CN110097124A - Based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation - Google Patents
Based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation Download PDFInfo
- Publication number
- CN110097124A CN110097124A CN201910367646.6A CN201910367646A CN110097124A CN 110097124 A CN110097124 A CN 110097124A CN 201910367646 A CN201910367646 A CN 201910367646A CN 110097124 A CN110097124 A CN 110097124A
- Authority
- CN
- China
- Prior art keywords
- image
- action type
- chain
- treatment effects
- distorted
- 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
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/25—Fusion techniques
- G06F18/254—Fusion techniques of classification results, e.g. of results related to same input data
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management
Landscapes
- Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Theoretical Computer Science (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Biology (AREA)
- Evolutionary Computation (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Image Processing (AREA)
- Editing Of Facsimile Originals (AREA)
- Image Analysis (AREA)
Abstract
The present invention relates to a kind of based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation.The method includes constructing the operation disjunctive model of the digital image manipulation chain based on blind source separating；Estimate the degree of correlation of digital characteristics of image, it is preliminary to identify distorted image action type；According to Dempster-Shafer evidence theory, estimation distorts operation confidence interval, accurately identifies distorted image action type.Compared with prior art, provided by the invention a kind of based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation, scene is distorted towards more actual jpeg image is multiple.Method of the invention is feasible and effective, identification image experience distort action type in terms of can obtain good effect.
Description
Technical field
The present invention relates to image forensics and multi-media safety technical field, and in particular to one kind is divided based on Treatment Effects are obscured
From image operational chain in action type recognition methods.
Background technique
With the prevalence of the image editing softwares such as Photoshop, GIMP, the editor and modification of digital image content are become
It is more and more convenient, and majority distorts pseudo- manufacturing operation all and will not cause the visual suspection of people.If digital picture is by malice
It distorts and wide-scale distribution, serious public trust crisis will be brought to society.Therefore, how combine digital image processing techniques and
Statistical analysis technique carries out the verifying of authenticity and integrity to image, is that image forensics and the great of multi-media safety field are chosen
War.
The statistical property of original digital image can be changed due to distorting forgery behavior, image is passively collected evidence according to number
These statistical properties of word image itself realize the verifying to digital image's authenticity and integrality.Passive forensic technologies have
Good adaptability was increasingly becoming the hot spot of image forensics technical field research in recent years.
However, most of passive forensic technologies of image be all based on image only by the hypothesis that single operation is distorted carry out operation take
Card, or only judge whether image undergoes to distort, it is thus possible to the multiple image forensics result distorted is undergone in limitation
Accuracy.For example, document " Yifang Chen, Xiangui Kang, Z.Jane Wang, and Qiong Zhang.
“Densely connected convolutional neural network for multi-purpose image
forensics under anti-forensic attacks.”ACM Workshop on Information Hiding and
Multimedia Security (IH&MMSEC), 2018. " disclose and a kind of singly distort behaviour based on dense convolutional neural networks
Make evidence collecting method, document " Haodong Li, Weiqi Luo, and Jiwu Huang. " Identification of
various image operations using residual-based features.”IEEE Transactions on
Circuits and Systems for Video Technology (TCSVT), 2018. " disclose one kind based on Image Residual
Characteristic of field singly distorts operation evidence collecting method, document " Marco Fontani, Tiziano Bianchi, Alessia De
Rosa, Alessandro Piva, and Mauro Barni. " A framework for decision Fusion in
image forensics based on dempster–shafer theory of evidence.”IEEE
Transactions on Information Forensics and Security (TIFS), 2013. ", which disclose one kind, is based on
The image true-false method of discrimination of Dempster-Shafer evidence theory.
In view of this, the present invention proposes one kind and is based on obscuring place towards deeper image operational chain evidence obtaining target
Manage effect separation image operational chain in action type recognition methods, with improve distorted in jpeg image operation detection it is accurate
Property.
Summary of the invention
The invention proposes a kind of based on separation method is operated in the image operational chain for obscuring Treatment Effects separation, to know
Include in other operational chain distorts action type, which comprises
(1) the operation disjunctive model of the digital image manipulation chain based on blind source separating is constructed；
(2) according to the degree of correlation of digital picture feature, action type is distorted in preliminary identification；
(3) according to Dempster-Shafer evidence theory, estimation distorts operation confidence interval, accurately identifies and distort operation
Type.
Particular content is as follows:
(1) the operation disjunctive model of the digital image manipulation chain based on blind source separating: the schematic diagram of the model such as Fig. 1 is constructed
It is shown, comprehensively consider the relevance between the operation separation of image operational chain and blind source separating, realizes in digital image manipulation chain
Distort action type identification.
Blind source separating refers to for unknown system, the Independent sources signal of its input is totally unknown or only a small amount of priori
In the case where knowledge, only restore the process for inputting source signal by output signal, that is, mixed signal of system.To digital picture
Operation may be distorted comprising multiple simultaneously when distort forgery, these operations form image operation with certain sequencing
Chain.In described image operational chain, different distorting is independent from each other between operation.Digital picture undergoes the multiple mistake distorted
Journey can be broadly considered as digital picture and multiple noise signals and mix the process being superimposed, to a certain extent image operational chain
Action type identification can also be considered as the separation identification to multiple noise signals of distorted image process addition.Therefore, image is grasped
The operation separation problem and blind source separating problem for making chain are similar.
The method will be considered as source signal in blind source separating problem to the multiple process for distorting operation that original image carries out
Mixed process, it is similar with a series of source signal that separating treatments are estimated is carried out to the mixed signal observed at this time,
When distorting operation separation of image operational chain is carried out, the feature for extracting the digital picture after multiple distort (obscures Treatment Effects
The composite character of generation), only composite character need to be inputted separation system, by separating treatment, can restore singly to distort operation spy
Sign.
It is described to obscure Treatment Effects to refer to that digital picture is subjected to a variety of when distorting processing, what distorted image was left in each operation
Trace, which is overlapped mutually, to be obscured or even subsequent distorts that operation reduction is previous to operate the trace left.
Definition: I={ I1,I2,...,ImIndicate raw image data set, wherein m indicates the number of digital image set
Amount.O={ o1,o2,...,onIndicating the distorted image operational set for including in image operational chain, operation is distorted in wherein n expression
Quantity.F=[f1,f2,...,fm] to concentrate the composite character set extracted from by the multiple digital picture distorted, whereinD indicates characteristic dimension.S=[s1,s2,...,sn] it is single operation characteristic set to be asked, whereinsi
Operation o is distorted in expressioniEstimation feature.There are a hybrid matrixMeet:
F=AS (1)
Separation needs to carry out matrix operation from the composite character that digital picture to be detected is extracted:
WF=WAS=S (2)
WhereinIt is the inverse matrix of hybrid matrix A to solve mixed matrix.Using the fixation based on fourth order cumulant
Point independent composition analysis algorithm, it is initial to dissolve mixed matrix, optimal solution is acquired by iteration and mixes matrix W:
Wherein t indicates the number of iterations,For the i-th row for solving mixed matrix, E [g] is the average statistical for calculating each iteration, |
| g | | it is calculating matrix norm.
Matrix W is mixed using optimal solution, matrixing is made to composite character, and then obtain respectively distorting operation in image operational chain
The approximate evaluation of feature realizes the separation for obscuring image operational chain Treatment Effects, provides direct card for single operation type identification
According to.
(2) according to the degree of correlation of digital picture feature, action type is distorted in preliminary identification: specific flow chart such as Fig. 2 institute
To show, the digital picture feature for comprehensively considering extraction separates degree of correlation between the operating characteristics of acquisition with using disjunctive model,
Realize the preliminary identification to distorted image action type.
Forgery is distorted to m original images, it is O={ o that operational chain included, which distorts operation,1,o2,...,on}.From experience
Composite character F=[f is extracted in the multiple digital picture distorted1,f2,...,fm].Utilize the digitized map based on blind source separating
As the operation disjunctive model of operational chain separates composite character, and then obtain estimation distorts operating characteristics S=[s1,
s2,...,sn], and be regarded as distorting the template characteristic of operation.
Given one experienced the multiple digital picture I' distorted, extract composite character f'.In order to identify that the image is undergone
Distort operation, measure the composite character f' extracted from image I' and the operation template characteristic s for being separated acquisitioniBetween correlation
Property.Degree of correlation is higher, then it represents that image I' may more experienced operation oiDistort forgery.By calculating f' and siPearson came
Related coefficient measures the degree of correlation between them, calculation formula are as follows:
Wherein cov (f', si) indicate to calculate feature f' and siCovariance, D (f') and D (si) indicate calculate feature f' and
siVariance.ρiIt is characterized f' and siRelated coefficient, value range be [- 1,1].Work as ρiValue closer to 1, indicate related journey
It spends higher.To distort operation oiOne threshold tau of corresponding settingi, work as ρi≥τiWhen, indicate that image I' is tampered operation oiDistort puppet
It makes.
(3) according to Dempster-Shafer evidence theory, estimation distorts operation confidence interval, accurately identifies and distort operation
Type: specific flow chart separates as shown in figure 3, being obscured Treatment Effects, primarily determines that may be present in image operational chain usurp
Change operation O={ o1,o2,K,onAfter, further, need to accurately identify these operations.
Joint singly distorts multiple detection algorithms of operation evidence obtaining, excavates different images feature, and study Dempster-
The Decision fusion of the composition rule of Shafer evidence theory differentiates in image operational chain according to fusion results and distorts action type.
The Dempster-Shafer evidence theory belongs to a kind of expansion form of probability theory, by by the trust of evidence
The value up and down of function and probability theory contacts, and explains multivalued mappings according to belief function and likelihood function, to form processing
The evidence theory of uncertain information.The Decision fusion of the composition rule includes melting for multiple operation evidence obtaining algorithm evidence obtaining evidences
It closes.Specific steps include:
That chooses Q kind classics distorts operation o to detectiEvidence obtaining algorithm, constitute Q evidence obtaining classifier, extract respectively
Feature calculates Q and distorts probability { p1,p2,...,pQ, image is without operation o at this timeiThe probability distorted is { 1-p1,
1-p2,...,1-pQ}。
Using the composition rule based on Dempster-Shafer evidence theory, the evidence obtaining result of Q evidence obtaining classifier is done
Integrated decision-making obtains a basic probability assignment m={ m0,m1, wherein m0Indicate image not by operation oiThe probability distorted, m1Table
Diagram picture is by operation oiThe probability distorted.
The belief function Bel and likelihood function Pl for being tampered and being not tampered with two kinds of situations are solved according to m, obtain confidence area
Between [Bel (m0),Pl(m0)] and [Bel (m1),Pl(m1)], wherein Bel (m0) and Bel (m1) respectively indicate not undergo image and usurp
Change operation oiDegree of belief and experience distort operation oiDegree of belief, Pl (m0) and Pl (m1) respectively indicate do not deny image without
It goes through and distorts operation oiDegree of belief and experience distort operation oiDegree of belief.Decision rule specifically:
If meeting decision rule, then it represents that exist in image operational chain and distort operation oi；Conversely, being then not present.
Compared with prior art, above-mentioned technical proposal at least has the advantages that
1, it is provided by the invention it is a kind of based on obscure Treatment Effects separation image operational chain in action type recognition methods,
By the multiple Treatment Effects of obscuring for distorting generation of separate picture, testing image feature is detected with distorting for estimation is separated and operates spy
The degree of correlation between sign, and the size according to the degree of correlation tentatively identify testing image experience distort action type.This method is not
It only experienced dependent on testing image that single operation is distorted it is assumed that solving the primary of more actual jpeg image operational chain evidence obtaining
The identification of problem, i.e. action type.
2, it is provided by the invention it is a kind of based on obscure Treatment Effects separation image operational chain in action type recognition methods,
The Decision fusion of composition rule according to Dempster-Shafer evidence theory to action type is distorted as a result, accurately known
Not.This method can effectively identify jpeg image experience distort operation.
Detailed description of the invention
Fig. 1 is the separation of the present invention " based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation "
Model schematic；
Fig. 2 is distorting for the present invention " based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation "
The preliminary identification process figure of action type；
Fig. 3 is distorting for the present invention " based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation "
Action type accurately identifies flow chart；
Fig. 4 is the present invention " based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation " embodiment
The true picture schematic diagram of middle Lena photo；
Fig. 5 is the present invention " based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation " embodiment
Image schematic diagram is forged in distorting for middle Lena photo.
Specific embodiment
The present invention is based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation.For the ease of saying
Bright, the present embodiment is realized obscure Treatment Effects separation as steps described below so that three are distorted the operational chain of operation composition as an example, but
It is that those skilled in the art should know the technical solution of the application distorts operation suitable for any number of.Specific separating step
Are as follows:
Step 1: the processing history that clear digital picture may be undergone.
For given digital picture I and distort operation set O={ o1,o2,o3, what may be undergone distorts processing history are as follows:
H0: image I, which is not undergone, to be distorted,
H1: image I undergoes o1It distorts,
H2: image I undergoes o2It distorts,
H3: image I undergoes o3It distorts,
H4: image I undergoes o1And o2It distorts,
H5: image I undergoes o1And o3It distorts,
H6: image I undergoes o2And o3It distorts,
H7: image I undergoes o1、o2And o3Distort
Step 2: according to the disjunctive model, Treatment Effects are obscured in separation, obtain the template characteristic for respectively distorting operation.
From M images of image data base A random selection, with operation o1、o2And o3Handle these images.Then, from this M
Composite character, the composite character f of every three images are extracted in imagei、fjAnd fkA composite character set is formed, according to formula
(3) it calculates and solves mixed matrix, then realize the separation to composite character according to formula (1) and (2), obtain respectively representing operation o1、o2With
o3Template characteristic s1、s2And s3.It repeats the step M/3 times, obtains average template featureWith
Below by taking the testing image of model Lena as an example, to distorted image action type recognition methods provided by the invention into
Row compliance test result, specifically:
Step 1: extracting the composite character of testing image.
Fig. 4 illustrates true image, and Fig. 5 illustrates the image after experience is distorted three times, comparison two
Image is only that be difficult to find Fig. 5 be the image after being tampered by naked eyes.In the present embodiment, using Fig. 5 as testing image,
Therefrom extract composite character f', to detect the image experience distort action type.
Step 2: action type in preliminary identification image operational chain.
Action type criterion of identification in the present embodiment are as follows:
WhereinIndicate the image detected experience distorts operation history, ρiIndicate feature f' andRelated coefficient,
τ1、τ2And τ3Respectively correspond operation o1、o2And o3Threshold value.
Composite character f' and template characteristic are calculated according to formula (4)WithRelated coefficient, obtain ρ1、ρ2And ρ3,
It was found that ρ1≥τ1、ρ2≥τ2And ρ3≥τ3, therefore deduce thatImage experienced o1、o2And o3Three kinds of operations are distorted.
Step 3: accurately identifying action type in image operational chain.
According to step 2, primarily determine that the operation of image operational chain is o1、o2And o3.At this point, in order to further bright
Really operation oiPresence hypothesis is given below for testing image I shown in fig. 5:
H0: image I does not undergo oiIt distorts,
H1: image I undergoes oiDistort
Q kind evidence obtaining algorithm is chosen, Q classifier is constituted, whether o is undergone to imageiIt distorts and classifies, obtain classification knot
Fruit { p1,p2,...,pQ}.Evidence obtaining result integrated decision-making based on Dempster-Shafer evidence theory to Q classifier, obtains
Basic probability assignment m={ m0,m1}.Belief function and likelihood function are calculated, according to decision rule shown in formula (5), discovery figure
As experienced operation oiDistort forgery.
In conclusion for more actual image operational chain evidence obtaining scene, the present invention devises one kind and is based on obscuring place
Manage action type recognition methods in the image operational chain of effect separation.Image is obscured Treatment Effects separation mutually to tie with blind source separating
Close, and the composition rule of comprehensive Dempster-Shafer evidence theory Decision fusion as a result, the method have it is feasible
Property, it is able to achieve effective identification to action type in jpeg image operational chain.
It will be understood by those skilled in the art that protection scope of the present invention is not limited to the specific embodiment.?
Under the premise of the principle of the present invention, those skilled in the art can carry out equivalent change or be replaced to the relevant technologies feature
It changes, it should be noted that the technical solution after change or replacement will fall within the scope of protection of the present invention.
Claims (4)
1. a kind of based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation, which is characterized in that described
Method includes:
(1) the operation disjunctive model of the digital image manipulation chain based on blind source separating is constructed；
(2) according to the degree of correlation of digital picture feature, action type is distorted in preliminary identification；
(3) according to Dempster-Shafer evidence theory, estimation distorts operation confidence interval, accurately identifies and distort action type.
2. it is according to claim 1 based on obscure Treatment Effects separation image operational chain in action type recognition methods,
It is characterized in that, the operation disjunctive model of the digital image manipulation chain of the building based on blind source separating, specifically includes:
Matrixing is carried out to the composite character extracted from digital picture, obtains the digital image manipulation chain based on blind source separating
Operation disjunctive model, and according to the model to digital picture operational chain obscure Treatment Effects separation, acquisition singly distort operation
Feature assessment, for singly distort action type identification positive evidence is provided.
3. according to claim 1 or 2 based on action type identification side in the image operational chain for obscuring Treatment Effects separation
Method, which is characterized in that the degree of correlation according to digital picture feature, preliminary identification are distorted action type, specifically included:
Measurement undergoes the composite character of the digital picture after multiple distort to distort with certain obtained according to the operation disjunctive model
Correlation between operating characteristics, obtains degree of correlation between the two, tentatively judges whether the testing image undergoes the behaviour
It distorts forgery and what may be undergone distorts action type.
4. according to claim 1 or 2 based on action type identification side in the image operational chain for obscuring Treatment Effects separation
Method, which is characterized in that described according to Dempster-Shafer evidence theory, estimation distorts operation confidence interval, accurately identifies and usurp
Change action type, specifically include:
Joint singly distorts multiple detection algorithms of operation evidence obtaining, excavates different images feature；Pass through Dempster-Shafer evidence
The Decision fusion of theoretical composition rule, according to the fusion results of multiple detection algorithms, operation Estimating Confidence Interval is distorted in acquisition,
Action type is distorted in accurate differentiation image operational chain.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910367646.6A CN110097124B (en) | 2019-05-05 | 2019-05-05 | Method for identifying operation type in image operation chain based on confusion processing effect separation |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910367646.6A CN110097124B (en) | 2019-05-05 | 2019-05-05 | Method for identifying operation type in image operation chain based on confusion processing effect separation |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110097124A true CN110097124A (en) | 2019-08-06 |
| CN110097124B CN110097124B (en) | 2023-05-26 |
Family
ID=67446854
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910367646.6A Active CN110097124B (en) | 2019-05-05 | 2019-05-05 | Method for identifying operation type in image operation chain based on confusion processing effect separation |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110097124B (en) |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1941693A (en) * | 2006-01-12 | 2007-04-04 | 大连理工大学 | Method for watermarking small wave threshold digital audio multiple mesh based on blind source separation |
| CN101056350A (en) * | 2007-04-20 | 2007-10-17 | 大连理工大学 | Digital image evidence collecting method for detecting the multiple tampering based on the tone mode |
| CN101950408A (en) * | 2010-08-12 | 2011-01-19 | 合肥工业大学 | Digital image creditability measurement method based on D-S evidence theory |
| CN102073980A (en) * | 2011-01-06 | 2011-05-25 | 哈尔滨工程大学 | Compression sensing theory-based interactive supported dual watermark generating and detecting method |
| CN105426912A (en) * | 2015-11-12 | 2016-03-23 | 河南师范大学 | Blind separation method for replacement aliasing image |
| US20170091588A1 (en) * | 2015-09-02 | 2017-03-30 | Sam Houston State University | Exposing inpainting image forgery under combination attacks with hybrid large feature mining |
| CN107292133A (en) * | 2017-05-18 | 2017-10-24 | 深圳中兴网信科技有限公司 | The obfuscation method and device of artificial intelligence |
| US20180107887A1 (en) * | 2016-10-14 | 2018-04-19 | ID Metrics Group Incorporated | Tamper detection for identification documents |
-
2019
- 2019-05-05 CN CN201910367646.6A patent/CN110097124B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1941693A (en) * | 2006-01-12 | 2007-04-04 | 大连理工大学 | Method for watermarking small wave threshold digital audio multiple mesh based on blind source separation |
| CN101056350A (en) * | 2007-04-20 | 2007-10-17 | 大连理工大学 | Digital image evidence collecting method for detecting the multiple tampering based on the tone mode |
| CN101950408A (en) * | 2010-08-12 | 2011-01-19 | 合肥工业大学 | Digital image creditability measurement method based on D-S evidence theory |
| CN102073980A (en) * | 2011-01-06 | 2011-05-25 | 哈尔滨工程大学 | Compression sensing theory-based interactive supported dual watermark generating and detecting method |
| US20170091588A1 (en) * | 2015-09-02 | 2017-03-30 | Sam Houston State University | Exposing inpainting image forgery under combination attacks with hybrid large feature mining |
| CN105426912A (en) * | 2015-11-12 | 2016-03-23 | 河南师范大学 | Blind separation method for replacement aliasing image |
| US20180107887A1 (en) * | 2016-10-14 | 2018-04-19 | ID Metrics Group Incorporated | Tamper detection for identification documents |
| CN107292133A (en) * | 2017-05-18 | 2017-10-24 | 深圳中兴网信科技有限公司 | The obfuscation method and device of artificial intelligence |
Non-Patent Citations (1)
| Title |
|---|
| SHANGDE GAO, ET AL.: "Real-time detecting one specific tampering operation in multiple operator chains", 《SPRINGER》 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110097124B (en) | 2023-05-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Carvalho et al. | Illuminant-based transformed spaces for image forensics | |
| Mushtaq et al. | Digital image forgeries and passive image authentication techniques: a survey | |
| Zhang et al. | Reliable detection of LSB steganography based on the difference image histogram | |
| CN103605958A (en) | Living body human face detection method based on gray scale symbiosis matrixes and wavelet analysis | |
| CN107944416A (en) | A kind of method that true man's verification is carried out by video | |
| CN104244016B (en) | A kind of H264 video contents altering detecting method | |
| CN110457996A (en) | Moving Objects in Video Sequences based on VGG-11 convolutional neural networks distorts evidence collecting method | |
| CN102959588A (en) | Method for detecting tampering with color digital image based on chroma of image | |
| CN102045357A (en) | Affine cluster analysis-based intrusion detection method | |
| CN112668557A (en) | Method for defending image noise attack in pedestrian re-identification system | |
| CN103390151A (en) | Face detection method and device | |
| CN110147570A (en) | It is a kind of that method for distinguishing is known based on the electronic component of texture and shape feature | |
| CN111914912A (en) | Cross-domain multi-view target identification method based on twin conditional countermeasure network | |
| Zhang et al. | Image splicing localization using noise distribution characteristic | |
| CN110097124A (en) | Based on action type recognition methods in the image operational chain for obscuring Treatment Effects separation | |
| Destruel et al. | Color noise-based feature for splicing detection and localization | |
| CN101527041B (en) | Picture counterfeiting detection method based on shadow matte consistency | |
| Yu et al. | A multi-scale feature selection method for steganalytic feature GFR | |
| CN106530199B (en) | Multimedia integration steganalysis method based on window type hypothesis testing | |
| Cozzolino et al. | A comparative analysis of forgery detection algorithms | |
| CN111325185B (en) | Face fraud prevention method and system | |
| Chen et al. | Identification of image global processing operator chain based on feature decoupling | |
| CN111597524B (en) | Verification method and system for seal sample sampling personnel | |
| CN110852203B (en) | Multi-factor suspicious person identification method based on video feature learning | |
| Wang et al. | Towards improved steganalysis: When cover selection is used in steganography |
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 |