CN110348511A - A kind of picture reproduction detection method, system and electronic equipment - Google Patents
A kind of picture reproduction detection method, system and electronic equipment Download PDFInfo
- Publication number
- CN110348511A CN110348511A CN201910613965.0A CN201910613965A CN110348511A CN 110348511 A CN110348511 A CN 110348511A CN 201910613965 A CN201910613965 A CN 201910613965A CN 110348511 A CN110348511 A CN 110348511A
- Authority
- CN
- China
- Prior art keywords
- picture
- detected
- reproduction
- feature
- fragment
- 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.)
- Withdrawn
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
Abstract
The present invention relates to pictures to detect identification technology field, in particular to picture reproduction detection method, system and electronic equipment.Comprising steps of providing picture to be detected；Sliding window sampling is carried out to picture to be detected, to obtain multiple picture fragments；Picture fragment is input in convolutional neural networks and carries out feature extraction to obtain the picture feature of each picture fragment；Determine whether picture to be detected is reproduction picture using disaggregated model based on picture feature, sliding window sampling is carried out to obtain multiple picture fragments to picture, multiple picture fragments effectively represent the characteristic of picture to be detected, multiple picture fragments are input in convolutional neural networks and carry out feature extraction to obtain the picture feature of each picture fragment, determine whether picture to be detected is reproduction picture using trained disaggregated model based on the picture feature of extraction, manpower can not only be saved and reduce cost, and offer reference picture is not provided and is compared, detection efficiency can be improved and be applicable in the flexibility ratio of scene.
Description
[technical field]
The present invention relates to pictures to detect identification technology field, in particular to a kind of picture reproduction detection method, system and electricity
Sub- equipment.
[background technique]
With the development of economy and universal, the personal reference of internet, the business such as remotely open an account be increasingly becoming finance,
The important service of the industries such as telecommunications and electric business.In these business, for examining for convenient, safety and laws and regulations etc.
Consider, it may be necessary to which user is shot and uploaded the certificate photo of oneself by smart phone, tablet computer or the first-class equipment of network shooting
Piece.However, what some certificate photographs were not obtained by being directed at true document photography, pass through reproduction computer screen or hand
Certificate picture on machine screen and formed.Certificate in these reproduction photos may be not belonging to user, it is also possible to once
It by editor, forges or distorts, do not have legal effect, therefore be identified as violation certificate photograph.Existing picture detection method
What mostly artificial examination or the picture for having reference based on one were verified, efficiency and accuracy rate are often relatively low, cost
It is relatively high.
[summary of the invention]
To overcome low efficiency existing for current picture detection method and defect at high cost, the present invention provides a kind of figure
Piece reproduction detection method, system and electronic equipment.
In order to solve the above-mentioned technical problem a kind of picture reproduction detection method is provided, include the following steps: that S1, offer are to be checked
Mapping piece；
S2, sliding window sampling is carried out to the picture to be detected, to obtain multiple picture fragments；S3, by the multiple picture
Fragment, which is input in convolutional neural networks, carries out feature extraction to obtain the picture feature for representing each picture fragment；And it S4, is based on
The picture feature of extraction determines whether each picture fragment is reproduction using trained disaggregated model respectively；And S5, root
Determine whether the picture to be detected is reproduction picture according to the sum for the picture fragment for being determined as reproduction in the step S4.
Preferably, the step S2 specifically comprises the following steps: step S21: the picture to be detected is carried out gray processing
Processing, obtains the gray level image of picture to be detected；And step S22: one rectangle frame of setting, by the rectangle frame described to be checked
Mapping on piece is moved by setting rule, and picture to be detected is successively obtained multiple figures according to the size acquisition of the rectangle frame
Piece fragment.
Preferably, the size of each picture fragment is identical, has overlapping region, any between two fragments of arbitrary neighborhood
Certain spacing is spaced between the edge connection of two adjacent picture fragments or two fragments of arbitrary neighborhood.
Preferably, the picture feature extracted in the step S3 include spectrum signature, textural characteristics, color characteristic and
At least one of edge feature.
Preferably, the picture reproduction detection method further include step S10, to the picture to be detected according to preset
Ratio zooms in and out, and the step S10 is between the step S1 and the step S2.
Preferably, the training of the disaggregated model includes: building picture database to be detected, the image data to be detected
Library includes whether one group of picture to be detected and the corresponding label of every picture to be detected, the label indicate picture to be detected
For reproduction picture；Calculate the picture feature of each picture to be detected in the picture database to be detected；Based on the label and
The picture feature constructs training set；And the classification mould is trained on the training set using supporting vector machine model
Type.
Preferably, the textural characteristics of acquisition include the bag of words of picture to be measured, and the color characteristic of acquisition includes that color is straight
Fang Tu.
In order to solve the above-mentioned technical problem, a kind of picture reproduction detection system is also provided, comprising: sampling module, for pair
The picture to be detected carries out sliding window sampling, to obtain multiple picture fragments；Characteristic extracting module is used for the multiple picture
Fragment, which is input in convolutional neural networks, carries out feature extraction to obtain the picture feature for representing each picture fragment；Picture classification mould
Block determines whether the picture to be detected is reproduction figure using trained disaggregated model for the picture feature based on extraction
Piece；First determination module, be configured to the picture feature based on extraction determined respectively using trained disaggregated model it is described every
Whether a picture fragment is the second determination module of reproduction, is configured to be determined as the picture fragment of reproduction according to the first determination module
Sum determine whether the picture to be detected is reproduction picture.
Preferably, further include Zoom module, be configured to the picture to be detected of offer according to preset ratio into
Row scaling.
In order to solve the above-mentioned technical problem the present invention also provides a kind of electronic equipment, including memory and processor, described
Computer program is stored in memory, the computer program is arranged to execute picture reproduction inspection as described above when operation
Survey method；The processor is arranged to execute picture reproduction detection method as described above by the computer program.
Compared with the existing technology, sliding window sampling is carried out to obtain multiple picture fragments to picture, multiple picture fragments are effective
Represent the characteristic of picture to be detected, by the multiple picture fragment be input to one by one in convolutional neural networks carry out feature mention
It takes to obtain the picture feature for representing each picture fragment, the picture feature based on extraction is determined using trained disaggregated model
Whether the picture to be detected is reproduction picture, and not needing to carry out whole picture analysis can be completed, and can be greatly improved point
Analyse speed；And determine whether the picture to be detected is reproduction, i.e., only needs based on the sum for the picture fragment for being determined as reproduction
The threshold value that sum reaches setting is considered that reproduction picture, the remaining picture fragment for not determining whether reproduction do not need again
Continue, also substantially increase analysis speed to a certain extent, manpower can not only be saved and reduce cost, and does not need to mention
Picture is compared for reference, can improve well detection efficiency and be applicable in scene flexibility it is higher.
The picture to be processed, which is carried out gray processing processing, can reduce the complexity handled picture to be detected, and set
One rectangle frame is moved using the rectangle frame according to default rule, so that the picture fragment size that acquisition obtains is consistent,
The picture fragment information of acquisition is stablized effectively, to improve the accuracy of analysis.
Between two fragments of arbitrary neighborhood with overlapping region, arbitrary neighborhood two picture fragments edge connection or
Be spaced certain spacing between two fragments of person's arbitrary neighborhood so that acquisition obtain picture fragment have the effect of it is different,
User can be very good the quality condition according to picture to be detected, select suitable rectangle frame movement rule, relatively reliable to obtain
Analysis as a result, for example, when according to artificial experience determine picture be reproduction probability it is not high when, two picture fragments can be made
Between there are a fixed spacings, both can guarantee accuracy in this way, can also improve efficiency well；It is understood that ought artificially pass through
Test determine picture be reproduction probability it is relatively high when, then make between two picture fragments of arbitrary neighborhood exist have overlay region
Domain, arbitrary neighborhood two picture fragments edge connection, improve to the sampling area of picture to be detected, to improve the standard of detection
True property.
The picture to be detected is zoomed in and out according to preset ratio, can be selected very well according to the quality of picture to be detected
Suitable scaling zooms in and out, so that more significant to the signature analysis of picture to be detected.
Picture reproduction detection system provided by the invention and electronic equipment have and the picture reproduction detection method phase
Same beneficial effect.
[Detailed description of the invention]
Fig. 1 is the flow diagram of the picture reproduction detection method provided in first embodiment of the invention；
Fig. 2 a is the details flow diagram of the picture reproduction detection method step S2 provided in first embodiment of the invention；
Fig. 2 b is the schematic diagram of the step S2 implementation process for the picture reproduction detection method that first embodiment of the invention provides；
Fig. 3 is the process signal of the variant embodiment of the picture reproduction detection method provided in first embodiment of the invention
Figure；
Fig. 4 is the module diagram of detection unit in the picture reproduction detection system provided in second embodiment of the invention；
Fig. 5 is the module diagram of the picture reproduction detection system variant embodiment provided in second embodiment of the invention；
Fig. 6 is the module diagram of the electronic equipment provided in third embodiment of the invention；
Fig. 7 is adapted for the structural schematic diagram for the computer system for realizing the server of the embodiment of the present invention.
Description of drawing identification:
100, picture to be measured；200, rectangle frame；300, first position；400, the second position；50, picture reproduction detection system
System；501, sampling module；502, characteristic extracting module；503, picture classification module；504, the first determination module；505, second sentences
Cover half block；506, unit for scaling；60, electronic equipment；601, memory；602, processor；800, computer system；801, in
Central Processing Unit (CPU)；802, memory (ROM)；803,RAM；804, bus；805, I/O interface；806, importation；
807, output par, c；808, storage section；809, communications portion；810, driver；811, detachable media.
[specific embodiment]
In order to make the purpose of the present invention, technical solution and advantage are more clearly understood, below in conjunction with attached drawing and embodiment,
The present invention will be described in further detail.It should be appreciated that the specific embodiments described herein are merely illustrative of the present invention,
It is not intended to limit the present invention.
Referring to Fig. 1, the picture reproduction detection method that first embodiment of the invention provides, includes the following steps:
S1, picture to be detected is provided；
S2, sliding window sampling is carried out to the picture to be detected, to obtain multiple picture fragments；
S3, it the multiple picture fragment is input in convolutional neural networks carries out feature extraction and to obtain represents each picture
The picture feature of fragment；And
Whether S4, the picture feature based on extraction determine each picture fragment using trained disaggregated model respectively
For reproduction；And S5, whether the picture to be detected is determined according to the sum for the picture fragment for being determined as reproduction in the step S4
For reproduction picture.
In step sl, the photo to be checked is identity card picture, passport picture, driver's license picture or other needs to examine
The picture of survey.
In the step S2, sliding window sampling is carried out to the picture to be detected, to obtain multiple picture fragments
(Patch).It is appreciated that the picture fragment (Patch) is to be acquired to obtain from picture to be detected, and therefore, energy
The truth of the picture to be detected is reacted well.
Fig. 2 a is please referred to, in some specific embodiments, the step S2 specifically comprises the following steps:
Step S2: sliding window sampling is carried out to the picture to be detected, to obtain multiple picture fragments.Step S2 is specifically included
Step S21~S22.It is appreciated that step S21~S22 is only a kind of embodiment of the embodiment, embodiments thereof are not
It is defined in step S21~S22.
Step S21: the picture to be detected is subjected to gray processing processing, obtains the gray processing picture of picture to be detected；And
Step S22: one rectangle frame of setting is moved the rectangle frame on the picture to be detected by setting rule
It is dynamic, successively picture to be detected is acquired according to the size of the rectangle frame to obtain multiple picture fragments.
Optionally, have overlapping region between the two neighboring picture fragment, the Duplication of the overlapping region be 0-1 it
Between, Duplication is 0.3-0.8 when best, then the movement of the rectangle frame is moved according to the Duplication, to figure to be detected
Piece is sampled.
For example, rectangle frame size may be set to the specification of 2x2 or 3x3, as shown in figure 2b, rectangle frame 200 is set as 3x3
Specification, rectangle frame 200 starts to move on the initial position of picture 100 to be detected, and initial position is located at picture 100 to be detected
Upper left position, initial position is defined as first position 300, it is mobile it is primary after, the rectangle frame 200 is moved to the second position
400, after repeatedly moving and sampling, successively picture 100 to be detected is obtained according to the size acquisition of the rectangle frame multiple
Picture fragment.
There is the region of overlapping, so that first position 300 and the second position between first position 300 and the second position 400
There is relevance between the 400 two picture fragments obtained, improve the accuracy to picture analyzing to be detected.
As an implementation, between two neighboring picture fragment between non-overlapping region or two neighboring second picture
Edge be just overlapped or two neighboring picture fragment between it is spaced apart.It is preferred that being treated using sliding sample mode
Detection picture is acquired to obtain the m × n picture fragments being sized.
In the step S3, the multiple picture fragment is input in convolutional neural networks and carries out feature extraction to obtain
Represent the picture feature of each picture fragment.Optionally, the convolutional neural networks include CNN convolutional neural networks.The figure
Piece feature includes one or more kinds of in color characteristic, contour feature, textural characteristics or other features.The CNN convolution mind
It include input layer and convolutional layer through network.The input layer is used to utilize convolution kernel for inputting picture fragment, the convolutional layer
Extract the picture feature of each picture fragment in every picture.
Due to the interference effect of the wave between camera photosensitive element and display, by reproduction computer or mobile phone screen
It will appear apparent periodical color fringe on picture obtained from certificate photograph, referred to as " moire fringes "." moire fringes " are differences
The important clue of true picture and reproduction picture.Since " moire fringes " are presented periodically, the characteristic of " moire fringes " in frequency domain
It can be more obvious.In addition, the color of " moire fringes " is also different from normal picture.Therefore, using due to camera photosensitive element with
The interference effect of wave between display and generate " moire fringes ", calculate obtain picture color characteristic, contour feature, texture
At least one of feature can effectively improve the precision of reproduction picture recognition.Illustratively, above-mentioned three kinds of features both can be only
It is vertical to use, it also may be incorporated in and come together to use as assemblage characteristic.
In step s3, it when obtaining the textural characteristics in the picture feature for representing each picture fragment, including calculates and obtains
The bag of words (Bag-of-words) of picture fragment.Bag of words are a kind of statistical presentation of picture textural characteristics, Ke Yiyou
The entirety and local characteristics of effect description picture.The acquisition of bag of words may include two key steps (a) and (b):
(a) it establishes code book: extracting a large amount of image descriptor (such as SIFT, HOG at random from a trained picture set
Deng), each image descriptor is a vector, is clustered using K-means clustering algorithm to these image descriptors, is obtained
To K classification (K is adjustable parameter, representative value 1024,2048,10000 etc.).Cluster centre is referred to as " word ", gathers
The all categories that class obtains form one " code book ".
(b) picture describes: for a width picture, extracting feature descriptor (such as SIFT, HOG) in dense mode；It is right
In each descriptor, most like cluster centre (namely word) is searched in the codebook.Different words are counted to occur in the picture
Frequency, form a histogram.L1 normalization is made to the histogram, the picture texture based on bag of words obtained to the end is special
Sign.
According to one embodiment of present invention, the color characteristic for obtaining picture to be detected may include obtaining picture fragment
Color histogram.Color histogram is a kind of statistical presentation of picture color feature, and described is different color in whole picture figure
Shared ratio in piece, and it is not relevant for spatial position locating for every kind of color.Color histogram is closely related with color space,
The color histogram for calculating certificate picture may include calculating RGB color histogram, and hsv color histogram and Lab color are straight
Side's figure etc..
In the step S4, the picture feature based on extraction is determined described each respectively using trained disaggregated model
Whether picture fragment is reproduction.
In the step S4, the disaggregated model can be obtained by training.The training of the disaggregated model includes such as
Lower step:
Step T01, samples pictures database is constructed, the samples pictures database includes one group of samples pictures and sample
The corresponding label of picture, whether the label instruction samples pictures are reproduction picture；
Step T02, the picture feature of each samples pictures in the samples pictures database is obtained；
Step T03, training set is constructed based on the label and the picture feature；And
Step T04, the disaggregated model is trained on the training set using supporting vector machine model.
Obtained in the step T02 in the samples pictures database method of the picture feature of each samples pictures and
The step S3's is identical, and specifically acquisition color characteristic, the mode of textural characteristics are also identical with method mentioned above,
This is repeated no more.
In the step T03, the training set can be indicated are as follows: { xi, yi, i=1,2 ... ... N, wherein N is sample to S=
The number of samples pictures in this picture database, Xi are the feature vector of samples pictures Ii, and Yi is the label of samples pictures Ii
(classification belonging to samples pictures Ii being indicated, for example whether being reproduction picture).
In step T04, the disaggregated model is trained on the training set using supporting vector machine model, using line
Property support vector machines (Linear SVM) model, on training set S training one disaggregated model C.Based on trained classification mould
Type C in step s 4 can be by the picture feature of each picture fragment obtained in step s3 for given samples pictures
It is input to disaggregated model C, obtains recognition result y (J).Y (J) indicates classification belonging to samples pictures J, for example, true picture or
Reproduction figure.
In the step S5, determined according to the sum for the picture fragment for being determined as reproduction in the step S4 described to be checked
Whether mapping piece is reproduction picture.The number is the threshold value that user rule of thumb sets, when the sum of picture fragment is greater than setting
Threshold value when, then it is assumed that reproduction has occurred, when the sum of picture fragment be less than setting threshold value when, then it is assumed that there is no turning over
It claps.
Referring to Fig. 3, further including step S10, to the mapping to be checked in picture reproduction detection method provided by the invention
Piece is zoomed in and out according to preset ratio.The step S10 is between the step S1 and the step S2.In this step,
Preset scaling is mainly to be determined according to the resolution ratio for the target for needing to identify in picture to be detected.For example, when to be checked
When the resolution ratio of mapping piece is relatively high, can increase scaling appropriate, can not be influenced in this way to the true of picture to be detected
Puppet determines also improve the analysis speed to picture to be detected.Certainly, when the resolution ratio of picture to be detected is lower, it should suitable
When reduction scaling, avoid picture to be detected distortion during scaling, influence final analysis result.Optionally, right
Picture to be detected is zoomed in and out can be carried out by ReSize function.
Referring to Fig. 4, the second embodiment of the present invention provides a kind of picture reproduction detection system 50 comprising: sampling
Module 501, characteristic extracting module 502, picture classification module 503, the first determination module 504, the second determination module 505.
Sampling module 501, for carrying out sliding window sampling to the picture to be detected, to obtain multiple picture fragments；
Characteristic extracting module 502 is mentioned for the multiple picture fragment to be input to progress feature in convolutional neural networks
It takes to obtain the picture feature for representing each picture fragment；
Picture classification module 503, for the picture feature based on extraction using trained disaggregated model determine it is described to
Detect whether picture is reproduction picture.
First determination module 504 is configured to the picture feature based on extraction and is sentenced respectively using trained disaggregated model
Whether fixed each picture fragment is reproduction；
Second determination module 505 is configured to be determined as the sum of the picture fragment of reproduction according to the first determination module 504
Determine whether the picture to be detected is reproduction picture.
Referring to Fig. 5, the picture reproduction detection system 50 further includes Zoom module 506, it is configured to the institute to offer
Picture to be detected is stated to zoom in and out according to preset ratio.
Referring to Fig. 6, the third embodiment of the present invention provides a kind of electronic equipment 60, including memory 601 and processor
602, computer program is stored in the memory 601, the computer program is arranged to be executed when operation as first is real
Apply picture reproduction detection method described in example；
The processor 602 is arranged to execute picture reproduction as in the first embodiment by the computer program
Detection method.
Below with reference to Fig. 7, it illustrates the terminal device/server computers for being suitable for being used to realize the embodiment of the present application
The structural schematic diagram of system 800.Terminal device/server shown in Fig. 7 is only an example, should not be to the embodiment of the present application
Function and use scope bring any restrictions.
As shown in fig. 7, computer system 800 includes central processing unit (CPU) 801, it can be read-only according to being stored in
Program in memory (ROM) 802 or be loaded into the program in random access storage device (RAM) 803 from storage section 808 and
Execute various movements appropriate and processing.In RAM 803, also it is stored with system 800 and operates required various program sum numbers
According to.CPU 801, ROM 802 and RAM 803 are connected with each other by bus 804.Input/output (I/O) interface 805 also connects
To bus 804.
I/O interface 805 is connected to lower component: the importation 806 including keyboard, mouse etc.；It is penetrated including such as cathode
The output par, c 807 of spool (CRT), liquid crystal display (LCD) etc. and loudspeaker etc.；Storage section 808 including hard disk etc.；
And the communications portion 809 of the network interface card including LAN card, modem etc..Communications portion 809 via such as because
The network of spy's net executes communication process.Driver 810 is also connected to I/O interface 805 as needed.Detachable media 811, such as
Disk, CD, magneto-optic disk, semiconductor memory etc. are mounted on as needed on driver 810, in order to read from thereon
Computer program be mounted into storage section 808 as needed.
Disclosed embodiment according to the present invention may be implemented as computer software above with reference to the process of flow chart description
Program.For example, embodiment disclosed by the invention includes a kind of computer program product comprising be carried on computer-readable medium
On computer program, which includes the program code for method shown in execution flow chart.In such reality
It applies in example, which can be downloaded and installed from network by communications portion 809, and/or from detachable media
811 are mounted.When the computer program is executed by central processing unit (CPU) 801, limited in execution the present processes
Above-mentioned function.It should be noted that computer-readable medium described herein can be computer-readable signal media or
Computer readable storage medium either the two any combination.Computer readable storage medium for example can be-but not
Be limited to-electricity, magnetic, optical, electromagnetic, infrared ray or semiconductor system, device or device, or any above combination.It calculates
The more specific example of machine readable storage medium storing program for executing can include but is not limited to: have the electrical connection, portable of one or more conducting wires
Formula computer disk, hard disk, random access storage device (RAM), read-only memory (ROM), erasable programmable read only memory
(EPROM or flash memory), optical fiber, portable compact disc read-only memory (CD-ROM), light storage device, magnetic memory device or
The above-mentioned any appropriate combination of person.
The calculating of the operation for executing the application can be write with one or more programming languages or combinations thereof
Machine program code, described program design language include object oriented program language-such as Java, Smalltalk, C+
+, it further include conventional procedural programming language-such as " such as " language or similar programming language.Program code can
Fully to execute, partly be executed on management end computer, as an independent software package on management end computer
It executes, partially part executes on the remote computer or completely in remote computer or server on management end computer
Upper execution.In situations involving remote computers, remote computer can pass through the network of any kind --- including local
Net (LAN) or the domain wide area network (WAN) are connected to management end computer, or, it may be connected to outer computer (such as using because
Spy nets service provider to connect by internet).
Flow chart and block diagram in attached drawing are illustrated according to the system of the various embodiments of the application, method and computer journey
The architecture, function and operation in the cards of sequence product.In this regard, each box in flowchart or block diagram can generation
A part of one module, program segment or code of table, a part of the module, program segment or code include one or more use
The executable instruction of the logic function as defined in realizing.It should also be noted that in some implementations as replacements, being marked in box
The function of note can also occur in a different order than that indicated in the drawings.For example, two boxes succeedingly indicated are actually
It can be basically executed in parallel, they can also be executed in the opposite order sometimes, and this depends on the function involved.Also it to infuse
Meaning, the combination of each box in block diagram and or flow chart and the box in block diagram and or flow chart can be with holding
The dedicated hardware based system of functions or operations as defined in row is realized, or can use specialized hardware and computer instruction
Combination realize.
Compared with the existing technology, sliding window sampling is carried out to obtain multiple picture fragments to picture, multiple picture fragments are effective
Represent the characteristic of picture to be detected, by the multiple picture fragment be input to one by one in convolutional neural networks carry out feature mention
It takes to obtain the picture feature for representing each picture fragment, the picture feature based on extraction is determined using trained disaggregated model
Whether the picture to be detected is reproduction picture, and not needing to carry out whole picture analysis can be completed, and can be greatly improved point
Analyse speed；And determine whether the picture to be detected is reproduction, i.e., only needs based on the sum for the picture fragment for being determined as reproduction
The threshold value that sum reaches setting is considered that reproduction picture, the remaining picture fragment for not determining whether reproduction do not need again
Continue, also substantially increase analysis speed to a certain extent, manpower can not only be saved and reduce cost, and does not need to mention
Picture is compared for reference, can improve well detection efficiency and be applicable in scene flexibility it is higher.
The picture to be processed, which is carried out gray processing processing, can reduce the complexity handled picture to be detected, and set
One rectangle frame is moved using the rectangle frame according to default rule, so that the picture fragment size that acquisition obtains is consistent,
The picture fragment information of acquisition is stablized effectively, to improve the accuracy of analysis.
Between two fragments of arbitrary neighborhood with overlapping region, arbitrary neighborhood two picture fragments edge connection or
Be spaced certain spacing between two fragments of person's arbitrary neighborhood so that acquisition obtain picture fragment have the effect of it is different,
User can be very good the quality condition according to picture to be detected, select suitable rectangle frame movement rule, relatively reliable to obtain
Analysis as a result, for example, when according to artificial experience determine picture be reproduction probability it is not high when, two picture fragments can be made
Between there are a fixed spacings, both can guarantee accuracy in this way, can also improve efficiency well；It is understood that ought artificially pass through
Test determine picture be reproduction probability it is relatively high when, then make between two picture fragments of arbitrary neighborhood exist have overlay region
Domain, arbitrary neighborhood two picture fragments edge connection, improve to the sampling area of picture to be detected, to improve the standard of detection
True property.
The picture to be detected is zoomed in and out according to preset ratio, can be selected very well according to the quality of picture to be detected
Suitable scaling zooms in and out, so that more significant to the signature analysis of picture to be detected.
Picture reproduction detection system provided by the invention and electronic equipment have and the picture reproduction detection method phase
Same beneficial effect.
The foregoing is merely present pre-ferred embodiments, are not intended to limit the invention, it is all principle of the present invention it
Any modification made by interior, equivalent replacement and improvement etc. should all be comprising within protection scope of the present invention.
Claims (10)
1. a kind of picture reproduction detection method, which comprises the steps of:
S1, picture to be detected is provided；
S2, sliding window sampling is carried out to the picture to be detected, to obtain multiple picture fragments；
S3, it the multiple picture fragment is input in convolutional neural networks carries out feature extraction to represent each picture broken to obtain
The picture feature of piece；And
S4, determine whether each picture fragment is to turn over respectively using trained disaggregated model based on the picture feature of extraction
It claps；And
S5, determine whether the picture to be detected is reproduction according to the sum for the picture fragment for being determined as reproduction in the step S4
Picture.
2. picture reproduction detection method as described in claim 1, it is characterised in that: the step S2 specifically includes following step
It is rapid:
Step S21: the picture to be detected is subjected to gray processing processing, obtains the gray level image of picture to be detected；And
Step S22: one rectangle frame of setting is moved the rectangle frame on the picture to be detected by setting rule, according to
It is secondary that picture to be detected is obtained into multiple picture fragments according to the size acquisition of the rectangle frame.
3. picture reproduction detection method as claimed in claim 3, it is characterised in that: the size of each picture fragment is identical, arbitrarily
With overlapping region, the edge connection of two picture fragments of arbitrary neighborhood or arbitrary neighborhood between two adjacent fragments
Certain spacing is spaced between two fragments.
4. picture reproduction detection method as described in claim 1, it is characterised in that: the picture extracted in the step S3
Feature includes at least one of spectrum signature, textural characteristics, color characteristic and edge feature.
5. picture reproduction detection method as described in claim 1, it is characterised in that: the picture reproduction detection method further includes
Step S10, the picture to be detected is zoomed in and out according to preset ratio, the step S10 is in the step S1 and described
Between step S2.
6. picture reproduction detection method as described in claim 1, it is characterised in that: the training of the disaggregated model includes:
Picture database to be detected is constructed, the picture database to be detected includes that one group of picture to be detected and every are to be detected
The corresponding label of picture, the label indicate whether picture to be detected is reproduction picture；
Calculate the picture feature of each picture to be detected in the picture database to be detected；
Training set is constructed based on the label and the picture feature；And
The disaggregated model is trained on the training set using supporting vector machine model.
7. the picture reproduction detection method stated such as claim 5, it is characterised in that: the textural characteristics of acquisition include picture to be measured
Bag of words, the color characteristic of acquisition include color histogram.
8. a kind of picture reproduction detection system characterized by comprising
Sampling module, for carrying out sliding window sampling to the picture to be detected, to obtain multiple picture fragments；
Characteristic extracting module carries out feature extraction for the multiple picture fragment to be input in convolutional neural networks to obtain generation
The picture feature of each picture fragment of table；
Picture classification module determines the picture to be detected using trained disaggregated model for the picture feature based on extraction
It whether is reproduction picture；
First determination module, be configured to the picture feature based on extraction determined respectively using trained disaggregated model it is described every
Whether a picture fragment is reproduction；
Second determination module, be configured to be determined as according to the first determination module the picture fragment of reproduction sum determine it is described to
Detect whether picture is reproduction picture.
9. picture reproduction detection system as claimed in claim 8, which is characterized in that further include:
Zoom module is configured to zoom in and out the picture to be detected of offer according to preset ratio.
10. a kind of electronic equipment, including memory and processor, it is characterised in that: be stored with computer journey in the memory
Sequence, the computer program are arranged to execute the detection of picture reproduction described in any one of claim 1 to 8 when operation
Method；
The processor is arranged to execute figure described in any one of claim 1 to 8 by the computer program
Piece reproduction detection method.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910613965.0A CN110348511A (en) | 2019-07-08 | 2019-07-08 | A kind of picture reproduction detection method, system and electronic equipment |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910613965.0A CN110348511A (en) | 2019-07-08 | 2019-07-08 | A kind of picture reproduction detection method, system and electronic equipment |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN110348511A true CN110348511A (en) | 2019-10-18 |
Family
ID=68178657
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910613965.0A Withdrawn CN110348511A (en) | 2019-07-08 | 2019-07-08 | A kind of picture reproduction detection method, system and electronic equipment |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110348511A (en) |
Cited By (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110991458A (en) * | 2019-11-25 | 2020-04-10 | 创新奇智(北京)科技有限公司 | Artificial intelligence recognition result sampling system and sampling method based on image characteristics |
| CN111062922A (en) * | 2019-12-14 | 2020-04-24 | 创新奇智(北京)科技有限公司 | Method and system for judging copied image and electronic equipment |
| CN111127398A (en) * | 2019-11-25 | 2020-05-08 | 深圳前海微众银行股份有限公司 | Method and device for detecting certificate photo duplication, terminal equipment and storage medium |
| CN111144425A (en) * | 2019-12-27 | 2020-05-12 | 五八有限公司 | Method and device for detecting screen shot picture, electronic equipment and storage medium |
| CN111524108A (en) * | 2020-04-15 | 2020-08-11 | 四川赛康智能科技股份有限公司 | Transformer substation equipment detection method and equipment |
| CN112036435A (en) * | 2020-07-22 | 2020-12-04 | 温州大学 | Brushless direct current motor sensor fault detection method based on convolutional neural network |
| CN112258481A (en) * | 2020-10-23 | 2021-01-22 | 北京云杉世界信息技术有限公司 | Portal photo reproduction detection method |
| CN113542142A (en) * | 2020-04-14 | 2021-10-22 | 中国移动通信集团浙江有限公司 | Portrait anti-counterfeiting detection method and device and computing equipment |
| CN113570001A (en) * | 2021-09-22 | 2021-10-29 | 深圳市信润富联数字科技有限公司 | Classification identification positioning method, device, equipment and computer readable storage medium |
| CN114066894A (en) * | 2022-01-17 | 2022-02-18 | 深圳爱莫科技有限公司 | Detection method for display image reproduction, storage medium and processing equipment |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105118048A (en) * | 2015-07-17 | 2015-12-02 | 北京旷视科技有限公司 | Method and device for identifying copying certificate image |
| CN108171689A (en) * | 2017-12-21 | 2018-06-15 | 深圳大学 | A kind of identification method, device and the storage medium of the reproduction of indicator screen image |
| CN109558794A (en) * | 2018-10-17 | 2019-04-02 | 平安科技（深圳）有限公司 | Image-recognizing method, device, equipment and storage medium based on moire fringes |
| CN109670430A (en) * | 2018-12-11 | 2019-04-23 | 浙江大学 | A kind of face vivo identification method of the multiple Classifiers Combination based on deep learning |
| CN109859227A (en) * | 2019-01-17 | 2019-06-07 | 平安科技（深圳）有限公司 | Reproduction image detecting method, device, computer equipment and storage medium |
| CN109886275A (en) * | 2019-01-16 | 2019-06-14 | 深圳壹账通智能科技有限公司 | Reproduction image-recognizing method, device, computer equipment and storage medium |
-
2019
- 2019-07-08 CN CN201910613965.0A patent/CN110348511A/en not_active Withdrawn
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105118048A (en) * | 2015-07-17 | 2015-12-02 | 北京旷视科技有限公司 | Method and device for identifying copying certificate image |
| CN108171689A (en) * | 2017-12-21 | 2018-06-15 | 深圳大学 | A kind of identification method, device and the storage medium of the reproduction of indicator screen image |
| CN109558794A (en) * | 2018-10-17 | 2019-04-02 | 平安科技（深圳）有限公司 | Image-recognizing method, device, equipment and storage medium based on moire fringes |
| CN109670430A (en) * | 2018-12-11 | 2019-04-23 | 浙江大学 | A kind of face vivo identification method of the multiple Classifiers Combination based on deep learning |
| CN109886275A (en) * | 2019-01-16 | 2019-06-14 | 深圳壹账通智能科技有限公司 | Reproduction image-recognizing method, device, computer equipment and storage medium |
| CN109859227A (en) * | 2019-01-17 | 2019-06-07 | 平安科技（深圳）有限公司 | Reproduction image detecting method, device, computer equipment and storage medium |
Cited By (17)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN110991458B (en) * | 2019-11-25 | 2023-05-23 | 创新奇智(北京)科技有限公司 | Image feature-based artificial intelligent recognition result sampling system and sampling method |
| CN111127398A (en) * | 2019-11-25 | 2020-05-08 | 深圳前海微众银行股份有限公司 | Method and device for detecting certificate photo duplication, terminal equipment and storage medium |
| CN110991458A (en) * | 2019-11-25 | 2020-04-10 | 创新奇智(北京)科技有限公司 | Artificial intelligence recognition result sampling system and sampling method based on image characteristics |
| CN111127398B (en) * | 2019-11-25 | 2023-09-26 | 深圳前海微众银行股份有限公司 | Certificate photo copying detection method and device, terminal equipment and storage medium |
| CN111062922A (en) * | 2019-12-14 | 2020-04-24 | 创新奇智(北京)科技有限公司 | Method and system for judging copied image and electronic equipment |
| CN111062922B (en) * | 2019-12-14 | 2024-02-20 | 创新奇智(北京)科技有限公司 | Method and system for distinguishing flip image and electronic equipment |
| CN111144425A (en) * | 2019-12-27 | 2020-05-12 | 五八有限公司 | Method and device for detecting screen shot picture, electronic equipment and storage medium |
| CN111144425B (en) * | 2019-12-27 | 2024-02-23 | 五八有限公司 | Method and device for detecting shot screen picture, electronic equipment and storage medium |
| CN113542142B (en) * | 2020-04-14 | 2024-03-22 | 中国移动通信集团浙江有限公司 | Portrait anti-fake detection method and device and computing equipment |
| CN113542142A (en) * | 2020-04-14 | 2021-10-22 | 中国移动通信集团浙江有限公司 | Portrait anti-counterfeiting detection method and device and computing equipment |
| CN111524108A (en) * | 2020-04-15 | 2020-08-11 | 四川赛康智能科技股份有限公司 | Transformer substation equipment detection method and equipment |
| CN112036435B (en) * | 2020-07-22 | 2024-01-09 | 温州大学 | Brushless direct current motor sensor fault detection method based on convolutional neural network |
| CN112036435A (en) * | 2020-07-22 | 2020-12-04 | 温州大学 | Brushless direct current motor sensor fault detection method based on convolutional neural network |
| CN112258481A (en) * | 2020-10-23 | 2021-01-22 | 北京云杉世界信息技术有限公司 | Portal photo reproduction detection method |
| CN113570001B (en) * | 2021-09-22 | 2022-02-15 | 深圳市信润富联数字科技有限公司 | Classification identification positioning method, device, equipment and computer readable storage medium |
| CN113570001A (en) * | 2021-09-22 | 2021-10-29 | 深圳市信润富联数字科技有限公司 | Classification identification positioning method, device, equipment and computer readable storage medium |
| CN114066894A (en) * | 2022-01-17 | 2022-02-18 | 深圳爱莫科技有限公司 | Detection method for display image reproduction, storage medium and processing equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110348511A (en) | A kind of picture reproduction detection method, system and electronic equipment | |
| Asghar et al. | Copy-move and splicing image forgery detection and localization techniques: a review | |
| Walia et al. | Digital image forgery detection: a systematic scrutiny | |
| CN105118048B (en) | The recognition methods of reproduction certificate picture and device | |
| CN106503703B (en) | System and method for recognizing credit card number and expiration date using terminal device | |
| US10878173B2 (en) | Object recognition and tagging based on fusion deep learning models | |
| CN107911753A (en) | Method and apparatus for adding digital watermarking in video | |
| US11455831B2 (en) | Method and apparatus for face classification | |
| Li et al. | Automatic comic page segmentation based on polygon detection | |
| CN108171208A (en) | Information acquisition method and device | |
| CN113221918A (en) | Target detection method, and training method and device of target detection model | |
| CN116543261A (en) | Model training method for image recognition, image recognition method device and medium | |
| WO2020071558A1 (en) | Business form layout analysis device, and analysis program and analysis method therefor | |
| Kim et al. | Generalized facial manipulation detection with edge region feature extraction | |
| Guha et al. | Computer Vision Based Automatic Margin Computation Model for Digital Document Images | |
| CN109064464A (en) | Method and apparatus for detecting battery pole piece burr | |
| Hung et al. | Automatic vietnamese passport recognition on android phones | |
| Ma et al. | A recognition method of hand gesture with CNN-SVM model | |
| CN112287905A (en) | Vehicle damage identification method, device, equipment and storage medium | |
| CN115273123B (en) | Bill identification method, device and equipment and computer storage medium | |
| CN116385957A (en) | X-ray image contraband detection method, device, equipment and medium | |
| CN115546554A (en) | Sensitive image identification method, device, equipment and computer readable storage medium | |
| CN114708420A (en) | Visual positioning method and device based on local variance and posterior probability classifier | |
| CN114638304A (en) | Training method of image recognition model, image recognition method and device | |
| CN110853115B (en) | Creation method and device of development flow page |
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 | ||
| WW01 | Invention patent application withdrawn after publication | ||
| WW01 | Invention patent application withdrawn after publication |
Application publication date: 20191018 |