CN109360199A - The blind checking method of image repeat region based on Wo Sesitan histogram Euclidean measurement - Google Patents
The blind checking method of image repeat region based on Wo Sesitan histogram Euclidean measurement Download PDFInfo
- Publication number
- CN109360199A CN109360199A CN201811197053.1A CN201811197053A CN109360199A CN 109360199 A CN109360199 A CN 109360199A CN 201811197053 A CN201811197053 A CN 201811197053A CN 109360199 A CN109360199 A CN 109360199A
- Authority
- CN
- China
- Prior art keywords
- sesitan
- image
- histogram
- barycentric coodinates
- formula
- 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
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T5/00—Image enhancement or restoration
- G06T5/40—Image enhancement or restoration by the use of histogram techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10004—Still image; Photographic image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/10—Image acquisition modality
- G06T2207/10024—Color image
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/20—Special algorithmic details
- G06T2207/20021—Dividing image into blocks, subimages or windows
Abstract
The invention discloses a kind of blind checking methods of image repeat region based on Wo Sesitan histogram Euclidean measurement, comprising the following steps: image to be detected step A, is divided into several fritters；Step B, each fritter got for step A calculates its Wo Sesitan barycentric coodinates, obtains the Wo Sesitan barycentric coodinates collection of image to be detected；Step C, maximum piece of the Wo Sesitan barycentric coodinates (removing itself block) for choosing euclidean distance metric are original block, the doubtful repetition block of matching PatchMatch algorithm search are searched for using dough sheet, until searching out repeat region.For prior art, inventive algorithm can accurately detect that the repetition object in image, algorithm have preferable robustness.
Description
Technical field
The present invention relates to a kind of Wo Sesitan image repeat region blind checking methods, belong to technical field of image processing.
Background technique
With the fast development of information technology and popularizing for intelligent electronic device, people are obtained and modification image becomes to be cured
Hair is easy.Meanwhile the appearance of various images modification softwares, arbitrarily distort and provide convenience to picture material, also to for media,
The image use of news, criminal investigation, the administration of justice etc. brings huge challenge.Therefore, image detection and evidence collecting method and technology are studied
It is highly desirable, it makes image provide the safety guarantee of content authenticity while facilitating people to live.
It is a kind of common image forge means that duplication reproducing image, which is forged, certain part duplication in image is pasted same
The other regions of piece image achieve the purpose that hide or add new content.The counterfeiting detection method packet of duplication mobile image at present
It includes: traversal search method, image block autocorrelation matrix method and image block matching method etc..
The traversal search method of image block is first by image block to be identified, each image block as retrieval template, traversal
Residual image block judges whether there is the identical image block with template；Similar image block if it exists then increases image block size, weight
New search, until searching for less than identical image block.This method principle is simple, but operand is big, to the Shandong of noise
Stick is poor.
Image block autocorrelation matrix rule according to the image block that duplication is pasted there is the detection of very high autocorrelation to repeat to scheme
As region.This method piecemeal first simultaneously sets auto-correlation discrimination threshold, and traversal searches all image blocks, if searching is more than to set
Determine the image block of threshold value, then it is assumed that there is duplication paste operation between image block.The algorithm calculation amount is small compared to traversal, and deficiency is
It can only detect the duplication lesser situation of sticking block block size.
Image block matching method indicates image block with matrix, more all image block matrix, to find out identical image
Block.The detection effect of above method is preferable, but the calculation amount of algorithm is O (m2) (m is image block number), in addition, block feature dimension
It is higher larger so as to cause algorithm space consuming.
Summary of the invention
It is a kind of frank based on Wo Sesi technical problem to be solved by the present invention lies in overcoming the deficiencies of the prior art and provide
The blind checking method of the image repeat region of square figure Euclidean measurement, the duplication in detection image, which is pasted, forges region.
The blind checking method of image repeat region based on Wo Sesitan histogram Euclidean measurement of the invention, including it is following
Step:
Step A, image to be detected is divided into several fritters；
Step B, each fritter got for step A calculates its Wo Sesitan barycentric coodinates, to obtain to be checked
The Wo Sesitan barycentric coodinates collection of altimetric image；
Step C, according to Wo Sesitan barycentric coodinates, from first BOB(beginning of block), successively calculate in addition to itself between other pieces |
|·||2Norm, according to norm measure distance find out distance in the smallest piece as detection original block；After determining original block, utilize
Image block around PatchMatch algorithm search original block expands search range, positions image repeat region；
Step D, the repeat region testing result of image is exported.
