CN106056639B - It is a kind of based on the anti-hidden decryption method collected evidence of camera source - Google Patents
It is a kind of based on the anti-hidden decryption method collected evidence of camera source Download PDFInfo
- Publication number
- CN106056639B CN106056639B CN201610346457.7A CN201610346457A CN106056639B CN 106056639 B CN106056639 B CN 106056639B CN 201610346457 A CN201610346457 A CN 201610346457A CN 106056639 B CN106056639 B CN 106056639B
- Authority
- CN
- China
- Prior art keywords
- image
- camera
- source
- fingerprint
- hidden
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T9/00—Image coding
Landscapes
- Engineering & Computer Science (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Image Processing (AREA)
Abstract
The present invention is a kind of to belong to Signal and Information Processing technical field based on the anti-hidden decryption method collected evidence of camera source, is related to the hidden secret skill art of source anti-forensics technology and jpeg image based on mode sensor noise.This method is the hidden decryption method for combining the anti-algorithm of collecting evidence of camera source with hidden close algorithm, utilize steganalysis technology feature lower to hidden close image detection rate in the case where camera model mismatch, first the source of image is forged, it is obtained using the fingerprint replacement algorithm in the anti-evidence collecting method in source based on mode sensor noise and forges image, finally, with the hidden close algorithm of nsF5 to image encryption, hidden close image is obtained.The method increase the safeties of hidden close image, are suitable for information security field, transmit secret information with effectively helping secret information transmitting personnel safety;To which the important confidential information transmitted in a network for politics, military, business etc. provides technical support.
Description
Technical field
The invention belongs to Signal and Information Processing technical field, it is related to that the source based on mode sensor noise is counter to collect evidence
The hidden secret skill art of technology and jpeg image.
Background technique
The arrival of information age so that information security becomes a worldwide problem, information security be national information into
The rapid development of the basic guarantee of journey and the information industry development, internet brings great convenience to our life, still
The wildness of hacker and computer virus makes the individual privacy of netizens probably will receive infringement, and politics, military, business etc. exist
The important confidential information transmitted in network may be intercepted and captured or be changed by criminal in transmittance process, bring to country huge
Big loss.The hidden secret skill art that Content of Communication safety can be realized by digital media, has attracted a large number of researchers in recent years
Concern.Due to digital picture acquisition facilitate it is from a wealth of sources, jpeg image again because its is small in size and facilitate storage and transmission the advantages that
Become the mainstream format of image, therefore the hidden secret skill art based on jpeg image has very big carrier selection space, stands hidden close
The position of analysis side is intended to find hidden close image in large nuber of images to be an extremely difficult thing, these advantages of jpeg image
So that its favor by hidden secret skill art researcher.
Have multinomial research achievement for the hidden close algorithm of jpeg image at present to deliver, classical hidden close algorithm is to utilize
The least significant bit of DCT coefficient and the method hiding information that image is modeled, due to having the embedded mobile GIS of hidden secret skill art
Very efficiently, hidden close algorithm main direction of studying at this stage has been converted to the calculating of hidden close distortion equation.Fridrich etc.
People article " Statistically undetectable jpeg steganography:dead ends challenges,
And opportunities ", Proceedings of the 9th workshop on Multimedia&security,
ACM proposes nsF5 algorithm in 2007, P3-14., and which employs l Water Paper coding, l Water Paper coding is divided into the data to be encoded
" dry " and " wet " two parts." wet " can only partially be used to recipient and read information, and sender cannot modify number therein
According to;" dry " is partially that normally, can not only modify, but also can read, this coding techniques ensure that the statistics of DCT histogram
Characteristic is constant, and the contraction of F5 algorithm is efficiently solved the problems, such as using l Water Paper coding.Phil Salle is in article " Model-
Based steganography ", Digital watermarking.Springer Berlin Heidelberg, 2003,
P154-167. it in, proposes MB algorithm, is a kind of Information hiding frame based on model, main thought is to scheme a width JPEG
Part and non-determined part are determined as being divided into, and are guaranteed when embedding information to determine that part is constant, are only changed non-determined part, by estimating
Model profile is counted, the position of embedding information is determined, but the hidden close image of MB algorithm is applied to will appear block effect, meeting in airspace
Easily it detected by corresponding concealed analysis method.The Patent No. CN201310275158.5 of Yu Nenghai et al., it is " a kind of
In the JPEG image steganography method of high security ", corresponding longitudinal mistake is defined using the quantization DCT coefficient xi and Xi that are rounded front and back
Very with lateral distortion, and using quantization round-off error building longitudinal distortion minimum be embedded in distortion model;According to longitudinal distortion
Minimum be embedded in distortion model, and the optimal modification of corresponding longitudinal distortion and the lateral each quantization DCT coefficient of distortion computation
Probability;The optimal modification probability of each quantization DCT coefficient of acquisition is converted into ternary distortion, and calls verification grid code STC,
It is embedded in classified information using the quantization DCT coefficient Xi after rounding as carrier, then is packaged into complete jpeg image, ensure that insertion
Good picture quality after the safety and insertion classified information of classified information, although the method guarantees image fault most as far as possible
Smallization, but the feature that uses of existing steganalysis method has reached dimensions up to ten thousand, insertion distortion model be difficult to it is exhaustive, very
It is easy to attend to one thing and lose sight of another.
Summary of the invention
The present invention in view of the drawbacks of the prior art, invents a kind of based on the anti-hidden decryption method collected evidence of camera source, hidden close side
Camera source is instead collected evidence and is combined with hidden close algorithm by method, under camera model match condition to the hidden close correct decision rate of image compared with
High problem proposes to reduce the correct decision rate of steganalysis, improves the safety of hidden close image.The present invention is suitable for information
Secret information is transmitted with effectively helping secret information transmitting personnel safety in security fields.
The technical scheme is that a kind of based on the anti-hidden decryption method collected evidence of camera source, characterized in that this method is
The hidden decryption method that the anti-algorithm of collecting evidence of camera source is combined with hidden close algorithm, using steganalysis technology in camera model mismatch
In the case where the feature lower to hidden close image detection rate, first the source of image is forged, using being based on mode sensor
Fingerprint replacement algorithm in the anti-evidence collecting method in the source of noise, which obtains, forges image, then with the hidden close algorithm of nsF5 to image encryption,
Obtain hidden close image;Detailed process is as follows for method:
First in camera source is instead collected evidence, if the fingerprint of source camera A is KA, the fingerprint of purpose camera B is KB, image
IAFrom source camera A, by image IAIn belong to the fingerprint of source camera A and be substituted for the fingerprint of purpose camera B, replaced
Forgery image I after fingerprintA', replacement formula are as follows:
IA′=IA-αKA+βKB (1)
Wherein, α is the fingerprint intensity of the source camera A subtracted, β be plus purpose camera B fingerprint intensity, 0≤α≤
1,0≤β≤1, the size of value depend on fingerprint replacement number;
Two variable linear regression P is constructed first
P=x0(x1+x2β)α+x3(x4+x5β) (2)
Wherein x0,x1,x2,x3,x4,x5For parameter.
Have after (2) formula is dismantled arrangement:
P=x3x4+x0x1α+x3x5β+x0x2αβ (3)
For convenience, by simplification of coefficient, then have:
Wherein, a0,a1,a2,a3And b0,b1,b2,b3For simplified coefficient, PtrueTo forge image and source camera A
Correlation, PfakeFor the correlation for forging image and purpose camera B;
P=[corr1 corr2 ... corrn]T (5)
Wherein, corr1,corr2,...,corrnImage and camera are forged when to select different fingerprints to replace intensity α and β
Correlation.
Y=[a1 a2 a3 a4]T (7)
Write as matrix form P=XY, X is sequency spectrum matrix, and Y is coefficient matrix, then least square method calculates the formula of Y
Are as follows: Y=(XTX)-1XTP (8)
It asks fingerprint to replace intensity α and β Y substitution formula (4), and enables PtrueIt is set as 0, by PfakeWhen being set as not altered
Correlation size between image and former camera;
If calculating fingerprint replacement intensity α and β is undesirable or not in 0≤α≤1, the zone of reasonableness of 0≤β≤1, is ensuring
PtrueIn the case where=0, by PfakeFair-sized is adjusted, if PfakeIt is excessive, β is turned down, if PfakeIt is too small then by α tune
Greatly, preliminary fingerprint replacement intensity α and β are obtained;
In order to improve the maximum insertion rate for forging image, then need to modify forgery figure under the premise of guaranteeing picture quality
Fingerprint as during replaces intensity α and β, but the anti-evidence obtaining safety for forging image can be reduced by modifying fingerprint replacement intensity α and β
Property, so wanting adjusting fingerprint replacement intensity α, β appropriate and maximum insertion rate σmRelationship between three is guaranteeing picture quality
While, so that forging the maximum insertion rate and anti-evidence obtaining safety maximization of image, its step are as follows:
The first step calculates initial fingerprint with the method for linear fit and replaces intensity α and β, and calculates corresponding PSNR, takes
Judgement threshold t is demonstrate,proved, image sources are judged to if correlation is greater than judgement threshold in respective camera, PSNR threshold value is set, calculates embedding
Enter the initial value of rate σ;
Second step, embedding information calculate the correlation corA (k) with source camera A, the correlation corB with purpose camera B
(k), wherein k indicates the correlation after kth time verification;
Third step, to needing condition to be iterated judgement, pseudocode is as follows:
Export fingerprint replacement intensity α, β and maximum insertion rate σmValue, with obtain fingerprint replacement intensity α and β to image
Source forgery is carried out, then obtains hidden close image with nsF5 algorithm embedding information.
The beneficial effects of the invention are as follows in the confrontation with steganalysis algorithm, match in camera source for hidden close algorithm
In the case where steganalysis to the higher problem of the hidden close correct decision rate of image, propose by camera source anti-forensics technology with it is hidden
The method that secret skill art combines.The present invention improves the safety of hidden close image, to be politics, military, business etc. in network
The important confidential information of middle transmitting provides technical support.
Detailed description of the invention
Fig. 1 instead collect evidence in hidden close algorithm and steganalysis model flow figure.
Fig. 2 a) when being α=0.04 as image and source camera or purpose camera correlation variation diagram are forged in the variation of β,
Fig. 2 b) when being α=0.02 as image and source camera or purpose camera correlation variation diagram are forged in the variation of β, in figure: horizontal seat
It is designated as the value of fingerprint replacement intensity β, ordinate is the correlation for forging image and camera, and solid line indicates change when α is constant with β
Change the correlation for forging image and source camera, dotted line indicates when α is constant as the phase of image and purpose camera is forged in the variation of β
Guan Xing.
Fig. 3 a) when being β=0.04 as image and source camera or purpose camera correlation variation diagram are forged in the variation of α,
Fig. 3 b) when being β=0.02 as image and source camera or purpose camera correlation variation diagram are forged in the variation of α, in figure: horizontal seat
It is designated as the value of fingerprint replacement intensity α, ordinate is the correlation for forging image and camera, and solid line indicates change when β is constant with α
Change the correlation for forging image and source camera, dotted line indicates when β is constant as the phase of image and purpose camera is forged in the variation of α
Guan Xing
Specific embodiment
A specific embodiment of the invention is described in detail below in conjunction with technical solution and attached drawing.
Camera source is instead collected evidence and is combined with hidden close algorithm by the present invention, and the process of method is as shown in Figure 1.
This process is divided into two stages: first stage, and the image that Alice is used when transmitting information to Bob all derives from
Source camera A in Alice hand, caretaker Wendy can also obtain these images during two people transmit information, long and long
Wendy obtain enough images for belonging to source camera A, Matching Model can be established to source camera A, if at this moment
Alice still uses the image of source camera A to transmit information, it is easy to detected by Wendy.After Alice has found such case,
The scheme of mismatch is established for the Matching Model of caretaker, in order not to allow Wendy to throw doubt upon, still with existing model, she
The image for calling in purpose camera B will belong to the image forge in source camera A originally into the image of purpose camera B, establish Wendy
Model failure.
Second stage, when Wendy have found image that Alice is transmitted from it is original different, and learn that Alice possesses B mesh
Camera, then with purpose camera B model, to Alice transmission image carry out steganalysis.Alice will finally realize hidden close
Transmission needs to complete image sources and forges and hidden close, needs to consider two key factors, in order to reach better hidden close effect,
Then to select preferable hidden close algorithm;In order to not only guarantee the quality of image but also achieve the purpose that image sources are forged, then to select
Appropriate fingerprint replaces intensity.
The selection of hidden close algorithm: the hidden close algorithm of the jpeg image of classics several to MB1, F5 and nsF5 carries out performance evaluation
Compared with, mainly in terms of the not sentience of algorithm and hidden capacity two detection algorithm performance.Due to nsF5 algorithm
Not sentience highest and hidden capacity is relatively high, the hidden close algorithm for selecting nsF5 to be embedded in as information.
Need to forge source image insertion secret information in experiment and obtain hidden close image, first have to the source of image into
Row is forged, and the used jpeg image of the present embodiment is as shown in table 1, is respectively derived from 4 cameras of different brands, every camera
500 width images.
The experiment of table 1 uses image
Camera | Image size | Resolution ratio | Amount of images |
CNPRO111 | 1024×768 | 180dpi | 500 |
NKE570023 | 1280×960 | 300dpi | 500 |
SNF82801 | 1280×960 | 72dpi | 500 |
KDDC29021 | 1792×1200 | 72dpi | 500 |
Respectively with the fingerprint of 200 four cameras of width Image estimation, calculates finger image and replace intensity.With source phase in Fig. 1
Machine A represents the camera for forging image, and purpose camera B represents the purpose camera for forging image, and source camera A and purpose camera B can
Think any camera.The present embodiment realizes the forgery of four groups of image sources in table 2, every group of 300 width image of forgery.
2 image sources of table are forged
Source camera A | Purpose camera B | Forge amount of images |
NKE570023 | SNF82801 | 300 |
SNF82801 | NKE570023 | 300 |
KDDC29021 | CNPR0111 | 300 |
CNPRO111 | KDDC29021 | 300 |
Wherein, the calculating of the fingerprint replacement intensity α and β that use when forging image, are first calculated just using formula (2)-(8)
The value of step, the image forged with the replacement intensity, embedding information discovery, with the Y-PSNR of the increase image of insertion rate
PSNR can decline, and insertion rate herein refers to that the percentage that embedding information accounts for image when expiring embedding, some images are very low in insertion rate
When PSNR just fall below 40dB hereinafter, human eye is it can be found that image changes.In order to improve under the premise of guaranteeing picture quality
The maximum insertion rate for forging image then needs to modify fingerprint replacement the intensity α and β forged in image process, but modifies fingerprint
Replacement intensity α and β can reduce the anti-evidence obtaining safety for forging image, so adjustings fingerprint appropriate is wanted to replace intensity α, β and most
Big insertion rate σmRelationship between three, while guaranteeing picture quality, so that the maximum insertion rate for forging image takes with counter
It demonstrate,proves safety to maximize, then preliminary fingerprint replacement intensity α and β is adjusted with the first, second and third previously described step again,
Obtain the maximum insertion rate σ of final fingerprint replacement intensity α and β and imagem, the results are shown in Table 3.
3 image sources of table forge fingerprint and replace intensity
Source camera A | Purpose camera B | Replace intensity α | Replace intensity β | Maximum insertion rate σm |
NKE570023 | SNF82801 | 0.0043 | 0.0151 | 0.86 |
SNF82801 | NKE570023 | 0.0020 | 0.0032 | 0.92 |
KDDC29021 | CNPRO111 | 0.0321 | 0.0410 | 0.90 |
CNPRO111 | KDDC29021 | 0.0094 | 0.0051 | 1.00 |
Source forgery is carried out to image with obtained fingerprint replacement intensity α and β, obtains forging image;Then, figure is being forged
It is embedded in secret information as in, obtains hidden close image.
Overall construction is carried out to hidden close image by the PEV-274 feature and SVM classifier of extracting image in experiment
Detection, verifies the validity of this method, and overall construction detects mistaken verdict rate PEIt is omission factor PFP, it is by hidden close image misjudgement
Carrier image and false alarm rate PFN, FNR misjudges carrier image for hidden close image, the average value of the two, it may be assumed that
Verification process first stage, the model of source camera A known to Wendy.
In the case where steganalysis personnel Wendy learns camera A and obtains its classifier, to the original graph of source camera
Picture and the overall construction for forging image are analyzed, as a result as shown in table 4, table 5.It can be seen that not in the comparison of two tables
All have significant improvement compared with original image with the overall construction of the forgery image of camera.
Original graph overall construction in the case of 4 known models A of table
Image library | PFP(%) | PFN(%) | PE(%) |
CNPRO111 | 3.0 | 4.0 | 3.5 |
NKE570023 | 15.0 | 13.0 | 14.0 |
SNF82810 | 2.0 | 12.0 | 7.0 |
KDDC29021 | 1.0 | 1.0 | 2.0 |
Figure overall construction is forged in the case of 5 known models A of table
Source camera A | Purpose camera B | PFP(%) | PNP(%) | PE(%) |
CNPRO111 | KDDC29021 | 0.0 | 100.0 | 50.0 |
NKE570023 | SNF82801 | 18.0 | 78.0 | 43.0 |
SNF82801 | NKE570023 | 10.0 | 64.0 | 37.0 |
KDDC29021 | CNPRO111 | 4.0 | 72.0 | 38.0 |
Overall construction analysis under different insertion rates, is CNPRO111, purpose camera using source camera
The forgery image of KDDC29021 is compared with the original image of camera CNPRO111, as shown in table 6, table 7.Two tables compare can
To find out, under different insertion rates, the overall construction for forging image is improved largely compared to the overall construction of original image.
The overall construction of original graph difference insertion rate in the case of 6 known models A of table
The overall construction for scheming different insertion rates is forged in the case of 7 known models A of table
Insertion rate | 0.05 (%) | 0.1 (%) | 0.2 (%) | 0.3 (%) | 0.4 (%) | 0.5 (%) |
PFP | 0.0 | 0.0 | 0.0 | 26.0 | 26.0 | 26.0 |
PFN | 100.0 | 100.0 | 100.0 | 74.0 | 70.0 | 68.0 |
PE | 50.0 | 50.0 | 50.0 | 50.0 | 48.0 | 47.0 |
Verification process second stage, the model of purpose camera B known to Wendy.
In the case where steganalysis personnel Wendy learns purpose camera B and obtains its classifier, to the non-of purpose camera
The overall construction for forging image and forgery image is analyzed, and the results are shown in Table 8.With the genuine image phase of purpose camera
Have than, the false detection rate for forging image and wants when big raising, at most improves 46.5%, it is minimum to improve 33%.
The overall construction of image is forged in the case of 8 known models B of table
Purpose camera | PFP(%) | PFN(%) | PE(%) | Genuine image PE(%) |
CNPRO111 | 0.0 | 100.0 | 50.0 | 3.5 |
NKE570023 | 32.0 | 62.0 | 47.0 | 14.0 |
SNF82810 | 10.0 | 80.0 | 45.0 | 7.0 |
KDDC29021 | 18.0 | 70.0 | 44.0 | 2.0 |
Overall construction analysis under different insertion rates, is CNPRO111, purpose camera using source camera
The forgery image of KDDC29021 is compared with the genuine image of camera KDDC29021, as shown in table 9, table 10.Two table ratios
Relatively as can be seen that under different insertion rates, the overall construction for forging image is pacified compared to the statistics of the genuine image of purpose camera
Full property is improved largely.
The overall construction of genuine figure difference insertion rate in the case of 9 known models B of table
Insertion rate | 0.05 (%) | 0.1 (%) | 0.2 (%) | 0.3 (%) | 0.4 (%) | 0.5 (%) |
PFP | 24.0 | 16.0 | 2.0 | 0.0 | 1.0 | 0.0 |
PNP | 23.0 | 8.0 | 0.0 | 1.0 | 1.0 | 1.0 |
PE | 23.5 | 12.0 | 1.0 | 0.5 | 0.5 | 0.5 |
The overall construction schemed under different insertion rates is forged in the case of 10 known models B of table
Insertion rate | 0.05 (%) | 0.1 (%) | 0.2 (%) | 0.3 (%) | 0.4 (%) | 0.5 (%) |
PFP | 32.0 | 30.0 | 18.0 | 28.0 | 30.0 | 28.0 |
PFN | 68.0 | 70.0 | 70.0 | 48.0 | 34.0 | 12.0 |
PE | 50.0 | 50.0 | 44.0 | 38.0 | 32.0 | 20.0 |
By can be seen that either for source camera still under the detection of purpose camera model, the present invention can in above-mentioned table
Enough effectively improve the safety of hidden close image.
Claims (1)
1. a kind of based on the anti-hidden decryption method collected evidence of camera source, characterized in that this method is by the anti-algorithm of collecting evidence of camera source
With the hidden decryption method that hidden close algorithm combines, hidden close image is examined in the case where camera model mismatch using steganalysis technology
The lower feature of survey rate, first forges the source of image, uses the anti-evidence collecting method in source based on mode sensor noise
In fingerprint replacement algorithm obtain forge image, finally, obtaining hidden close image to image encryption with the hidden close algorithm of nsF5;Method
Detailed process is as follows:
First in camera source is instead collected evidence, if the fingerprint of source camera A is KA, the fingerprint of purpose camera B is KB, image IACome
Derived from source camera A, by image IAIn belong to the fingerprint of source camera A and be substituted for the fingerprint of purpose camera B, obtain replacement fingerprint
Forgery image I afterwardsA', replacement formula are as follows:
IA′=IA-αKA+βKB (1)
Wherein, α is the source camera A subtracted fingerprint replaces intensity, β be plus the fingerprint of purpose camera B replace intensity, 0
≤ α≤1,0≤β≤1, the size of value depend on fingerprint replacement number;
Two variable linear regression P is constructed first
P=x0(x1+x2β)α+x3(x4+x5β) (2)
Wherein, x0,x1,x2,x3,x4,x5For parameter;
Have after (2) formula is dismantled arrangement:
P=x3x4+x0x1α+x3x5β+x0x2αβ (3)
For convenience, by simplification of coefficient, then have:
Wherein, a0,a1,a2,a3And b0,b1,b2,b3For simplified coefficient, PtrueFor the phase for forging image and source camera A
Guan Xing, PfakeFor the correlation for forging image and purpose camera B;
P=[corr1 corr2...corrn]T (5)
Wherein, corr1,corr2,...,corrnImage and camera A or B are forged when to select different fingerprints to replace intensity α and β
Correlation;
Y=[a1 a2 a3 a4]T (7)
Write as matrix form P=XY, X is sequency spectrum matrix, and Y is coefficient matrix, then least square method calculates the formula of Y are as follows: Y=
(XTX)-1XTP (8)
It asks fingerprint to replace intensity α and β Y substitution formula (4), and enables PtrueIt is set as 0, by PfakeImage when being set as not altered
IAThe correlation size between the camera A of source;
If calculating fingerprint replacement intensity α and β is undesirable or not in 0≤α≤1, the zone of reasonableness of 0≤β≤1, is ensuring Ptrue=
In the case where 0, by PfakeFair-sized is adjusted, if PfakeIt is excessive, β is turned down, if PfakeIt is too small, α is tuned up, is obtained
To preliminary α and β;
In order to improve the maximum insertion rate for forging image under the premise of guaranteeing picture quality, then needs to modify and forge image mistake
Fingerprint in journey replaces intensity α and β, but the anti-evidence obtaining safety for forging image can be reduced by modifying fingerprint replacement intensity α and β,
So wanting adjusting α, β appropriate and maximum insertion rate σmRelationship between three, while guaranteeing picture quality, so that forging
The maximum insertion rate of image and anti-evidence obtaining safety maximize, and its step are as follows:
The first step calculates initial fingerprint with the method for linear fit and replaces intensity α and β, and calculates corresponding PSNR, and evidence obtaining is sentenced
Other thresholding t is judged to image sources in respective camera if correlation is greater than judgement threshold, sets PSNR threshold value, calculate insertion rate σ
Initial value;
Second step, embedding information calculate the correlation corA (k) with source camera A, the correlation corB (k) with purpose camera B,
Wherein, k indicates the correlation after kth time verification;
Third step, to needing condition to be iterated judgement, pseudocode is as follows:
S1, under conditions of PSNR threshold value is greater than threshold value and corB (k) and is greater than evidence obtaining judgement threshold t, if corA (k) is less than
T- θ, θ are to adjust constant,
α=α-ε is then executed, σ=σ+γ, ε and γ are to adjust constant;
S2, under conditions of PSNR threshold value is greater than threshold value and corB (k) and is greater than evidence obtaining judgement threshold t,
If corA (k) is greater than t- θ, corA (k) is less than t, and corB (k) is greater than t+ θ,
β=β-δ is then executed, σ=σ+γ, σ are to adjust constant;
If corA (k) is greater than t- θ, corA (k) is less than t, and corB (k) is less than or equal to t+ θ,
β=β-δ is then executed, second step is returned;
S3, under conditions of PSNR threshold value is greater than threshold value and corB (k) and is greater than evidence obtaining judgement threshold t, if corA (k) is greater than
T,
α=α+ε is then executed, second step is returned;
S5, it is less than or equal under conditions of threshold value or corB (k) be less than or equal to evidence obtaining judgement threshold t in PSNR threshold value,
σ=σ-γ is then executed, second step is returned;
S6, when meeting condition αk-αk-1≤10-4Or βk-βk-1≤10-4When, output fingerprint replaces intensity α, β and maximum insertion rate
σmValue, source forgery carried out to image with obtained fingerprint replacement intensity α and β, then hidden with the acquisition of nsF5 algorithm embedding information
Close image.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610346457.7A CN106056639B (en) | 2016-05-24 | 2016-05-24 | It is a kind of based on the anti-hidden decryption method collected evidence of camera source |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201610346457.7A CN106056639B (en) | 2016-05-24 | 2016-05-24 | It is a kind of based on the anti-hidden decryption method collected evidence of camera source |
Publications (2)
Publication Number | Publication Date |
---|---|
CN106056639A CN106056639A (en) | 2016-10-26 |
CN106056639B true CN106056639B (en) | 2019-05-14 |
Family
ID=57174204
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201610346457.7A Active CN106056639B (en) | 2016-05-24 | 2016-05-24 | It is a kind of based on the anti-hidden decryption method collected evidence of camera source |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN106056639B (en) |
Families Citing this family (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110659679B (en) * | 2019-09-16 | 2022-02-11 | 大连理工大学 | Image source identification method based on adaptive filtering and coupling coding |
CN111178166B (en) * | 2019-12-12 | 2023-05-26 | 中国科学院深圳先进技术研究院 | Camera source identification method based on image content self-adaption |
Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103903271A (en) * | 2014-04-11 | 2014-07-02 | 北京航空航天大学 | Image forensics method for natural image and compressed and tampered image based on DWT |
-
2016
- 2016-05-24 CN CN201610346457.7A patent/CN106056639B/en active Active
Patent Citations (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103903271A (en) * | 2014-04-11 | 2014-07-02 | 北京航空航天大学 | Image forensics method for natural image and compressed and tampered image based on DWT |
Non-Patent Citations (3)
Title |
---|
Anti-Forensics with Steganographic Data Embedding in Digital Images;Hung-Min Sun 等;《IEEE Journal on Selected Areas in Communications》;20110831;第29卷(第7期);全文 |
基于DCT系数大小关系的自适应JPEG隐写;程森 等;《武汉大学学报(理学版)》;20121231;第58卷(第6期);摘要,引言 |
数字图像的反取证技术研究;杨弘;《中国优秀硕士学位论文全文数据库 信息科技辑》;20150315(第03期);摘要,正文第3章 |
Also Published As
Publication number | Publication date |
---|---|
CN106056639A (en) | 2016-10-26 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Li et al. | Steganalysis over large-scale social networks with high-order joint features and clustering ensembles | |
Cho et al. | Block-based image steganalysis: Algorithm and performance evaluation | |
Schmid et al. | Performance analysis of iris-based identification system at the matching score level | |
CN108154080B (en) | Method for quickly tracing to source of video equipment | |
Zhang et al. | Face anti-spoofing detection based on DWT-LBP-DCT features | |
CN101835040B (en) | Digital video source evidence forensics method | |
Rehman et al. | Deep learning for face anti-spoofing: An end-to-end approach | |
Vishi et al. | Multimodal biometric authentication using fingerprint and iris recognition in identity management | |
CN104636764A (en) | Image steganography analysis method and device | |
CN107122725B (en) | Face recognition method and system based on joint sparse discriminant analysis | |
Huang et al. | Exploring feature coupling and model coupling for image source identification | |
CN106056639B (en) | It is a kind of based on the anti-hidden decryption method collected evidence of camera source | |
Qiao et al. | Identifying individual camera device from raw images | |
Long et al. | Detection of Face Morphing Attacks Based on Patch‐Level Features and Lightweight Networks | |
CN110490053B (en) | Human face attribute identification method based on trinocular camera depth estimation | |
Zeng et al. | A framework of camera source identification Bayesian game | |
Wu | Unsupervised steganographer identification via clustering and outlier detection | |
US11935331B2 (en) | Methods and systems for real-time electronic verification of content with varying features in data-sparse computer environments | |
Kumar et al. | Steganography based on payload transformation | |
Liu et al. | Anti‐noise image source identification | |
Jia et al. | Enhanced face morphing attack detection using error-level analysis and efficient selective kernel network | |
Liu et al. | Splicing forgery localization via noise fingerprint incorporated with CFA configuration | |
CN110598043A (en) | Video processing method and device, computer equipment and storage medium | |
Fernandez et al. | Biometric watermarking: an application-based review | |
Hertina et al. | Verifying the authenticity of digital certificate and transcript using background subtraction method |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
C06 | Publication | ||
PB01 | Publication | ||
C10 | Entry into substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |