CN1920850A - Irrelevant technique method of image pickup device in fingerprint recognition algorithm - Google Patents
Irrelevant technique method of image pickup device in fingerprint recognition algorithm Download PDFInfo
- Publication number
- CN1920850A CN1920850A CN 200610068892 CN200610068892A CN1920850A CN 1920850 A CN1920850 A CN 1920850A CN 200610068892 CN200610068892 CN 200610068892 CN 200610068892 A CN200610068892 A CN 200610068892A CN 1920850 A CN1920850 A CN 1920850A
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- image
- matched
- algorithm
- feature
- 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
Images
Abstract
The invention relates to an irrelative technique of image collector in fingerprint recognize algorism, wherein it comprises: image collector irrelative technique of character extract algorism and image collector irrelative of fingerprint matching algorism, while the image obtained in first step will be matched by second step; the first step uses large-window frequency analysis algorism, to realize self-adaptive selection of block size; the second image collector irrelative of fingerprint matching algorism uses triangle method to calculate the ratio between two fingerprint character modes. The invention can apply one fingerprint recognize algorism to different fingerprint collectors, to improve the application of fingerprint recognize algorism.
Description
Technical field
The present invention relates to automatic fingerprint recognition field, irrelevant technique method of image pickup device in specifically a kind of algorithm for recognizing fingerprint.
Background technology
We know, in recent years, are that the biometrics identification technology of representative receives much concern with automatic fingerprint identification technology.Automatically fingerprint identification technology is meant that crestal line, the valley line distribution pattern of utilizing the finger tip surface lines confirm to be identified a kind of biological identification technology of the identity of object.The human fingerprint that uses has had very long history as the means of identification, and the legitimacy of using fingerprint to carry out identification has also obtained approval widely already.
Automatically fingerprint identification technology process years of researches and development, ripe relatively.But all these algorithm systems all are based on same prerequisite: input fingerprint and template fingerprint that is to say that all from a collector fingerprint image is on all four on image resolution ratio and image size.Up to now, the independence problem that relates to the image capture device of many collectors does not cause enough attention.
The image capture device independence problem of so-called algorithm, say intuitively, be exactly under the prerequisite of using same set of recognizer, carry out the fingerprint registration with a fingerprint collecting equipment, discern and gather fingerprint image, and can guarantee the correctness discerned with the different with it collecting device of another acquisition principle, image resolution ratio.In itself, this is the adaptability problem of a cover algorithm for recognizing fingerprint to fingerprint collecting equipment, promptly for various model differences, acquisition principle difference, image resolution ratio difference, the picture quality obtained is different and even the also different fingerprint collecting equipment of size of the fingerprint actual area of gathering, one overlap algorithm for recognizing fingerprint whether correctly characteristic information extraction, correctly discern.If a cover algorithm for recognizing fingerprint has such ability, we just say that this cover algorithm for recognizing fingerprint possesses the image capture device independence.
Image capture device independence technology comprises the image capture device independence technology of feature extraction algorithm and the image capture device independence technology of fingerprint matching algorithm.Feature extraction is meant after fingerprint image preprocessing, adopts special algorithm to extract effective finger print information of some in order to characterize fingerprint image.Fingerprint image after conventional detail characteristics of fingerprints extraction algorithm needs to adopt earlier method that streakline follows the tracks of to refinement carries out the streakline reparation, and then realizes that minutia extracts.Streakline is repaired not only complex steps, and more consuming time.At this problem, Yin Yilong etc. have proposed a kind of improved detail characteristics of fingerprints extraction algorithm.This algorithm at first directly extracts original details feature point set on the fingerprint image after the refinement; Each noise like and the characteristics thereof that exist in the analysis image are then summed up the regularity of distribution of pseudo-characteristic point; At last, in conjunction with local streakline directional information, at different noises, adopt algorithm targetedly, the pseudo-characteristic point that each noise like causes is deleted respectively, the feature point set that finally remains promptly is considered as real feature point set.
Selection problem for the fingerprint image block size, if a cover fingerprint characteristic extraction algorithm just is used on a certain fixing fingerprint image acquisition equipment, we always can utilize experience is that it determines a suitable relatively image block size, in most of the cases can satisfy application demand.
But, if require a cover feature extraction algorithm can directly be used on the multiple fingerprint image acquisition equipment, the i.e. fingerprint image of gathering for multiple distinct device, require that algorithm can both correctly be handled, characteristic information extraction reliably, then select suitable image block size just to become a relatively problem of difficulty.Because different fingerprint image acquisition equipment, its acquisition resolution possibility is different even have bigger difference (roughly from 250dpi to 900dpi), and it obviously is impracticable that the fingerprint image from different acquisition equipment is used identical image block size.
Fingerprint matching is last step in the fingerprint recognizer, and the effective finger print information that refers to utilize feature extraction algorithm to extract is judged whether homology of two width of cloth fingerprint images.1986, Ai Senna (Isenor) and match gram (Zaky) proposed a kind of figure matching process.2000, gold people such as (Jain) proposed the matching process based on bank of filters, and this method carries out 8 trend pass filterings and extract feature mating to entire image.At present, the Point Pattern Matching based on minutiae point (streakline end points and streakline bifurcation) is the main flow mode of fingerprint coupling.In the fingerprint matching process, at first determine the reference point between the pattern to be matched, calculate the rotation translation parameters then and carry out stance adjustment, calculate the coupling mark at last and determine whether homology of two width of cloth fingerprint images, more than the problem that solved be same model fingerprint capturer fingerprint matching problem.
If two feature modes to be matched only are different on ratio, and structurally are identical words, the proportionate relationship of calculating between them is not difficult.But in fingerprint matching, owing to be subjected to all multifactor influences, even information is from same finger, two feature modes to be matched structurally neither be identical, often exists the disappearance of real unique point, existence, the positioning feature point of indivedual pseudo-characteristic points exists deviation, two patterns to be matched to have only problems such as part coupling.This makes the difficulty of dealing with problems increase greatly.Under a kind of like this complex conditions, how to calculate two proportionate relationships between the feature mode to be matched effectively, be the key point that solves image capture device independence in the fingerprint matching algorithm.
Along with popularizing gradually of rapid development of network technique and network condition, shift to network environment gradually the application scenario of authentication.With the ecommerce is the online identity authentication of representative, will become the important applied field of fingerprint recognition gradually.The application of fingerprint recognition under network environment, should be similar to digital signature, can be progressively trend " pattern of finger print identifying " center "+remote terminal: promptly " the finger print identifying " center " is responsible for the collection of all sample fingerprint, storage and management, verification process then be mostly by in long-range service terminal images acquired, extract feature and realize.If algorithm for recognizing fingerprint can not be realized the independence of image capture device, then can have a strong impact on the application of fingerprint recognition in network environment.
Summary of the invention
The present invention provides irrelevant technique method of image pickup device in a kind of method algorithm for recognizing fingerprint simple, easy to use for overcoming above-mentioned the deficiencies in the prior art.
The objective of the invention is to adopt following technical proposals to realize: irrelevant technique method of image pickup device in a kind of algorithm for recognizing fingerprint, it comprises the image capture device independence technical step of feature extraction algorithm and the image capture device independence technical step of fingerprint matching algorithm, the characteristic extraction step of algorithm for recognizing fingerprint by the image capture device independence feature that takes the fingerprint, the coupling step by the image capture device independence realizes fingerprint comparison.
The image capture device independence technical step of feature extraction algorithm adopts big window frequency spectrum analysis method:
At first view picture fingerprint image or most of fingerprint image are wherein carried out Fourier transform, obtain its spectral image;
Analyze and handle at its spectral image then, the i.e. annulus of calculating energy aggregation extent maximum or part annulus on its spectral image, according to annulus or part annulus respective radius and carry out relation between the image size of Fourier transform, calculate the average ridge distance of this width of cloth fingerprint image, with the foundation of average ridge distance, realize the adaptively selected of image block size in this way again as the selection window size;
Utilize the detail characteristics of fingerprints extraction algorithm to extract streakline end points and bifurcation sign fingerprint characteristic dot pattern at last.
The irrelevant technique method of image pickup device of fingerprint matching algorithm adopts the trigonometric ratio method of computational geometry to solve the computational problem of proportionate relationship between two fingerprint characteristic patterns to be matched:
At first two fingerprint characteristic dot patterns to be matched are regarded as two discrete point sets in the plane domain, used trigonometric ratio that it is divided into the triangle gridding structure;
For a fingerprint feature point, if it has n adjacent unique point hypothesis, in the triangle gridding of this part, have two groups of geometric elements to can be used for solving the proportionate relationship problem of two feature modes to be matched: 1) line between this fingerprint feature point and adjacent n the unique point is leg-of-mutton limit, be n bar line segment, it can be considered as n dimension length vector; 2) in this n bar line segment, formed angle between any two adjacent segments also is n, it can be considered as n dimension angle vector;
From two patterns to be matched, respectively get a unique point arbitrarily, it is right to form a unique point, utilize above two groups of geological informations, the average ridge distance information of the attribute information of fingerprint feature point itself such as unique point type and previous calculations judges whether a pair of unique point mates;
Between two feature modes to be matched, determine that some unique points to positive match are right, calculate two proportionate relationships between the feature mode to be matched;
At last, two patterns to be matched are carried out scaling and realize coupling under the same ratio.
The reliable computing method of described average ridge distance are:
1) uses big window to carry out spectrum analysis, cover the view picture fingerprint image as far as possible, even when guaranteeing that the fingerprint image mass ratio is relatively poor, the energy part also has enough intensity, is convenient to detect and calculate in the frequency domain figure picture;
2) the maximum direction of calculating energy, i.e. basic frequency of signal direction, and the information on all the other directions is given up;
3) utilize radial function, realize calculating average ridge distance.
Comprise the image capture device independence technical step of feature extraction algorithm and the image capture device independence technical step of fingerprint matching algorithm in the irrelevant technique method of image pickup device of the present invention.
(1) feature extraction is meant after fingerprint image preprocessing, adopts special algorithm to extract effective finger print information of some in order to characterize fingerprint image.
The present invention utilizes the information of pending original fingerprint image itself, for it determines suitable block size.The selection of fingerprint image block size mainly is to decide according to the mean distance between fingerprint ridge.Theoretically, the image block size should be 2 to 4 times of image ridge distance.At the concrete fingerprint image of a width of cloth,, just can select suitable image block size for it if can calculate its average ridge distance reliably.The adaptively selected problem of image block size just changes the problem of calculating the average ridge distance of a width of cloth fingerprint image how reliably into like this.
(2) fingerprint matching is meant the fingerprint characteristic that utilize to extract, and differentiates whether homology of two width of cloth fingerprint images.
The present invention is in the matching algorithm based on unique point, and two fingerprint characteristic patterns to be matched can be regarded two discrete point sets on the plane domain as.Like this, the problem of calculating proportionate relationship between two fingerprint characteristic patterns to be matched just can be regarded the proportionate relationship problem of calculating between two plane discrete point sets as." trigonometric ratio " method in the computational geometry has the characteristic of the following aspects, makes it have certain advantage on addressing this problem:
1) " trigonometric ratio " method has uniqueness to the division result of discrete point set.Promptly for two identical dot patterns, use same triangle division criterion, division result is identical.This is the build-in attribute of " trigonometric ratio " method, also is the basis of calculating the proportionate relationship between two discrete point sets
2) " trigonometric ratio " method has good locality to the division result of fingerprint characteristic dot pattern.Triangle division has multiple different criterion, intends adopting " minimum angle maximum " (max-min) criterion in the research of this project, and the plane point set that fingerprint feature point is formed carries out triangle division, and the triangle gridding that division is obtained has good locality.In the fingerprint characteristic dot pattern, this division rule can make the appearance of the disappearance of the real unique point of minority, indivedual pseudo-characteristic points and the factors such as deviations of unique point that the influence of dividing the result is confined in the little regional area as far as possible, is essentially identical thereby make most of dividing region result of two feature modes (if they are from same finger) to be matched
3) " trigonometric ratio " method has constant rate to the division result of fingerprint characteristic dot pattern.This is the build-in attribute of " trigonometric ratio " method, and also we are needed exactly
The invention has the beneficial effects as follows: solved the problem that under existing algorithm system a kind of algorithm for recognizing fingerprint depends critically upon a cover fingerprint collecting equipment, make the fingerprint collecting equipment that a kind of algorithm for recognizing fingerprint can the multiple different model of simultaneous adaptation, improved the versatility of algorithm for recognizing fingerprint.
Description of drawings
Fig. 1 is the fingerprint image A of the same finger that collects of different mining storage;
Fig. 2 is the fingerprint image B of the same finger that collects of different mining storage;
Fig. 3 is the fingerprint image C of the same finger that collects of different mining storage;
Fig. 4 is the ROC curve map of A of the present invention, B, AB;
Fig. 5 is the ROC curve map of A of the present invention, C, AC;
Fig. 6 is the ROC curve map of A of the present invention, B, C, ABC.
Embodiment
The present invention is further described below in conjunction with drawings and Examples.
Embodiment: use self-built storehouse to test, two collector that use in self-built storehouse is respectively FPR-361/300 electric capacity fingerprint acquisition instrument and safe BIOCA-120 electric capacity fingerprint acquisition instrument of producing of Shanghai cutting edge of a knife or a sword and a U are U 4000 Sensor optical fingerprint collectors of just producing in the Hangzhou of being produced by middle control science and technology, called after A, B and C, its resolution is respectively 450dpi, 500dpi and 512dpi, as Fig. 1, Fig. 2 and shown in Figure 3.Fingerprint image derives from 16 student volunteers (24 years old mean age, the men and women respectively accounts for half).The finger that every volunteer is gathered is the forefinger and the middle finger (totally 4 fingers) of both hands, and each finger collection 10 width of cloth fingerprint image amounts to 640 width of cloth fingerprint images.Testing scheme comprises: test separately and cross-beta.Test A, B and C are meant test performance index on fingerprint image storehouse A, B and C respectively separately; Cross-beta AB is meant that thereby coupling is from the corresponding fingerprint image test performance index of same finger on fingerprint image storehouse A and B.In like manner, can draw cross-beta AC and ABC.As Fig. 4, Fig. 5, shown in Figure 6, similar by the ROC curve of cross-beta gained to the ROC curve of testing gained separately, cross-beta AB, AC, ABC gained ROC curve are respectively between the independent centre of testing A, B, C, wherein, the technical indicator EER that obtains on fingerprint base A, C, AC (error rate) is respectively 1.1%, 5.4% and 4.2%, proves absolutely that the effect of the fingerprint image that the recognition effect that the present invention is directed to the fingerprint image of being gathered by different fingerprint collecting equipment and same fingerprint collecting equipment are gathered is very nearly the same.In general, image capture device independence technology has obtained good performance index on self-built storehouse.
Claims (2)
1. irrelevant technique method of image pickup device in the algorithm for recognizing fingerprint, it comprises the image capture device independence technical step of feature extraction algorithm and the image capture device independence technical step of fingerprint matching algorithm, it is characterized in that the image that the image capture device independence technical step of feature extraction algorithm obtains is realized the coupling of image by the image capture device independence technical step of fingerprint matching algorithm;
The image capture device independence technical step of feature extraction algorithm adopts big window frequency spectrum analysis method:
At first view picture fingerprint image or most of fingerprint image are wherein carried out Fourier transform, obtain its spectral image;
Analyze and handle at its spectral image then, the i.e. annulus of calculating energy aggregation extent maximum or part annulus on its spectral image, according to annulus or part annulus respective radius and carry out relation between the image size of Fourier transform, calculate the average ridge distance of this width of cloth fingerprint image, with the foundation of average ridge distance, realize the adaptively selected of image block size in this way again as the selection window size;
Utilize the detail characteristics of fingerprints extraction algorithm to extract streakline end points and bifurcation sign fingerprint characteristic dot pattern at last;
The irrelevant technique method of image pickup device of fingerprint matching algorithm adopts the trigonometric ratio method of computational geometry to solve the computational problem of proportionate relationship between two fingerprint characteristic patterns to be matched:
At first two fingerprint characteristic dot patterns to be matched are regarded as two discrete point sets in the plane domain, used trigonometric ratio that it is divided into the triangle gridding structure;
For a fingerprint feature point, if it has n adjacent unique point hypothesis, in the triangle gridding of this part, have two groups of geometric elements to can be used for solving the proportionate relationship problem of two feature modes to be matched: 1) line between this fingerprint feature point and adjacent n the unique point is leg-of-mutton limit, be n bar line segment, it can be considered as n dimension length vector; 2) in this n bar line segment, formed angle between any two adjacent segments also is n, it can be considered as n dimension angle vector;
From two patterns to be matched, respectively get a unique point arbitrarily, it is right to form a unique point, utilize above two groups of geological informations, the average ridge distance information of the attribute information of fingerprint feature point itself such as unique point type and previous calculations judges whether a pair of unique point mates;
Between two feature modes to be matched, determine that some unique points to positive match are right, calculate two proportionate relationships between the feature mode to be matched;
At last, two patterns to be matched are carried out scaling and realize coupling under the same ratio.
2. irrelevant technique method of image pickup device in the algorithm for recognizing fingerprint according to claim 1 is characterized in that, the reliable computing method of described average ridge distance are:
1) uses big window to carry out spectrum analysis, cover the view picture fingerprint image as far as possible, even when guaranteeing that the fingerprint image mass ratio is relatively poor, the energy part also has enough intensity, is convenient to detect and calculate in the frequency domain figure picture;
2) the maximum direction of calculating energy, i.e. basic frequency of signal direction, and the information on all the other directions is given up;
3) utilize radial function, realize calculating average ridge distance.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100688924A CN100370472C (en) | 2006-09-18 | 2006-09-18 | Irrelevant technique method of image pickup device in fingerprint recognition algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100688924A CN100370472C (en) | 2006-09-18 | 2006-09-18 | Irrelevant technique method of image pickup device in fingerprint recognition algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1920850A true CN1920850A (en) | 2007-02-28 |
| CN100370472C CN100370472C (en) | 2008-02-20 |
Family
ID=37778570
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006100688924A Expired - Fee Related CN100370472C (en) | 2006-09-18 | 2006-09-18 | Irrelevant technique method of image pickup device in fingerprint recognition algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100370472C (en) |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100592323C (en) * | 2008-07-01 | 2010-02-24 | 山东大学 | Method for identifying fingerprint facing image quality |
| CN101539993B (en) * | 2008-03-20 | 2012-05-23 | 中国科学院自动化研究所 | Multi-acquisition-instrument fingerprint crossing-matching method based on size scaling estimation |
| CN104933389A (en) * | 2014-03-18 | 2015-09-23 | 北京思而得科技有限公司 | Identity recognition device and method based finger veins |
| CN105913047A (en) * | 2016-05-12 | 2016-08-31 | 林梓梁 | Fingerprint identification method and device |
| CN108090475A (en) * | 2018-01-17 | 2018-05-29 | 深圳市爱克信智能股份有限公司 | A kind of algorithm for recognizing fingerprint |
| CN108121940A (en) * | 2016-11-29 | 2018-06-05 | 深圳指芯智能科技有限公司 | A kind of method and apparatus of fingerprint image analysis |
| CN111259359A (en) * | 2020-01-21 | 2020-06-09 | 恒大智慧科技有限公司 | Community-based fingerprint management method and system, community server and storage medium |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP3952293B2 (en) * | 2003-01-06 | 2007-08-01 | ソニー株式会社 | Fingerprint verification apparatus and method, recording medium, and program |
-
2006
- 2006-09-18 CN CNB2006100688924A patent/CN100370472C/en not_active Expired - Fee Related
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101539993B (en) * | 2008-03-20 | 2012-05-23 | 中国科学院自动化研究所 | Multi-acquisition-instrument fingerprint crossing-matching method based on size scaling estimation |
| CN100592323C (en) * | 2008-07-01 | 2010-02-24 | 山东大学 | Method for identifying fingerprint facing image quality |
| CN104933389A (en) * | 2014-03-18 | 2015-09-23 | 北京思而得科技有限公司 | Identity recognition device and method based finger veins |
| CN104933389B (en) * | 2014-03-18 | 2020-04-14 | 北京细推科技有限公司 | Identity recognition method and device based on finger veins |
| CN105913047A (en) * | 2016-05-12 | 2016-08-31 | 林梓梁 | Fingerprint identification method and device |
| CN108121940A (en) * | 2016-11-29 | 2018-06-05 | 深圳指芯智能科技有限公司 | A kind of method and apparatus of fingerprint image analysis |
| CN108090475A (en) * | 2018-01-17 | 2018-05-29 | 深圳市爱克信智能股份有限公司 | A kind of algorithm for recognizing fingerprint |
| CN111259359A (en) * | 2020-01-21 | 2020-06-09 | 恒大智慧科技有限公司 | Community-based fingerprint management method and system, community server and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN100370472C (en) | 2008-02-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Jain et al. | Intelligent biometric techniques in fingerprint and face recognition | |
| Yang et al. | A fingerprint verification algorithm using tessellated invariant moment features | |
| CN100370472C (en) | Irrelevant technique method of image pickup device in fingerprint recognition algorithm | |
| AU2002348288B2 (en) | Image identification system | |
| US20160070968A1 (en) | Image-based liveness detection for ultrasonic fingerprints | |
| Maltoni | A tutorial on fingerprint recognition | |
| Jiang et al. | Reference point detection for fingerprint recognition | |
| CN101464945A (en) | Identification characteristic identification method based on finger back arthrosis veins | |
| Yang et al. | Fingerprint verification based on invariant moment features and nonlinear BPNN | |
| Wang et al. | Fingerprint matching using orientationcodes and polylines | |
| Maltoni et al. | Advances in fingerprint modeling | |
| CN116704557A (en) | Low-quality fingerprint matching method based on texture information | |
| CN1209732C (en) | Iris identifying method based on ripple analysis and zero passage description | |
| Maddala et al. | Implementation and evaluation of NIST Biometric Image Software for fingerprint recognition | |
| Shin et al. | Detecting fingerprint minutiae by run length encoding scheme | |
| Zhou et al. | Crease detection from fingerprint images and its applications in elderly people | |
| Mali et al. | Fingerprint recognition using global and local structures | |
| Qi et al. | Segmentation of fingerprint images using the gradient vector field | |
| Surajkanta et al. | A digital geometry-based fingerprint matching technique | |
| Kumar et al. | Fingerprint recognition using minutiae extraction | |
| Yang et al. | Fingerprint matching using invariant moment features | |
| Banđur et al. | Improving minutiae extraction in fingerprint images through robust enhancement | |
| Viswanathan et al. | Morlet wavelet fingerprint invariant automated authentication system | |
| Min et al. | Intelligent fingerprint recognition system by using geometry approach | |
| Preetha et al. | Selection and extraction of optimized feature set from fingerprint biometrics-a review |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| 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 | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20080220 Termination date: 20130918 |