As the further detailed protocol of method of the invention, step B falls into a trap the Wo Sesitan barycentric coodinates collection of nomogram picture,
Specifically according to the following steps:
Step B1, for appointing image block pi, piIt is divided into tri- Color Channels of B, G, R, and calculates the histogram in each channel
Figure；
Step B2, the Wo Sesitan center of gravity of B, G, R channel histogram is calculated separately；
Step B3, it is calculated according to the Wo Sesitan center of gravity being calculated using Wo Sesitan barycentric coodinates homing method
To image block Wo Sesitan barycentric coodinates.
As the further detailed protocol of method of the invention, the Wo Sesi of the histogram in each channel is calculated in step B2
Smooth barycentric coodinates are the Sinkhorn algorithms for using extension, specific as follows:
Wo Sesitan center of gravity is defined as,
Wherein, ∑NFor the simplex of normalization histogram；psIt is ∑NS-th interior of histogram, p are ∑sNInterior histogram；W
(p,ps) it is histogram p, psEntropy regularization OT distance；W(p,ps) be defined as
Wherein C is the cost for quantifying transmission quality between histogram, and γ is regularization parameter；T is N × N positive definite square of histogram
Matrix representation, H are the negentropy of T；
Wo Sesitan center of gravity, specific formula is calculated according to the differentiation of Sinkhorn algorithm are as follows: for all s≤S,The sum of histogram is represented for l >=0, s≤S, S, does following iteration:
And
Wherein K=e-C/γ, thenThus the accurate center of gravity of Wo Sesitan is obtained.
As the further detailed protocol of method of the invention, step B3 is specific as follows:
Step B301, according to the formula of Wo Sesitan barycentric coodinates:The formula
Be it is non-convex, the stationary point of the formula can be obtained by gradient descent method；
Step B302, local derviation is sought using λ of the chain type derivation rule to Wo Sesitan barycentric coodinates formula, obtains P (λ) gradient
Formula:WhereinIt isJacobian matrix,It is
psIn the gradient of P (λ)；
Step B303, P (λ) gradient formula provided using step B302 executes l iteration get Wo Sesitan center of gravity P(l)
(λ) replaces P (λ)；Formula in step 301,302 is become:ε (λ)=L (P(l)(λ),ps),
Step B304, it is calculated using backward recursiveTo obtainSpecific algorithm are as follows:
It defines first:
And
And
It can be calculated by formula, i.e.,Wherein v(L)Initial value isIt is calculated using following backward recursive:
Step B305, the Wo Sesitan barycentric coodinates for calculating each image block, obtain the Wo Sesitan barycentric coodinates of image
Collection.
Compared with prior art, the blind Detecting of the image repeat region of the invention based on Wo Sesitan histogram Euclidean measurement
Method has the advantages that
Method proposed by the present invention can accurately detect that forgery region is pasted in the duplication of image, and to rigid-object object
The detection of body has preferable robustness.
Detailed description of the invention
Fig. 1 is the blind checking method flow chart of Wo Sesitan image repeat region of the invention.
Specific embodiment
Technical solution of the present invention is described in detail with reference to the accompanying drawing:
Those skilled in the art can understand that unless otherwise defined, all terms used herein (including skill
Art term and scientific term) there is meaning identical with the general understanding of those of ordinary skill in fields of the present invention.Also
It should be understood that those terms such as defined in the general dictionary should be understood that have in the context of the prior art
The consistent meaning of meaning will not be explained in an idealized or overly formal meaning and unless defined as here.
Thinking of the invention is the Wo Sesitan barycentric coodinates image block as description, is made with multiple orthogonal Gaussian kernels
Base is indicated for Wo Sesitan.With image block indicating as Wo Sesitan barycentric coodinates at Wo Sesitanji, and it is used as its figure
As feature.According between image block characteristics | | | |2Norm determines original block position, is expanded using PatchMatch algorithm
Region of search obtains image repeat region testing result.
For the ease of public understanding, before technical solution of the present invention is described in detail, below first to the technology of the present invention
Related art involved in scheme is briefly described.
1. Wo Sesitan distance
Image is as a kind of data with high dimensional feature, except the information characteristics with general data midpoint, such as color and
Outside brightness, also have point and point, while while between geological information.Feature extraction or study based on European measurement system are calculated
Method all cannot characteristically be effectively retained geological information.Wo Sesitan measurement no longer simply regards image as common higher-dimension
Feature vector, and regarded as a statistic, the geological information of sample distribution can be retained in solution procedure.In the present invention
Wo Sesitan distance is entropy regularization optimal transmission distance, is specifically defined are as follows:
Matrix Ci,j=(Ci,j)i,jFor quantifying the cost of transmission quality between histogram, if section is in Euclid
Some position (x in spacei)iSampling, then C is usually Ci,j=| | xi-xj||α(α is usually positive number).Here regularization parameter
γ is usually positive number, this ensure optimal solution be it is unique, parallel computation is faster easier.(q, that is, p heres, represent s-th of histogram
Figure, similarly hereinafter)
2. Wo Sesitan center of gravity
Wo Sesitan center of gravity, also referred to as optimal transmission interpolation.Optimal transmission is optimal transmission theory (Optimal
Transport), it regards probability histogram as sand heap, and by consider from the mobile all sand grains of a histogram with by its
The generally the least expensive method of another histogram is re-formed to quantify distance between the two.If two distributions are compared to a pile
Two states of sand, then optimal transmission theory, which is substantially reflected, is transported to another state from a state this heap sand
" function " done.During acting, the available reservation of geometry state of sand, the geometric properties being distributed in Interpolation Process will not
It is destroyed.The specific definition of Wo Sesitan center of gravity are as follows:
The uniqueness of P (λ) is determined by the strong convexity for defining energy on the right side of formula (2).And Wo Sesitan center of gravity with it is common
Linear averaging ∑sλspsDifference, linear averaging correspond to l2Center of gravity in meaning cannot reflect geometrical characteristic, and Wo Sesitan is heavy
The heart is then without this problem.
3. improved PatchMatch algorithm
Barnes etc. proposes a kind of Fast Block match search algorithm PatchMatch, its biggest characteristic is that can use phase
Correlation between adjacent block carrys out communication sphere information.PatchMatch algorithm passes through random initializtion, propagation and random search
To match the image block in two images.The present invention is initialized using Wo Sesitan barycentric coodinates instead of PatchMatch algorithm
Random initializtion, then in same width image propagate and search for, determine repeat region.
The Blind Detect Algorithm of image repeat region based on Wo Sesitan histogram Euclidean measurement of the invention, process is such as
Shown in Fig. 1, specifically includes the following steps:
Step A, image to be detected is divided into several fritters using block codes；
Step B, for each fritter divided in step A, the feature of each fritter is extracted:
Existing various methods, such as Fridrich et al. can be used in the Feature Descriptor extraction of image block of the invention
[FridrichJ, SoukalD, Lukas J.Detectingofcopy-move forgeryindigital images,
Proceedings ofDigitalForensic ResearchWorkshop, 2003] it proposes and a kind of is become based on discrete cosine
The image duplication for changing (DCT) coefficient characteristics, which is pasted, forges detection algorithm, and the algorithm calculating speed is very fast, the disadvantage is that working as flat site
When existing in the image, error hiding rate is high, and performance decline is serious.Mohammad et al. [HashmiMF, Anand V,
KeskarAG.Copy-move ImageForgeryDetectionUsing anEfficientandRobustMethod Com
biningUndecimatedWaveletTransformand ScaleInvariantFeature Transform[J]
.AASRIProcedia, 2014,12 (9): 84-91.] propose the image of dynamic wavelet transformation coupling Scale invariant features transform
Detection algorithm is forged, and region is forged in positioning in airspace, it is difficult to characteristics of image dimension is reduced, has time complexity is high to lack
It falls into, and is difficult to detect the forgery region that affine geometric distortion distorts form, reduce the robustness of algorithm.Anand and Hashmi
Et al. [AnandV, HashmiMF, KeskarAG.ACopyMoveForgeryDetectionto Overcome
SustainedAttacksUsingDyadic WaveletTransformand SIFTMethods.IntelligentInfor
MationandDatabase Systems.2014:530-542.] WDT and SIFT are applied in the detection to distorted image,
The shortcomings that Detection accuracy with higher extracts the key point of characteristics of image by SIFT transformation, the technology is to mould
The forgeries forms such as paste, rotation are difficult to effectively identify.[Liu Xiaoxia, Li Feng, Xiong Bing are based on weber local feature for Liu Xiao rosy clouds et al.
Image mosaic detects [J] computer engineering and application, 2013,49 (12): 140-143.] one kind is proposed based on local weber
The image mosaic forensic technologies of feature (WLD), firstly, input picture discrete cosine transform is obtained DCT coefficient matrix, so
Characteristics of image is extracted using WLD afterwards, and establishes svm classifier model, this method detection accuracy is improved, but its calculation amount
Larger, WLD is only preferable to Edge Gradient Feature effect, can not be accurately extracted to the other feature of image so as to lead to erroneous detection
It surveys.King appoints China et al. [application study computer of Wang Renhua, Huo Hongtao, Jiang Min the .RANSAC algorithm in the duplication identification of same figure
Application study, 2014,31 (7): 2209-2212.] using RANSAC removal region is mismatched, it obtains model parameter, is accurately positioned pseudo-
Region is made, simulation result shows that RANSAC is used for image forge detection by it, can be effectively reduced feature and mismatch rate, improve algorithm
Detection accuracy, however RANSAC strategy there is still a need for setting threshold value, detection accuracy is still influenced by threshold value, and has been difficult to
Effect detection scaling etc. forges region.[Li Yan, Liu Nian, Zhang Bin wait the duplication of image mirrors to paste in tampering detection to Li Yan et al.
FI-SURF algorithm [J] communicates journal, 2015,36 (5): 54-65.] characteristic point, which is extracted, using SURF method and generates feature retouches
After stating symbol, duplication sticking area is overturn, then carries out characteristic matching, this method can improve the correct of characteristic point detection
Property, the error detection in detection process is reduced, but this method needs to sort to feature descriptor, so that duplication sticking area mirror
As overturning, the process computation complexity is higher.
In view of the deficiency of above-mentioned existing method, the present invention provides a kind of figure based on Wo Sesitan histogram Euclidean measurement
As the Blind Detect Algorithm of repeat region, with Wo Sesitan barycentric coodinates as image feature descriptor, with Wo Sesitan histogram
Euclidean metric blocks distance, with PatchMatch method expanded search repeat region.This method can consider that the geometry of image is special very well
Sign, is effectively detected out many places repeat region present in image, and have to the image containing Gaussian noise and lossy compression
Preferable robustness.This method is specific as follows:
Step B1, for image block pi, piResolve into the image p in tri- channels B, G, Rib、pigAnd pir；
Step B2, p is calculatedib、pigAnd pirWo Sesitan barycentric coodinates, as piFeature.Calculate Wo Sesitan
Barycentric coodinates are using the Sinkhorn algorithm extended.Specific method be (by taking channel B as an example, the method in other channels and this phase
Together):
Step B201, p is calculatedib、pigAnd pirHistogram simultaneously normalizes；
Step B202, it is secondly defined according to Wo Sesitan center of gravity and calculates histogram barycentric coodinates.
Wo Sesitan center of gravity is defined as,Wherein ∑NFor normalization histogram
Simplex；psIt is ∑NS-th interior of histogram, p are ∑sNInterior histogram；W (p, q) be histogram p, q entropy regularization OT away from
From.W (p, q) is defined asWherein C is transmitted between quantization histogram
The cost of quality；γ is regularization parameter.(q, that is, p heres, represent s-th of histogram) and due to the definition of Wo Sesitan center of gravity
It is not closure expression formula, therefore can be calculated according to the differentiation of Sinkhorn algorithm, specific formula are as follows: for all s
≤ S,For l >=0, s≤S, following iteration is done:
And
Wherein K=e-C/γ, thenThus the accurate center of gravity of Wo Sesitan is obtained.
Step B3, according to the Wo Sesitan center of gravity being calculated, the method returned using Wo Sesitan barycentric coodinates is calculated
The Wo Sesitan barycentric coodinates of a width small images are obtained, specific as follows:
Step B301, according to the formula of Wo Sesitan barycentric coodinates:The formula
Be it is non-convex, the stationary point of the formula can be obtained by gradient descent method；
Step B302, local derviation is sought using λ of the chain type derivation rule to Wo Sesitan barycentric coodinates formula, obtained:WhereinIt isJacobian matrix,It is q in P
The gradient of (λ).
Proposition 001: assuming that Wo Sesitan center of gravity P (λ) can be accurately calculated, for simplification, being firstly introduced into function (Φ, Ψ),
Wo Sesitan center of gravity iterative calculation in step B202 is written as:
P(l)(λ)=Ψ (b(l)(λ), λ) wherein
P(l+1)(λ)=Φ (b(l)(λ), λ) wherein
AndDefinition:
WhereinWithCorresponding to its opposite the first and second component Φ (b, λ) (Ψ (b,
Partial derivative λ)).In practice, accurate calculating above formula mode is not used, because these vectors only can be enough by iteration
Number carrys out approximate calculation, converges to enough accuracy.AndCalculating need to solve N × N linear system, and N is usually
The order of magnitude of 10^6.
Step B303, in view of the iterative calculation P (λ) provided in proposition 001, the l iteration proposed using step B202
Get Wo Sesitan center of gravity P(l)(λ) replaces P (λ).Step 301, the formula in 302 become:,
Step B304, it is calculated using backward recursiveTo obtainSpecific algorithm are as follows: first
Definition:
Wherein
Wherein
It can be calculated by formula:Wherein u(L)In step B303
It is defined, v(L)It is initialized to It is calculated using following backward recursive:
It calculatesThe method of algorithm be that forward direction is first carried out to recycle to calculate Wo Sesitan center of gravity P(l)(λ),
Then an inverse circulation is executed to realize v(l)And add up, it obtainsIt is specific as follows:
Input:
(w,r)←(0S,0S×N)
For l=1,2 ..., L
For l=L, L-1 ..., 1
g←∑srs
return
Step B305, image block Wo Sesitan barycentric coodinates are calculated, Wo Sesitan barycentric coodinates set is obtained.
Step C, according to the Wo Sesitan barycentric coodinates collection that is calculated, using PatchMatch expanded search image block,
Obtain the repeat region of image.
Step C1, according to the Wo Sesitan barycentric coodinates that are calculated, from first BOB(beginning of block), successively calculate except itself in addition to
Between other pieces | | | |2Norm, according to norm measure distance find out distance in the smallest piece as detection original block.
Step C2, after determining original block, using the image block around PatchMatch algorithm search original block, to expand
Search range positions image repeat region.
Those skilled in the art can understand that can realize these structure charts with computer program instructions
And/or the combination of each frame and these structure charts and/or the frame in block diagram and/or flow graph in block diagram and/or flow graph.It can be with
These computer program instructions are supplied to the processing of general purpose computer, special purpose computer or other programmable data processing methods
Device generates machine, creates to be performed instruction by the processor of computer or other programmable data processing methods
For realizing the method specified in the frame or multiple frames of structure chart and/or block diagram and/or flow graph.
Embodiments of the present invention are explained in detail above in conjunction with attached drawing, but the invention is not limited to above-mentioned
Embodiment within the knowledge of a person skilled in the art can also be before not departing from present inventive concept
It puts and makes a variety of changes.
Claims (4)
1. a kind of blind checking method of the image repeat region based on Wo Sesitan histogram Euclidean measurement, it is characterised in that including
Following steps:
Step A, image to be detected is divided into several fritters；
Step B, each fritter got for step A calculates its Wo Sesitan barycentric coodinates, to obtain mapping to be checked
The Wo Sesitan barycentric coodinates collection of picture；
Step C, according to Wo Sesitan barycentric coodinates, from first BOB(beginning of block), successively calculate in addition to itself between other pieces | | |
|2Norm, according to norm measure distance find out distance in the smallest piece as detection original block；After determining original block, utilize
Image block around PatchMatch algorithm search original block expands search range, positions image repeat region；
Step D, the repeat region testing result of image is exported.
2. the method as described in claim 1, which is characterized in that step B falls into a trap the Wo Sesitan barycentric coodinates collection of nomogram picture, tool
Body is according to the following steps:
Step B1, for appointing image block pi, piIt is divided into tri- Color Channels of B, G, R, and calculates the histogram in each channel；
Step B2, the Wo Sesitan center of gravity of B, G, R channel histogram is calculated separately；
Step B3, figure is calculated using Wo Sesitan barycentric coodinates homing method according to the Wo Sesitan center of gravity being calculated
As block Wo Sesitan barycentric coodinates.
3. method according to claim 2, which is characterized in that calculate the Wo Sesitan of the histogram in each channel in step B2
Barycentric coodinates are the Sinkhorn algorithms for using extension, specific as follows:
Wo Sesitan center of gravity is defined as,
Wherein, ∑NFor the simplex of normalization histogram；psIt is ∑NS-th interior of histogram, p are ∑sNInterior histogram；W(p,
ps) it is histogram p, psEntropy regularization OT distance；W(p,ps) be defined as
Wherein C is the cost for quantifying transmission quality between histogram, and γ is regularization parameter；T is N × N positive definite square of histogram
Matrix representation, H are the negentropy of T；
Wo Sesitan center of gravity, specific formula is calculated according to the differentiation of Sinkhorn algorithm are as follows: for all s≤S,The sum of histogram is represented for l >=0, s≤S, S, does following iteration:
And
Wherein K=e-C/γ,
According toThus the accurate center of gravity of Wo Sesitan is obtained.
4. method as claimed in claim 3, which is characterized in that step B3 is specific as follows:
Step B301, according to the formula of Wo Sesitan barycentric coodinates:The formula right and wrong
It is convex, the stationary point of the formula can be obtained by gradient descent method；
Step B302, local derviation is sought using λ of the chain type derivation rule to Wo Sesitan barycentric coodinates formula, obtains P (λ) gradient public affairs
Formula:WhereinIt isJacobian matrix, ▽ L (P (λ), ps) it is ps
In the gradient of P (λ)；
Step B303, P (λ) gradient formula provided using step B302 executes l iteration get Wo Sesitan center of gravity P(l)(λ) comes
Instead of P (λ)；Formula in step 301,302 is become:ε (λ)=L (P(l)(λ),ps),u(l)=▽ L (P(l)(λ),ps)；
Step B304, it is calculated using backward recursiveTo obtain ▽ εL(λ), specific algorithm are as follows:
It defines first:
And
And
▽εL(λ) can be calculated by formula, i.e.,Wherein v(L)Initial value is It is calculated using following backward recursive:
Step B305, the Wo Sesitan barycentric coodinates for calculating each image block, obtain the Wo Sesitan barycentric coodinates collection of image.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811197053.1A CN109360199B (en) | 2018-10-15 | 2018-10-15 | Blind detection method of image repetition region based on Watherstein histogram Euclidean measurement |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811197053.1A CN109360199B (en) | 2018-10-15 | 2018-10-15 | Blind detection method of image repetition region based on Watherstein histogram Euclidean measurement |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109360199A true CN109360199A (en) | 2019-02-19 |
| CN109360199B CN109360199B (en) | 2021-10-08 |
Family
ID=65349450
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811197053.1A Active CN109360199B (en) | 2018-10-15 | 2018-10-15 | Blind detection method of image repetition region based on Watherstein histogram Euclidean measurement |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109360199B (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110555792A (en) * | 2019-08-16 | 2019-12-10 | 广东外语外贸大学南国商学院 | Image tampering blind detection method based on normalized histogram comprehensive feature vector |
| CN111754441A (en) * | 2020-06-29 | 2020-10-09 | 国网甘肃省电力公司电力科学研究院 | Passive detection method for image copy-paste forgery |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150293884A1 (en) * | 2014-04-14 | 2015-10-15 | Kyoto University | Method to compute the barycenter of a set of histograms |
| CN106469443A (en) * | 2015-08-13 | 2017-03-01 | 微软技术许可有限责任公司 | Machine vision feature tracking systems |
| CN107230215A (en) * | 2016-03-24 | 2017-10-03 | 想象技术有限公司 | Learning characteristic motion detection |
-
2018
- 2018-10-15 CN CN201811197053.1A patent/CN109360199B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150293884A1 (en) * | 2014-04-14 | 2015-10-15 | Kyoto University | Method to compute the barycenter of a set of histograms |
| CN106469443A (en) * | 2015-08-13 | 2017-03-01 | 微软技术许可有限责任公司 | Machine vision feature tracking systems |
| CN107230215A (en) * | 2016-03-24 | 2017-10-03 | 想象技术有限公司 | Learning characteristic motion detection |
Non-Patent Citations (3)
| Title |
|---|
| DAVIDE COZZOLINO等: "《COPY-MOVE FORGERY DETECTION BASED ON PATCHMATCH》", 《2014 IEEE INTERNATIONAL CONFERENCE ON IMAGE PROCESSING(ICIP)》 * |
| 焦丽鑫 等: "《基于均值漂移的图像复制粘贴伪造盲检测》", 《计算机应用》 * |
| 赵春江 等: "《基于Wasserstein距离的目标识别中的研究》", 《NCIRCS2004第一届全国信息检索与内容安全学术会议论文集》 * |
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110555792A (en) * | 2019-08-16 | 2019-12-10 | 广东外语外贸大学南国商学院 | Image tampering blind detection method based on normalized histogram comprehensive feature vector |
| CN110555792B (en) * | 2019-08-16 | 2022-05-17 | 广东外语外贸大学南国商学院 | Image tampering blind detection method based on normalized histogram comprehensive feature vector |
| CN111754441A (en) * | 2020-06-29 | 2020-10-09 | 国网甘肃省电力公司电力科学研究院 | Passive detection method for image copy-paste forgery |
| CN111754441B (en) * | 2020-06-29 | 2023-11-21 | 国网甘肃省电力公司电力科学研究院 | Image copying, pasting and forging passive detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN109360199B (en) | 2021-10-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110349136A (en) | A kind of tampered image detection method based on deep learning | |
| CN112052831B (en) | Method, device and computer storage medium for face detection | |
| CN107862680B (en) | Target tracking optimization method based on correlation filter | |
| AlSawadi et al. | Copy-move image forgery detection using local binary pattern and neighborhood clustering | |
| Yao et al. | Detecting copy-move forgery using non-negative matrix factorization | |
| CN107301643A (en) | Well-marked target detection method based on robust rarefaction representation Yu Laplce's regular terms | |
| CN106056523A (en) | Digital image stitching tampering blind detection method | |
| Mushtaq et al. | Image copy move forgery detection: a review | |
| CN109360199A (en) | The blind checking method of image repeat region based on Wo Sesitan histogram Euclidean measurement | |
| Isaac et al. | A key point based copy-move forgery detection using HOG features | |
| Lian | Pedestrian detection using quaternion histograms of oriented gradients | |
| Kovacs et al. | Orientation based building outline extraction in aerial images | |
| CN106127236B (en) | Classification of Polarimetric SAR Image method based on Di Li Cray MRF mixed model | |
| Kaur et al. | Illumination invariant face recognition | |
| Zhao et al. | A comprehensive study on third order statistical features for image splicing detection | |
| Muzaffer et al. | Recent keypoint based copy move forgery detection techniques | |
| CN110555792B (en) | Image tampering blind detection method based on normalized histogram comprehensive feature vector | |
| Liang et al. | Robust hashing with local tangent space alignment for image copy detection | |
| Rouis et al. | Local geometry analysis for image tampering detection | |
| Yang et al. | A robust scheme for copy detection of 3D object point clouds | |
| Kale et al. | Copy Move Forgery Detection-A Robust Technique | |
| CN112906508A (en) | Face living body detection method based on convolutional neural network | |
| CN112257688A (en) | GWO-OSELM-based non-contact palm in-vivo detection method and device | |
| Ke et al. | An efficient blind detection algorithm of median filtered image | |
| KR100711223B1 (en) | Face recognition method using Zernike/LDA and recording medium storing the method |
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 |