selected steganography detection characteristic.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111213537.2A CN114051082B (en) | 2021-10-19 | 2021-10-19 | Steganography detection feature selection method and device based on distortion degree and information gain ratio |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111213537.2A CN114051082B (en) | 2021-10-19 | 2021-10-19 | Steganography detection feature selection method and device based on distortion degree and information gain ratio |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114051082A CN114051082A (en) | 2022-02-15 |
| CN114051082B true CN114051082B (en) | 2023-10-27 |
Family
ID=80205652
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111213537.2A Active CN114051082B (en) | 2021-10-19 | 2021-10-19 | Steganography detection feature selection method and device based on distortion degree and information gain ratio |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114051082B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114627340B (en) * | 2022-03-19 | 2024-04-30 | 河南师范大学 | Image steganography detection feature self-adaptive selection method based on triple measurement |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110334805A (en) * | 2019-05-05 | 2019-10-15 | 中山大学 | A kind of JPEG domain image latent writing method and system based on generation confrontation network |
| CN112950445A (en) * | 2021-03-11 | 2021-06-11 | 河南师范大学 | Compensation-based detection feature selection method in image steganalysis |
| WO2021155832A1 (en) * | 2020-02-07 | 2021-08-12 | 华为技术有限公司 | Image processing method and related device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8379940B2 (en) * | 2009-06-02 | 2013-02-19 | George Mason Intellectual Properties, Inc. | Robust human authentication using holistic anthropometric and appearance-based features and boosting |
| US8958546B2 (en) * | 2009-12-11 | 2015-02-17 | Stegosytems, Inc. | Steganographic messaging system using code invariants |
-
2021
- 2021-10-19 CN CN202111213537.2A patent/CN114051082B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110334805A (en) * | 2019-05-05 | 2019-10-15 | 中山大学 | A kind of JPEG domain image latent writing method and system based on generation confrontation network |
| WO2021155832A1 (en) * | 2020-02-07 | 2021-08-12 | 华为技术有限公司 | Image processing method and related device |
| CN112950445A (en) * | 2021-03-11 | 2021-06-11 | 河南师范大学 | Compensation-based detection feature selection method in image steganalysis |
Non-Patent Citations (4)
| Title |
|---|
| A Multi-Scale Feature Selection Method for Steganalytic Feature GFR;Xinquan Yu;《IEEE ACCESS》;全文 * |
| Comprehensive Criteria-Based Generalized Steganalysis Feature Selection Method;Yihao Wang;《 IEEE Access》;全文 * |
| 基于W2ID准则的Rich Model隐写检测特征选取方法;马媛媛等;《计算机学报》;全文 * |
| 高容量可逆信息隐藏模型的研究与实现;赵凯峰;《中国优秀硕士学位论文全文数据库》;全文 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114051082A (en) | 2022-02-15 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108961137B (en) | Image steganalysis method and system based on convolutional neural network | |
| CN111652290B (en) | Method and device for detecting countermeasure sample | |
| CN108717680B (en) | Airspace image steganalysis method based on full-dense connection network | |
| CN111460446B (en) | Malicious file detection method and device based on model | |
| CN104661037B (en) | The detection method and system that compression image quantization table is distorted | |
| Ali et al. | A review of digital forensics methods for JPEG file carving | |
| CN110968845B (en) | Detection method for LSB steganography based on convolutional neural network generation | |
| CN112950445B (en) | Compensation-based detection feature selection method in image steganalysis | |
| Muralidharan et al. | The infinite race between steganography and steganalysis in images | |
| CN114051082B (en) | Steganography detection feature selection method and device based on distortion degree and information gain ratio | |
| CN107808100B (en) | Steganalysis method for specific test sample | |
| CN110532772B (en) | File detection method, model, device and computer readable storage medium | |
| Zhang et al. | Coverless video steganography based on audio and frame features | |
| CN115038091A (en) | Method and system for sensing wireless communication frequency spectrum on arctic sea | |
| Biswas et al. | Coverless image steganography based on DWT approximation and pixel intensity averaging | |
| CN108733843B (en) | File detection method based on Hash algorithm and sample Hash library generation method | |
| CN112907431B (en) | Steganalysis method for robust countersteganalysis | |
| Qin et al. | Robustness enhancement against adversarial steganography via steganalyzer outputs | |
| CN108305207B (en) | Airspace image steganalysis credibility evaluation method | |
| Cho et al. | Block-based image steganalysis for a multi-classifier | |
| Xia et al. | A learning-based steganalytic method against LSB matching steganography | |
| Zeng et al. | Towards Secure and Robust Steganography for Black-box Generated Images | |
| Steinebach et al. | The need for steganalysis in image distribution channels | |
| CN112597498A (en) | Webshell detection method, system and device and readable storage medium | |
| Luo et al. | On F5 steganography in images |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |