CN108305207A - A kind of spatial domain picture steganalysis credibility evaluation method - Google Patents
A kind of spatial domain picture steganalysis credibility evaluation method Download PDFInfo
- Publication number
- CN108305207A CN108305207A CN201810037216.3A CN201810037216A CN108305207A CN 108305207 A CN108305207 A CN 108305207A CN 201810037216 A CN201810037216 A CN 201810037216A CN 108305207 A CN108305207 A CN 108305207A
- Authority
- CN
- China
- Prior art keywords
- image
- matrix
- steganalysis
- spatial domain
- residual
- 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.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T1/00—General purpose image data processing
- G06T1/0021—Image watermarking
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2201/00—General purpose image data processing
- G06T2201/005—Image watermarking
- G06T2201/0065—Extraction of an embedded watermark; Reliable detection
Abstract
The invention discloses a kind of spatial domain picture steganalysis credibility evaluation methods, test image is treated first carries out adaptive residual computations, then rounding and break-in operation are carried out to the residual error of acquisition and counts its level, vertically, the symbiosis probability matrix on the secondary diagonal four direction of main diagonal sum.By being averaging to the addition of four symbiosis probability matrixs, the image further is calculated by the reliability assessment numerical value after Stego-detection using this Mean Matrix.The present invention can make steganalysis, and person distinguishes which sample in testing result has higher or lower certainty factor, to promote the application value realistic of steganalysis.
Description
Technical field
The invention belongs to computerized information concealing technology fields, are related to a kind of steganalysis blind checking method, more particularly to
A kind of spatial domain picture steganalysis credibility evaluation method.
Background technology
With the rapid development of network technology, the communication technology and multimedia signal processing technique, Information hiding is as a kind of
Emerging cryptographic technique has become one new research hotspot of information security field.Steganography is Information Hiding Techniques
One important branch, main research how in disclosed multi-medium data hiding information is to realize covert communications.And it is corresponding
Steganalysis research be attack to Steganography, i.e., how to detect, extract or destroy the secret information hidden.
Spatial domain picture steganography is a kind of technology of secret information embedded on image pixel.It generally utilizes least significant bit
(LSB, Lest Significant Bit), which rewrite, realizes secret information insertion.In order to improve the safety of steganographic algorithm,
Hidden writer often attempts the modification amount for reducing steganographic algorithm to carrier image using various coding techniques, and simultaneous selection is comparatively safe
Embedded location carry out steganography.Steganalysis is then the countermeasure techniques of image latent writing, mainly passes through various mathematical statistical models
The exception that steganography is brought is captured, realizes Stego-detection.Rule of doing the most classical is to obtain image high frequency using high-pass filter
Then residual matrix uses symbiosis probability matrix as steganalysis feature.Steganography operation influences phase for HF noise signal
To larger, steganography operation can destroy the distribution of symbiosis probability to a certain extent.Therefore, even using SVM or random forest
It is that deep learning network can be carried out training and classify, realizes image whether by the detection of steganography.
However, existing steganalysis algorithm comments the testing result of any piece image with same confidence level
Estimate.And there are greatest differences for the masking action of steganography signal for the degree of strength of image inherently high-frequency signal, this makes
Different carriers can correspond to different Detection accuracies.The smaller image detection difficulty of noise is smaller, steganalysis testing result
It is more credible.Conversely, the bigger confidence level of noise is lower.
Invention content
In order to solve the above-mentioned technical problem, the present invention proposes a kind of assessment side of image latent writing analysis detection confidence
Method treats test image and carries out adaptive residual computations first, then carries out rounding and break-in operation to the residual error of acquisition and unites
Count the symbiosis probability matrix on the secondary diagonal four direction of its horizontal, vertical, main diagonal sum.By to four symbiosis probability matrixs
It is added and is averaging, further calculate the image by the reliability assessment numerical value after Stego-detection using this Mean Matrix.
The technical solution adopted in the present invention is：A kind of spatial domain picture steganalysis credibility evaluation method, feature exist
In including the following steps：
Step 1：It treats test image and carries out adaptive residual computations, obtain residual matrix D；
Step 2：Rounding and truncation are carried out to the residual error of acquisition, residual matrix is blocked in acquisition
Step 3：Statistics blocks residual matrixSymbiosis probability on the secondary diagonal four direction of horizontal, vertical, main diagonal sum
Matrix；
Step 4：Calculate the Mean Matrix of four symbiosis probability matrixs；
Step 5：Image to be tested is calculated by the confidence value S after Stego-detection.
The present invention can make steganalysis, and person distinguishes which sample in testing result has higher or lower certainty factor,
To promote the application value realistic of steganalysis.
Description of the drawings
Fig. 1 is that the present invention implements confidence level calculating process flow chart.
Specific implementation mode
Understand for the ease of those of ordinary skill in the art and implement the present invention, with reference to the accompanying drawings and embodiments to this hair
It is bright to be described in further detail, it should be understood that implementation example described herein is merely to illustrate and explain the present invention, not
For limiting the present invention.
The present invention can be used for general steganalysis and the specific verification and measurement ratio of individual specimen estimated, when it is implemented, always
Body process includes two processes of training and detection；Training is to prepare the process of grader (detector), oneself system is understood after training
Make some known sample test accuracy rates.
(1) training process；
Prepare training image and training grader.
It gets out normal picture 5000 and opens open with corresponding hidden image 5000, using being disclosed steganalysis detection side
Method trains detection grader.The steganalysis algorithm that the step uses can voluntarily prepare without particular requirement, user.
Whether grader is used for judgement sample by steganography, and it is close whether detection sample contains, and the grader trained is utilized to help
We judge whether an image has been embedded into secret information.Classifier training in the present invention is mainly that subsequent accuracy rate is surveyed
Have a fling at the experiment under preparation namely laboratory environment.
(2) detection process；
Setup test image simultaneously calculates all test image S numerical value.
Detection process includes (1) (2) (3) three parts, and (1) in addition prepares 5000 normal samples and 5000 steganography samples
For testing, this 10000 sample in total is calculated separately into respective confidence level S values, and according to several sections of S values point.
For purposes of illustration only, this example hypothesis be divided into [0,0.4), [0.4,0.6], (0.6,1] three sections.The image being then located therein
Sample correspond to respectively it is low, in, the method test of high confidence level (2) test is low, in, the corresponding accuracy rate of high confidence level.(3) real
When border is tested, for an image pattern to be tested, the S values of test sample are calculated first, then according to the corresponding section of S values
Accuracy rate is assessed as the test sample accuracy rate.Therefore (1) and (2) is the test job done in the lab, for practical
Advance assessment before detection；(3) be practical application, accuracy rate is totally unknown, need consult (2) test data, and with work
It is assessed for actual accuracy rate.
Referring to Fig.1, confidence level S value calculating process includes the following steps：
Step 1：It treats test image and carries out adaptive residual computations, obtain residual matrix D；
For gray level image to be tested (or Color Channel for coloured image), it is denoted as I=(Ii,j)H×W, 1≤i here
≤ H and 1≤j≤W is respectively the height and width of image I.Adaptive convolutional calculation filtering image I ' through the invention=
(I′i,j)H×W, wherein adaptive convolution is as shown by the equation：
Here ka,bFor the element numerical value in adaptive convolution kernel, calculated according to following formula：
Work as Ii+a,j+bBeyond image I ranges, i.e. i+a=0 or i+a>H or j+a>W or
When j+b=0, I is enabledi+a,j+b=0.Adaptive convolution residual error is subtracted each other by original image and convolved image
It obtains, i.e. D=I-I '=(Di,j)H×W。
Step 2：Rounding and truncation are carried out to the residual error of acquisition, residual matrix is blocked in acquisition
For each element D in residual matrix Di,j, blocked using following formula：
[] is floor operation, and residual matrix D obtains blocking residual matrix after blockingObviously
Element presence -3, -2, -1,0 ,+1 ,+2 ,+3 totally 7 kinds of states.
Step 3：Statistics blocks residual matrixSymbiosis probability on the secondary diagonal four direction of horizontal, vertical, main diagonal sum
Matrix；
For the residual matrix blockedIt is as follows to calculate symbiosis probability matrix formula.
Here Ch(u, v), Cv(u, v), Cd(u, v) and Cm(u, v) indicates to block symbiosis of the residual error on four direction respectively
Probability, wherein u and v are integer, and u, v ∈ [- 3 ,+3] (similarly hereinafter).Due to blocking each element presence -3 of residual error, -2, -1,
0 ,+1 ,+2 ,+3 totally 7 kinds of states, therefore each co-occurrence matrix dimension is 7 × 7 sizes.
Step 4：Calculate the Mean Matrix of four symbiosis probability matrixs；
Obtain in step 3 four symbiosis probability matrixs are added to be averaging and can obtain Mean MatrixIts
Middle u, v ∈ [- 3 ,+3].
Step 5：Image to be tested is calculated by the confidence value S after Stego-detection；
S numerical value is between 0 and 1, and the S numerical value of testing image is bigger, and testing result is more reliable, otherwise detection confidence is got over
It is low.
The problem of present invention is influenced for image to be detected Detection accuracy by image-carrier itself, utilizes image spatial domain
Filter noise is detected the assessment of credible result degree.It obtains Image Residual using adaptive convolution method proposed by the present invention
As the noise signal of image, the statistical distribution data of symbiosis probability extraction noise signal are reused, the probability point is finally utilized
The statistical data of cloth calculates reliability assessment numerical value of the image for steganalysis.It is an advantage of the present invention that for a batch
Image pattern to be detected, can effectively assess check accuracy rate height or distinguishing tests sample set in high detection rate (can
Reliability) sample and low probability of detection (confidence level) sample, the practical manifestation of detection algorithm is promoted, its actual application ability is enhanced.
It should be understood that the part that this specification does not elaborate belongs to the prior art.
It should be understood that the above-mentioned description for preferred embodiment is more detailed, can not therefore be considered to this
The limitation of invention patent protection range, those skilled in the art under the inspiration of the present invention, are not departing from power of the present invention
Profit requires under protected ambit, can also make replacement or deformation, each fall within protection scope of the present invention, this hair
It is bright range is claimed to be determined by the appended claims.
Claims (6)
1. a kind of spatial domain picture steganalysis credibility evaluation method, which is characterized in that include the following steps：
Step 1：It treats test image and carries out adaptive residual computations, obtain residual matrix D；
Step 2：Rounding and truncation are carried out to the residual error of acquisition, residual matrix is blocked in acquisition
Step 3：Statistics blocks residual matrixSymbiosis probability square on the secondary diagonal four direction of horizontal, vertical, main diagonal sum
Battle array；
Step 4：Calculate the Mean Matrix of four symbiosis probability matrixs；
Step 5：Image to be tested is calculated by the confidence value S after Stego-detection.
2. spatial domain picture steganalysis credibility evaluation method according to claim 1, it is characterised in that：In step 1, note
One Color Channel of gray level image or coloured image to be tested is I=(Ii,j)H×W, wherein H, W be respectively image I height and
Width, 1≤i≤H, 1≤j≤W；
Pass through adaptive convolutional calculation filtering image I '=(I 'i,j)H×W, wherein adaptive Convolution Formula is：
Wherein ka,bFor the element numerical value in adaptive convolution kernel, calculated according to following formula：
Work as Ii+a,j+bBeyond image I ranges, i.e. i+a=0 or i+a>H or j+a>W or
When j+b=0, I is enabledi+a,j+b=0；
Residual matrix D subtracts each other acquisition, i.e. D=I-I '=(D by image I and image I 'i,j)H×W。
3. spatial domain picture steganalysis credibility evaluation method according to claim 2, it is characterised in that：It is right in step 2
Each element D in residual matrix Di,j, blocked using following formula：
Wherein, T is predetermined threshold value, and [] is floor operation, and residual matrix D obtains blocking residual matrix after blocking
4. spatial domain picture steganalysis credibility evaluation method according to claim 2, it is characterised in that：It is right in step 3
In blocking residual matrixCalculating symbiosis probability matrix formula is：
Wherein, Ch(u,v)、Cv(u,v)、Cd(u,v)、Cm(u, v) indicates to block symbiosis probability of the residual error on four direction respectively,
Wherein u and v is integer, and u, v ∈ [- T ,+T], T are predetermined threshold value.
5. spatial domain picture steganalysis credibility evaluation method according to claim 4, it is characterised in that：It, will in step 4
The four symbiosis probability matrixs obtained in step 3, which are added, to be averaging, and Mean Matrix is obtained
6. spatial domain picture steganalysis credibility evaluation method according to claim 5, it is characterised in that：In step 5,
Confidence level S numerical value is between 0 and 1, and the S numerical value of testing image is bigger, and detection confidence is higher, otherwise detection confidence
It is lower.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810037216.3A CN108305207B (en) | 2018-01-15 | 2018-01-15 | Airspace image steganalysis credibility evaluation method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810037216.3A CN108305207B (en) | 2018-01-15 | 2018-01-15 | Airspace image steganalysis credibility evaluation method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108305207A true CN108305207A (en) | 2018-07-20 |
| CN108305207B CN108305207B (en) | 2021-07-20 |
Family
ID=62869141
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810037216.3A Active CN108305207B (en) | 2018-01-15 | 2018-01-15 | Airspace image steganalysis credibility evaluation method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108305207B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111127286A (en) * | 2019-11-27 | 2020-05-08 | 杨春芳 | Color image steganography detection method based on differential channel weight distribution |
| CN112199693A (en) * | 2020-09-30 | 2021-01-08 | 东南数字经济发展研究院 | Steganography method based on cartoon image |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103281473A (en) * | 2013-06-09 | 2013-09-04 | 中国科学院自动化研究所 | General video steganalysis method based on video pixel space-time relevance |
| CN103914839A (en) * | 2014-03-27 | 2014-07-09 | 中山大学 | Image stitching and tampering detection method and device based on steganalysis |
| CN106131553A (en) * | 2016-07-04 | 2016-11-16 | 武汉大学 | A kind of video steganalysis method based on motion vector residual error dependency |
| CN106548445A (en) * | 2016-10-20 | 2017-03-29 | 天津大学 | Spatial domain picture general steganalysis method based on content |
-
2018
- 2018-01-15 CN CN201810037216.3A patent/CN108305207B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN103281473A (en) * | 2013-06-09 | 2013-09-04 | 中国科学院自动化研究所 | General video steganalysis method based on video pixel space-time relevance |
| CN103914839A (en) * | 2014-03-27 | 2014-07-09 | 中山大学 | Image stitching and tampering detection method and device based on steganalysis |
| CN106131553A (en) * | 2016-07-04 | 2016-11-16 | 武汉大学 | A kind of video steganalysis method based on motion vector residual error dependency |
| CN106548445A (en) * | 2016-10-20 | 2017-03-29 | 天津大学 | Spatial domain picture general steganalysis method based on content |
Non-Patent Citations (4)
| Title |
|---|
| SONGTAO WU 等: "Steganalysis via Deep Residual Network", 《2016 IEEE 22ND INTERNATIONAL CONFERENCE ON PARALLEL AND DISTRIBUTED SYSTEM》 * |
| 张星星: "面向图像后处理操作类型的盲隐写分析算法可信度评价", 《中国优秀硕士学位论文全文数据库信息科技辑》 * |
| 汪然 等: "基于分割的空域图像隐写分析", 《自动化学报》 * |
| 王丽娜 等: "《信息隐藏技术与应用》", 31 May 2012, 武汉大学出版社 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111127286A (en) * | 2019-11-27 | 2020-05-08 | 杨春芳 | Color image steganography detection method based on differential channel weight distribution |
| CN112199693A (en) * | 2020-09-30 | 2021-01-08 | 东南数字经济发展研究院 | Steganography method based on cartoon image |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108305207B (en) | 2021-07-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105095856B (en) | Face identification method is blocked based on mask | |
| Eklund et al. | Can parametric statistical methods be trusted for fMRI based group studies? | |
| CN107292287A (en) | Face identification method, device, electronic equipment and storage medium | |
| Holmes et al. | Generalized nonlinear modeling with multivariate free-knot regression splines | |
| CN102722858B (en) | Blind steganalysis method based on symmetric neighborhood information | |
| CN113538202A (en) | Image steganography method and system based on generative steganography confrontation | |
| CN111783853B (en) | Interpretability-based method for detecting and recovering neural network confrontation sample | |
| CN108305207A (en) | A kind of spatial domain picture steganalysis credibility evaluation method | |
| CN105374047A (en) | Improved bilateral filtering and clustered SAR based image change detection method | |
| CN107172024B (en) | For in BitTorrent file sharing process based on the covert communications detection method of Have message coding | |
| CN110852364A (en) | Method and device for identifying water source of water burst in mine and electronic equipment | |
| Gunawan et al. | Mobile application detection of road damage using canny algorithm | |
| Haberstroh et al. | Line detection in noisy and structured backgrounds using Græco-Latin squares | |
| CN106022217A (en) | Civil airport runway area detection method free from supervision multistage classification | |
| CN106097236A (en) | Frequency domain robust image reversible water mark method based on Non-negative Matrix Factorization | |
| Yu et al. | Skin detection for adult image identification | |
| CN106203349A (en) | Face identification method based on sparse error matrix | |
| Parekh et al. | Sample size requirement for achieving multisite harmonization using structural brain MRI features | |
| Sarkar et al. | Comparison of various Edge Detection Techniques for maximum data hiding using LSB Algorithm | |
| Ojeda‐Magaña et al. | Pore detection in 3‐D CT soil samples through an improved sub‐segmentation method | |
| He et al. | Multiple Testing of Local Extrema for Detection of Structural Breaks in Piecewise Linear Models | |
| Amerini et al. | Acquisition source identification through a blind image classification | |
| Arbogast et al. | Model‐checking techniques for stratified case‐control studies | |
| Gülay et al. | An improved method to set significance thresholds for β diversity testing in microbial community comparisons | |
| Tomsey et al. | Case work guidelines and interpretation of short tandem repeat complex mixture analysis |
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 |