CN1384467A - Micro image characteristic extracting and recognizing method - Google Patents
Micro image characteristic extracting and recognizing method Download PDFInfo
- Publication number
- CN1384467A CN1384467A CN 02120886 CN02120886A CN1384467A CN 1384467 A CN1384467 A CN 1384467A CN 02120886 CN02120886 CN 02120886 CN 02120886 A CN02120886 A CN 02120886A CN 1384467 A CN1384467 A CN 1384467A
- Authority
- CN
- China
- Prior art keywords
- image
- feature
- micro
- identification
- feature extraction
- 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
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000000605 extraction Methods 0.000 claims abstract description 39
- 238000012545 processing Methods 0.000 claims abstract description 12
- 238000013519 translation Methods 0.000 claims description 11
- 238000004088 simulation Methods 0.000 claims description 6
- 230000003321 amplification Effects 0.000 claims description 3
- 239000000284 extract Substances 0.000 claims description 3
- 238000003199 nucleic acid amplification method Methods 0.000 claims description 3
- 238000010606 normalization Methods 0.000 claims description 2
- 238000003672 processing method Methods 0.000 claims 1
- 238000005516 engineering process Methods 0.000 abstract description 2
- 238000001514 detection method Methods 0.000 description 9
- 239000000523 sample Substances 0.000 description 5
- 238000006243 chemical reaction Methods 0.000 description 4
- 238000003708 edge detection Methods 0.000 description 4
- 238000004321 preservation Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 230000000052 comparative effect Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000002474 experimental method Methods 0.000 description 3
- 239000004744 fabric Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000006116 polymerization reaction Methods 0.000 description 2
- 238000011160 research Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000008878 coupling Effects 0.000 description 1
- 238000010168 coupling process Methods 0.000 description 1
- 238000005859 coupling reaction Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 238000009499 grossing Methods 0.000 description 1
- 239000000463 material Substances 0.000 description 1
- 230000003278 mimic effect Effects 0.000 description 1
- 210000005036 nerve Anatomy 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 239000010409 thin film Substances 0.000 description 1
Images
Landscapes
- Image Analysis (AREA)
- Character Discrimination (AREA)
Abstract
The present invention relates to image processing technology, and the micro image characteristic extracting and recognizing method includes two parts, the extraction of micro image characteristic and the recognition of characteristics. The extraction includes the micro enlargement of image, digitizing the analog micro image, binarization processing, characterizing the position information of digital image with normalized coordinates and information re-processing and template preserving. The recognition includes micro enlargement of the image to be recognized, digitizing the image, edge information extraction of the digitized image and the comparison between the edge information and the stored characteristic template. Maximum error limit criterion is used in judging the truth of image. The present invention is especially suitable for recognition and anti-fake of volualbe security and document.
Description
Technical field
The invention belongs to technical field of image processing, particularly the feature extraction of micro image and feature identification.
Background technology
Micro image all is widely used at aspects such as the microscopic feature analysis of remote sensing, the research of reservoir fractal characteristic, optical element surface micro-profile The Characteristics, Nano thin film interfacial structure, material tests.At different micro image feature and application purpose, the method for feature extraction and identification is also inequality, does not have unified pattern.With a kind of fingerprint identification method based on architectural feature is example, and key diagram is as Feature Extraction and identification.The feature extraction of this fingerprint identification method may further comprise the steps: gather fingerprint image and convert to digital picture input computing machine, based on the binaryzation of edge search, to the image of binaryzation carry out smoothing processing, refinement, the feature that takes the fingerprint (comprising features such as limit, bifurcated, the shape of the mouth as one speaks, thorn, cross, bridge type).The identification of the coupling of fingerprint comprises the feature of extracting fingerprint to be measured, compares, exports recognition result with the characteristic of preservation.When identification, to consider the influence of angle rotation and distortion etc.
The common methods of relevant Image Edge-Detection has: method of differential operator; The model matching method; Border and curve enhancement techniques, as the iteration technique frontier probe, the relaxation method border strengthens, or utilizes parameter space, and edge pixel is done polymerization according to border or parameter of curve; The continuous wavelet rim detection; Edge focusing; The texture rim detection; Neural network rim detection etc.
Method of differential operator is to utilize multi-form differentiating operator, calculates the space derivative of each pixel gray scale, provides differential sharpening image.As gradient operator:
G (i, j)=Δ
xF (i, j)
2+ Δ
yF (i, j)
2(1) Laplacian operator:
G(i，j)＝Δ
2 xf(i，j)+Δ
2 yf(i，j) (2)
=f (i+1, j)+f (i-1, j)+f (i, j+1)-f (i, j-1) wherein f (i j) is point (i, the gray-scale value of j) locating.Method of differential operator mainly is at gray level image, the projecting edge feature, though the image edge after the processing has strengthened, the gray level at each edge is disunity still, and the interference of background still exists, and is difficult to discern automatically the border.Method of differential operator is difficult to determine the position on border, though the image human eye after handling looks that the border is apparent in view, the Computer Automatic Recognition border is difficulty relatively.
The model matching method utilizes desirable small fringe region to constitute the edge model, and the matching degree of each pixel and model provides edge image in the detection image.As the direction model, can detect the edge on the different directions.But, can not make model in advance for edge image with random character.
Border and curve enhancement techniques utilize the neighborhood information of pixel that pixel is done enhancing on the basis of edge detection, or utilize parameter space that edge pixel is done polymerization according to border or parameter of curve, image after handling like this has with the similar feature of differentiating operator result, and marginal position is difficult to automatic identification.Continuous wavelet rim detection and edge focusing also have similar weakness.
The texture edge detection method requires object to have different textures with background, therefore also is not suitable for directly applying to random image.As for the neural network edge detection method, though existing at present many algorithms can be converted into neural network and realize, but do not reflect the essence of nerve network system, the characteristic detection method of actual configuration mimic biology vision system is still waiting further research, and this respect does not have ready-made effective method so far yet.
Summary of the invention
The objective of the invention is microcosmic random character characteristic, propose a kind of method of feature extraction and identification of micro image according to image, have realize simple, edge contour is clear and the identification reliable characteristics, is applicable to feature extraction and identification to micro image.Also can be applicable to common computer picture recognition, be particularly useful for to the true or false of securities and file distinguish and false proof.
The present invention proposes a kind of method of feature extraction and identification of micro image, and it comprises micro image Feature Extraction and Feature Recognition two parts, it is characterized in that: the feature extraction of said micro image may further comprise the steps;
1) specific image amplified become the microcosmic Simulation image,
2) this micro image is carried out digitized processing and becomes digitized image,
3) this digitized image binaryzation is become binary image,
4) this binary image is carried out the feature extraction of marginal position,
5) and to this feature carry out normalized and preserve the feature model;
The identification of said micro image comprises and may further comprise the steps:
1) image to be identified is carried out microcosmic amplification becoming microcosmic Simulation image,
2) this micro image is carried out digitized processing and becomes digitized image,
3) this digitized image binaryzation is become binary image,
4) this binary image is carried out the feature extraction of marginal position,
5) and with the feature model of this feature and storage compare, judge the true and false of this image to be identified.
Principle of the present invention and characteristics:
Main contribution of the present invention is a microcosmic random character characteristic of having found image, and with the basis of this characteristic as feature extraction of the present invention and characteristic recognition method, makes its method simple and reliable.
The image that the present invention relates to is meant with duplicating or the non-means of duplicating are formed on all lines, point, writings and image on the image-carrier (for example paper).Find that after deliberation edge on edge of image or internal feature, especially the image microcosmic or internal feature present a kind of character at random.As shown in Figure 1, Fig. 1 (a) is the macrograph of two width same straight line 1,2 that is 0.3mm, and Fig. 1 (b) is the microcosmic enlarged drawing of 11,21 parts, two straight line left parts, as can be seen, the end profile shape of two figure is different fully under microstate, and both edge degrees of correlation are very low.Can find out that also the distribution of image border point is at random, not have regularity to say.In addition, these lines are continuous on macroscopic view, but on microcosmic but are not, a lot of discontinuity zones is arranged in the lines, and its distribution also is at random.
Because the distribution of these marginal points and inner discontinuity zone all is at random, the such image of any two width of cloth, the possibility of its coincident is very low, and what interior zone was identical may be also very low.The rule of this image border point stochastic distribution is unique, and any two width of cloth images are all inequality.The feature that this randomness of micro image edge contour and uniqueness can be used as image, this feature that the present invention just is being based on image is used as the method basis to image recognition.
The extracting method of characteristics of image of the present invention mainly comprises generation and the preservation to the detection of image border profile and feature extraction and feature model, Feature Recognition then at first adopts the step identical with characteristic extraction procedure to extract the feature that is identified image, then this feature and the feature model of being preserved are compared, provide recognition result.
Description of drawings
Fig. 1 is the micro image of two same straight line and end points thereof; Wherein, (a) being macrograph, (b) is local microcosmic enlarged drawing.
Fig. 2 is an image characteristics extraction process flow diagram of the present invention.
Fig. 3 is characteristics of image identification process figure of the present invention.
Fig. 4 is the result of the micro image binary conversion treatment among Fig. 1.
Fig. 5 is the extracting method and the result of marginal position information.
Fig. 6 is the image rotating key diagram.
Embodiment
Below by a preferred embodiment of the invention and describe content of the present invention in conjunction with the accompanying drawings in detail and realize principle:
The feature extraction of a kind of micro image that the present invention proposes and the method for identification, it comprises micro image Feature Extraction and Feature Recognition two parts, as shown in Figure 3, wherein, image characteristics extraction may further comprise the steps:
1) specific image amplified become the microcosmic Simulation image,
2) this micro image is carried out digitized processing and becomes digitized image,
3) this digitized image binaryzation is become binary image,
4) this binary image is carried out the feature extraction of marginal position,
5) and to this feature carry out normalized and preserve the feature model;
Characteristics of image identification may further comprise the steps:
6) image to be identified is carried out microcosmic amplification becoming microcosmic Simulation image,
7) this micro image is carried out digitized processing and becomes digitized image,
8) this digitized image binaryzation is become binary image,
9) this binary image is carried out the feature extraction of marginal position,
10) read template data;
11) with the positional information and the directly relatively identification of template data of image to be identified, recognition result is that " very " then finishes;
12) if recognition result is " puppet ", then the positional information and the template data of image to be identified are carried out translation relatively,
Recognition result is that " very " then finishes;
13) if recognition result is " puppet ", then positional information and the template data with image to be identified is rotated comparison,
Obtain recognition result;
14) output recognition result.
So that two straight lines among Fig. 1 are compared as a kind of embodiment each step of feature extraction of the present invention and feature identification is elaborated below.One, feature extraction
Straight line 1 is zoomed into micro image by optical system, this micro image collection is become gray-scale image 11 with image capture device; This gray-scale image is input in the computing machine, it is carried out binary conversion treatment, this gray-scale image is transformed to black and white two-value (0 and 1) image according to a predetermined threshold value.
The present invention proposes a kind of selection of threshold method, as the formula (3),
p
T=p
Min+ α (p
Max-p
Min) (3) p wherein
TBe the threshold value of choosing; p
MinBe minimum gradation value in the image; p
MaxBe maximum gradation value; α is a constant, can determine according to experimental result.The principle that α chooses is: the image after the binaryzation can separate useful information and background, does not lose marginal information.A certain class image is done experiment, change the size of α, choose a suitable numerical value, make it reach above-mentioned requirements according to the result of binaryzation.The threshold value that this method is determined has the advantages that to regulate threshold size automatically with the light intensity variation.
Fig. 4 is the image after Fig. 1 (b) utilizes formula (3) binary conversion treatment, can see that two rectilinears after the binaryzation are obvious as 13,23 marginal information, and both edge degrees of correlation is very low.
After being made binary conversion treatment, image just can carry out the random character information extraction of marginal position to this binary image.Concrete grammar is: the every row along binary image is searched for from left to right, stop search when gray-scale value is 0 point when running into, and with this point as marginal point, (the image upper left corner is the coordinate axis initial point to write down its positional information xi, x is axially right for just, y is axially down for just), as the positional information of extraction; This method is actually that the leftmost stain of every row is used as is marginal point.
To different lines, because its width difference, the edge is counted also just different, if the positional information of being had a few of left hand edge is all extracted preservation, the data number of preserving is just inequality, is not easy to comparison and unification like this, and the starting point up and down at searching edge is relatively more difficult.So in this preferred embodiment, the information of a part is preserved in the middle of the intercepting left hand edge, promptly extracts ordinate between y
aAnd y
bBetween marginal information.Shown in Fig. 5 (a), the ordinate of straight line A and B position is respectively y
aAnd y
b, from y
aRow is to y
bOK, search for from left to right respectively, obtain the marginal position coordinate x of every row
a..., x
i... x
bFig. 5 (b) is according to coordinate figure x
a..., x
i... x
bThe edge contour figure that draws.
Then above-mentioned coordinate figure is carried out normalized, method is: compare the size of these edge point position coordinates, find out minimum value x wherein
n, with each coordinate figure x
iAll cut this minimum value, obtain new data x
a-x
n..., x
i-x
n... x
b-x
n, the feature model when preserving these new data as identification.Two, feature identification,
During identification, at first use the step identical with above-mentioned feature extraction to extract the random character information of image to be tested, the feature model with prior preservation compares then.
Whether consistent, if coincide in the error allowed band, recognition result is " very " just, otherwise is " puppet " if relatively discerning the edge point position that is actually comparison present image and the original image of having preserved.
When relatively discerning, because the influence of front and back environment etc., all data can not overlap fully, so will determine a maximum error limit according to concrete requirement and experimental result.As long as recognition result meets this requirement, just identification is passed through, otherwise recognition failures.This maximum error limit should satisfy two requirements: erroneous judgement one, can not occur, be true with the image recognition of vacation; Two, discrimination height, promptly genuine image recognition are very low even be zero for pseudo-probability.
The identifying of present embodiment is described in detail as follows:
Among Fig. 5, the y axial coordinate value of A and B is respectively y
aAnd y
bTherefore, the marginal point of gathering adds up to (y
b-y
a+ 1).
When carrying out image recognition, at first extract the left hand edge feature (y of sample straight line 2 to be identified according to the step identical with image characteristics extraction
b-y
a+ 1 characteristic), carry out the normalization coordinate then and handle the position influence that to avoid the sample move left and right to bring like this.Then read the template data of having preserved, the criterion (formula (4) and formula (5)) according to the maximum error limit compares identification then.
Though the image border that newly obtains is similar with former edge shape,, therefore when identification, stipulate a maximum deflection difference value x because the influence of acquisition condition etc. can not overlap fully with original edge
iAs long as each marginal point satisfies formula (4), just think that the edge overlaps.
| x
s-x
m|≤x
i(4) x wherein
sBe the feature template data of preserving; x
mThe data that arrive for current detection; x
iBe the maximum deviation that allows.x
iThe principle of choosing should be able to prevent erroneous judgement, can identify real object again.In a preferred embodiment, x
iValue be 3.According to the quality and the lot of experiment results of image, x
iCan do small fluctuation, key is under the prerequisite that prevents to judge by accident, improves discrimination as far as possible.
As long as last comparative result satisfies formula (5), just think that its edge overlaps, recognition result is " very ".
N
c〉=N
i* p (5) is N wherein
cSum for the marginal point that satisfies formula (4); N
iBe total data number (y
b-y
a+ 1); The accuracy rate of p for needing to satisfy, in a preferred embodiment, the value of p is 90%.The value of p is big more, and False Rate is low more, but can reduce discrimination; The value of p is more little, and discrimination is high more, but erroneous judgement might occur.Should determine the p value of a certain class sample according to a large amount of experiments, under the prerequisite that does not occur judging by accident, improve discrimination.
When image characteristics extraction and feature identification, the position that sample is put always has up and down the deviation with angle, therefore will do rotation and translation to image.
If directly Bi Jiao result is " very ", finish comparison procedure; If directly Bi Jiao result is " puppet ", then carry out translation relatively.
Detection obtains the y coordinate from y
A-nTo y
B+n, y altogether
b-y
aThe positional information x of+2n+1 marginal point
A-n..., x
a..., x
b..., x
B+n, wherein n is an integer, is the amount that need move up and down.Relatively the time, therefrom choose y successively
b-y
a+ 1 numerical value compares with the feature model of preserving.Like this, total 2n+1 group data, they are respectively: x
A-n～x
B-n, x
A-n+1～x
B-n+1..., x
a～x
b..., x
A+n-1～x
B+n-1And x
A+n～x
B+n, it is from y that each group data is all regarded the y coordinate as
aTo y
b, compare with the feature templates data.As long as wherein there is one group of data to satisfy formula (5), just think that the edge is overlapping, recognition result is " very ".More in fact translation has comprised the process of direct comparison.
If translation result relatively is " very ", finish comparison procedure; Otherwise be rotated comparison.
Speed ratio is earlier entire image to be rotated an angle, and then carries out translation relatively.
As shown in Figure 6, A point and the coordinate of B point in the same coordinate system be respectively (x, y) and (x
0, y
0), OA is with respect to angle θ of OB rotation, and then the coordinate of the two has following relational expression:
x＝x
0cosθ-y
0sinθ
y＝x
0sinθ+y
0cosθ (6)
When image rotating, at first determine an angle step α, then with entire image according to formula (6) anglec of rotation α or-α, next carry out translation relatively, if comparative result be " very ", the end comparison; If recognition result is " puppet ", again original image is revolved shape 2 α and-2 α, carry out translation relatively.If comparative result is " very ", finish relatively; If recognition result be " puppet ", again original image is rotated 3 α and-3 α ... n α and-n α, and carry out the translation comparison respectively.If to the anglec of rotation be n α and-during n α, still discern successfully, export recognition result " puppet ", the end identifying.
Claims (4)
1, the method for a kind of feature extraction of micro image and identification, it comprises micro image Feature Extraction and Feature Recognition two parts, it is characterized in that: the feature extraction of said micro image may further comprise the steps;
1) specific image amplified become the microcosmic Simulation image,
2) this micro image is carried out digitized processing and becomes digitized image,
3) this digitized image binaryzation is become binary image,
4) this binary image is carried out the feature extraction of marginal position,
5) and to this feature carry out normalized and preserve the feature model;
The identification of said micro image comprises and may further comprise the steps:
6) image to be identified is carried out microcosmic amplification becoming microcosmic Simulation image,
7) this micro image is carried out digitized processing and becomes digitized image,
8) this digitized image binaryzation is become binary image,
9) this binary image is carried out the feature extraction of marginal position,
10) and with the feature model of this feature and storage compare, judge the true and false of this image to be identified.
2, the method for micro image feature extraction as claimed in claim 1 and identification, it is characterized in that: the feature extracting methods of said marginal position, specifically comprise: the every row along binary image is searched for from left to right, stop search when gray-scale value is 0 point when running into, and with this point as marginal point, write down the coordinate information x of its position
i, as the positional information of extracting; Choose set line number positional information as the feature model wherein, the image upper left corner is the coordinate axis initial point, x is axially right for just, y axially under for just.
3, the method for micro image feature extraction as claimed in claim 1 and identification, it is characterized in that: said normalization processing method is: extract the location coordinate information x that sets line number
i, relatively more all coordinate information x
iSize, find out minimum value x wherein
n, with each coordinate figure x
iAll cut this minimum value, obtain new data x
i-x
nFeature model when preserving these new data as identification.
4, the method for micro image feature extraction as claimed in claim 1 and identification is characterized in that: said feature and the method that compares of the feature model of storage with image to be identified may further comprise the steps;
1) determines maximum error limit criterion;
2) at first the feature of image to be identified and the feature model of storage are directly compared, if it satisfies said
Mistake limit criterion, recognition result is that " very " then finishes;
3) if recognition result is " puppet ", then make the feature relative translation of image to be identified, compare recognition result again
For " very " then finishes;
4) if recognition result is " puppet ", the feature of image to be identified is rotated relatively, carry out the translation of step 3) again
Relatively, obtain recognition result.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB021208867A CN1141665C (en) | 2002-06-07 | 2002-06-07 | Micro image characteristic extracting and recognizing method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB021208867A CN1141665C (en) | 2002-06-07 | 2002-06-07 | Micro image characteristic extracting and recognizing method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1384467A true CN1384467A (en) | 2002-12-11 |
| CN1141665C CN1141665C (en) | 2004-03-10 |
Family
ID=4744838
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB021208867A Expired - Fee Related CN1141665C (en) | 2002-06-07 | 2002-06-07 | Micro image characteristic extracting and recognizing method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN1141665C (en) |
Cited By (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100405405C (en) * | 2005-12-29 | 2008-07-23 | 兆日科技（深圳）有限公司 | Fiber image antifake method |
| CN101976336A (en) * | 2010-10-21 | 2011-02-16 | 西北工业大学 | Fuzzy enhancement and surface fitting-based image edge characteristic extraction method |
| CN102034108A (en) * | 2010-12-06 | 2011-04-27 | 哈尔滨工业大学 | Multi-resolution network characteristic registration-based method for sorting face values and face directions of notes in sorter |
| CN102214305A (en) * | 2011-04-08 | 2011-10-12 | 大连理工大学 | Method for taking evidence for source of printing paper sheet by using grain characteristic |
| WO2015089859A1 (en) * | 2013-12-20 | 2015-06-25 | 深圳市华星光电技术有限公司 | Boundary acquisition method and detection method for alignment film |
| CN104903775A (en) * | 2013-01-04 | 2015-09-09 | Lg电子株式会社 | Head mounted display and method for controlling the same |
| WO2016177301A1 (en) * | 2015-05-01 | 2016-11-10 | 励元科技(上海)有限公司 | Micro-texture anti-counterfeit method |
| WO2016177253A1 (en) * | 2015-05-01 | 2016-11-10 | 励元科技(上海)有限公司 | Ink edge anti-counterfeiting method and ink edge anti-counterfeiting network identification system |
| CN106250932A (en) * | 2016-08-04 | 2016-12-21 | 上海华虹宏力半导体制造有限公司 | A kind of method and device of scanogram identification |
| CN109409158A (en) * | 2018-09-29 | 2019-03-01 | 武汉保诚信网络科技有限公司 | A kind of method for anti-counterfeit based on two dimensional code edge roughness |
| CN111504608A (en) * | 2019-01-31 | 2020-08-07 | 中强光电股份有限公司 | Brightness uniformity detection system and brightness uniformity detection method |
-
2002
- 2002-06-07 CN CNB021208867A patent/CN1141665C/en not_active Expired - Fee Related
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100405405C (en) * | 2005-12-29 | 2008-07-23 | 兆日科技（深圳）有限公司 | Fiber image antifake method |
| CN101976336A (en) * | 2010-10-21 | 2011-02-16 | 西北工业大学 | Fuzzy enhancement and surface fitting-based image edge characteristic extraction method |
| CN102034108A (en) * | 2010-12-06 | 2011-04-27 | 哈尔滨工业大学 | Multi-resolution network characteristic registration-based method for sorting face values and face directions of notes in sorter |
| CN102214305A (en) * | 2011-04-08 | 2011-10-12 | 大连理工大学 | Method for taking evidence for source of printing paper sheet by using grain characteristic |
| CN104903775B (en) * | 2013-01-04 | 2018-03-27 | Lg电子株式会社 | Head mounted display and its control method |
| CN104903775A (en) * | 2013-01-04 | 2015-09-09 | Lg电子株式会社 | Head mounted display and method for controlling the same |
| WO2015089859A1 (en) * | 2013-12-20 | 2015-06-25 | 深圳市华星光电技术有限公司 | Boundary acquisition method and detection method for alignment film |
| WO2016177301A1 (en) * | 2015-05-01 | 2016-11-10 | 励元科技(上海)有限公司 | Micro-texture anti-counterfeit method |
| WO2016177253A1 (en) * | 2015-05-01 | 2016-11-10 | 励元科技(上海)有限公司 | Ink edge anti-counterfeiting method and ink edge anti-counterfeiting network identification system |
| RU2703113C2 (en) * | 2015-05-01 | 2019-10-15 | Хаинан Яюан Анти-Цоунтерфеит Теч Рес Инст | Ink edge anti-counterfeiting method and ink edge anti-counterfeiting network identification system |
| CN106250932A (en) * | 2016-08-04 | 2016-12-21 | 上海华虹宏力半导体制造有限公司 | A kind of method and device of scanogram identification |
| CN109409158A (en) * | 2018-09-29 | 2019-03-01 | 武汉保诚信网络科技有限公司 | A kind of method for anti-counterfeit based on two dimensional code edge roughness |
| CN109409158B (en) * | 2018-09-29 | 2021-08-31 | 武汉保诚信网络科技有限公司 | Anti-counterfeiting method based on two-dimensional code edge roughness |
| CN111504608A (en) * | 2019-01-31 | 2020-08-07 | 中强光电股份有限公司 | Brightness uniformity detection system and brightness uniformity detection method |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1141665C (en) | 2004-03-10 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110163853B (en) | Edge defect detection method | |
| Gangwar et al. | IrisSeg: A fast and robust iris segmentation framework for non-ideal iris images | |
| CN110543822A (en) | finger vein identification method based on convolutional neural network and supervised discrete hash algorithm | |
| CN1141665C (en) | Micro image characteristic extracting and recognizing method | |
| CN109389033B (en) | Novel pupil rapid positioning method | |
| CN110929713B (en) | Steel seal character recognition method based on BP neural network | |
| CN105404885B (en) | A kind of two dimension character graphics identifying code complex background noise jamming minimizing technology | |
| CN110348289B (en) | Finger vein identification method based on binary image | |
| CN106127193B (en) | A kind of facial image recognition method | |
| CN1421815A (en) | Fingerprint image enhancement method based on knowledge | |
| CN105809625B (en) | Fragment reconstructing method based on local grain mode | |
| CN113516619A (en) | Product surface flaw identification method based on image processing technology | |
| CN1144158C (en) | Method for identifying human body biological characteristics | |
| CN108985294B (en) | Method, device and equipment for positioning tire mold picture and storage medium | |
| CN1246800C (en) | Synergic fingerprint identification method | |
| CN114418975A (en) | Cutting method of large-area animal fur contour image based on radius increment constraint | |
| Zhu et al. | Palmprint recognition based on PFI and fuzzy logic | |
| KRISHNA et al. | Digitization, Preservation and Character Recognition in Ancient Documents Using Image Processing Techniques–A Review | |
| CN110619273A (en) | Efficient iris recognition method and device | |
| Gonçalves et al. | Pixel-level Class-Agnostic Object Detection using Texture Quantization | |
| Dey et al. | Fast and accurate personal identification based on iris biometric | |
| CN110120118A (en) | A kind of identifying system for splicing coin | |
| Alkazzaz et al. | Precise pupil boundary detection using angular integral projection and active contour model | |
| CN110415229B (en) | Texture image quality estimation method based on Haar features and AdaBoost algorithm | |
| Kanagalakshmi et al. | Log-Gabor orientation with run-length code based fingerprint feature extraction approach |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C19 | Lapse of patent right due to non-payment of the annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |