CN105975909B - A kind of fingerprint classification method and fingerprint three-level classification method based on fractal dimension - Google Patents
A kind of fingerprint classification method and fingerprint three-level classification method based on fractal dimension Download PDFInfo
- Publication number
- CN105975909B CN105975909B CN201610270838.1A CN201610270838A CN105975909B CN 105975909 B CN105975909 B CN 105975909B CN 201610270838 A CN201610270838 A CN 201610270838A CN 105975909 B CN105975909 B CN 105975909B
- Authority
- CN
- China
- Prior art keywords
- fingerprint
- classification
- image
- fractal dimension
- dimension
- 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.)
- Active
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1347—Preprocessing; Feature extraction
- G06V40/1359—Extracting features related to ridge properties; Determining the fingerprint type, e.g. whorl or loop
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/12—Fingerprints or palmprints
- G06V40/1365—Matching; Classification
Landscapes
- Engineering & Computer Science (AREA)
- Human Computer Interaction (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Theoretical Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Collating Specific Patterns (AREA)
Abstract
Classification method provided by the invention rejects low quality fingerprint using the progressive Evaluation Algorithm of Fingerprint Quality of substep of more Assessing parameters, to improve the accuracy of fingerprint recognition.Three-level classification is carried out for up-to-standard fingerprint, fingerprint is divided into six classes according to Finger print feature by the first order；Classified characterized by the crestal line number between fingerprint singularity the second level；The third level is classified characterized by the fractal dimension of fingerprint image quality stability region, and consecutive sort and redundant search can be achieved in wherein second and third grade classification, to effectively improve the accuracy and robustness of genealogical classification.Experiment on NIST DB4 shows that classification proposed in this paper is complete, and retrieval rate is fast, adaptable, has stronger robustness, the fast automatic identification particularly suitable for large capacity fingerprint base.
Description
Technical field
The present invention relates to fingerprint analysis field, more particularly, to a kind of fingerprint classification method based on fractal dimension and
Fingerprint three-level classification method.
Background technique
Precise Identification personal identification is the critical issue that must be solved in the Network Information epoch to ensure information security.
Although effective means of the Automated Fingerprint Identification System as identification, in public security, finance, e-commerce and personal information security
Equal fields show huge development space and application prospect, but existing large capacity fingerprint base auto Fingerprint Identification System is fast
Also not fully up to expectations in terms of speed, accuracy and practicability, wherein the fingerprint classification problem of most critical is as area of pattern recognition
Difficult point[1,2], obvious breakthrough is not yet obtained, the extensive of the auto Fingerprint Identification System based on large capacity fingerprint base is seriously constrained
Using.
Domestic and foreign scholars have conducted extensive research fingerprint classification algorithm, propose rule-based, syntax, structure, statistics,
The fingerprint classification method of neural network and multi-categorizer[3,4].Document [5] proposes a kind of improved fingerprint based on fingerprint singularity
Fingerprint is divided into 12 classes by disaggregated model；Document [6] is divided into 8 classes by establishing three level list, by fingerprint；Document [7] is by height
Quality fingerprinting establishes three level list, fingerprint is divided into 151-226 class, document [8] gives the search strategy of large-scale fingerprint base.This
A little methods have two o'clock insufficient mostly: first is that causing classification to be imitated since the distribution probability of fingerprint classification in a natural environment is uneven
Rate reduces；Second is that distinguishing and reducing between class when classification number increases, the difficulty of Accurate classification is increased, and it is non-to may cause classification results
Uniquely.
Summary of the invention
The present invention is to solve the defect of the above prior art, provides a kind of fingerprint classification method based on fractal dimension,
This method classification is complete, adaptable, and recall precision is high, match cognization space is small, has stronger robustness, is suitable for big
The systematic searching and match cognization of capacity fingerprint base.
To realize the above goal of the invention, the technical solution adopted is that:
A kind of fingerprint classification method based on fractal dimension, comprising the following steps:
(1) using the coordinate of fingerprint singularity as the center of circle, using the threshold value set as radius on fingerprint image after binarization,
Intercept a border circular areas；
(2) spectrum analysis is carried out to the border circular areas intercepted, acquires the principal direction of fingerprint image in the region；
(3) image wide along fingerprint image principal direction interception unit pixel, draws position-grayscale image, calculates its box counting dimension
FDp；
(4) image wide along the vertical direction interception unit pixel of fingerprint image principal direction, draws position-grayscale image, counts
Calculate its box counting dimension FDv；
(5) according to box counting dimension FDp, box counting dimension FDvClassify to fingerprint.
Meanwhile the present invention also provides a kind of fingerprint three-level classification method, concrete scheme is as follows: respectively with fingerprint line
Crestal line number and fractal dimension are characteristic of division between type, fingerprint singularity, are classified to fingerprint image.
Preferably, the classification method specifically includes the following steps:
(1) being rounded a fingerprint base is set to be sorted, is denoted as A；
(2) one piece of fingerprint image is taken out from A and carries out quality evaluation using the progressive Evaluation Algorithm of Fingerprint Quality of substep, if matter
It measures unqualified, is included into low quality fingerprint image set B；Otherwise judge its line type classification, the line type classification is respectively left-handed,
One of dextrorotation, whirlpool, arch, account arch or miscellany；
(3) if fingerprint classification result obtained by previous step is left-handed, dextrorotation, whirlpool thrin, fingerprint image surprise is calculated
Crestal line number between dissimilarity, then classifies according to crestal line number；
(4) the fingerprint image fractal dimension of fingerprint quality stability region is calculated, and is classified according to fractal dimension；
(5) all fingerprints to be sorted are carried out with the operation of (2)-(4).
Preferably, fingerprint is divided into five classes according to the location information of fingerprint singularity by the basic line type classification.
Preferably, in the step (5), after the classification for completing all fingerprints to be sorted, classify to according to crestal line number
Fingerprint in obtained each classification set, is ranked up according to its crestal line number；Meanwhile to according to fingerprint image fractal dimension
Obtained each fingerprint in gathering of classifying of classifying is carried out, is ranked up according to its fractal dimension.
Compared with prior art, the beneficial effects of the present invention are:
Classification method provided by the invention rejects low-quality using the progressive Evaluation Algorithm of Fingerprint Quality of substep of more Assessing parameters
Fingerprint is measured, to improve the accuracy of fingerprint recognition.Three-level classification is carried out for up-to-standard fingerprint, the first order is according to fingerprint
Fingerprint is divided into six classes by line type feature；Classified characterized by the crestal line number between fingerprint singularity the second level；The third level is to refer to
The fractal dimension of print image quality stability region, which is characterized, classifies, and consecutive sort can be achieved in wherein second and third grade classification
And redundant search, to effectively improve the accuracy and robustness of genealogical classification.Experiment on NIST DB4 shows this hair
The classification of bright proposition is complete, and retrieval rate is fast, adaptable, has stronger robustness, particularly suitable for large capacity
The fast automatic identification of fingerprint base.
Detailed description of the invention
Fig. 1 is the schematic diagram of the fingerprint image of local self similarity.
Fig. 2 is the schematic diagram for calculating fingerprint image fractal dimension.
Fig. 3 is the flow chart of classification method.
Fig. 4 is the statistical results chart of fingerprint image fractal dimension classification.
Specific embodiment
Embodiment 1
One, the selection and extraction of multiclass classification feature
Towards large capacity fingerprint base realize multiclass classification index used by characteristic of division answer it is with good stability and
Separability.The present invention chooses the FRACTAL DIMENSION of crestal line number and fingerprint image central area between the intrinsic line type classification of fingerprint, singular point
For number as characteristic of division at different levels, these feature stability are good and mutually indepedent, have very strong separability.
The classification of 1.1 line types
Fingerprint is divided into five classes according to the location information of fingerprint singularity by basic line type classification.The present invention uses document
[9] the complex filter method proposed carrys out the central point and triangulation point of location fingerprint image, and this method can not only calculate singular point
Type, position and number, and the direction of singular point can be sought；Point based on singular point information proposed using document [10]
Fingerprint is divided into left-handed, dextrorotation, whirlpool, arch, five class of account arch by class method, and all fingerprint images that cannot be divided into preceding 5 class
Image set, which closes, is labeled as miscellany, amounts to 6 classes.
The classification of 1.2 crestal line numbers
Crestal line number is a kind of brief expression of distance between two points on fingerprint image[11].When the central point of a width fingerprint image
Exist with triangulation point and when can be accurately positioned, the crestal line number of point-to-point transmission is definite value, the shadow of deformation factor such as is not translated, rotated
It rings, therefore can be used as characteristic of division and discrimination and robustness are good.
Since only left-handed, dextrorotation, whirlpool and account arch fingerprint have 2 or 2 or more singular points, and account arch refers to
Line NATURAL DISTRIBUTION probability very little and between its central point and triangulation point almost without crestal line, therefore pressing line type carry out the first fraction
On the basis of class, crestal line number only is calculated to preceding 3 class fingerprint, realizes secondary classification.
The classification of 1.3 fractal dimensions
1. point shape basis
No characteristic dimension but has the geometric figure of self-similar structure to be known as a point shape[12], the most important characteristics of point shape are parts
There are self-similarities between entirety[13].It is divided to shape that can be divided into rule to be divided to shape and be irregularly divided to shape two classes, it is tight that rule divides shape to have
The self-similarity of lattice, irregularly the self-similarity of point shape exists only in scale invariant region, is approximate self-similarity or statistics
Self-similarity in meaning.
So far, dividing shape, there is no a stringent stringent definition.Nineteen eighty-two, Mandelbrot will divide the Shape definition to be
Hausdorff dimension is greater than the set of topological dimension[14], Hausdorff dimension is not usually integer, that is, shape is divided to have score
Dimension.Later, Mandelbrot had also been proposed a widely definition in 1986: a point shape is part and is integrated with certain mode
Similar shape, this definition refer not only to the self similarity of geometric figure it is emphasised that self-similarity in figure between part and entirety
Property, it can also include the self-similarity of the figures such as track in phase space, i.e. the self-similarity of dynamic process.
2. the fractals of fingerprint image
Fingerprint image is a kind of typical texture image, it is made of the staggered streakline with certain orientation of black and white.
In certain range scale, fingerprint image texture shows apparent fractal characteristic[15], such as fingerprint image shown in Fig. 1 (a)
Through shown in partially enlarged streakline form such as Fig. 1 (b), this kind of streakline is in whole picture fingerprint image everywhere as it can be seen that showing as line
Line curvature changes everywhere and has similar appearance, shows self-similarity[16]。
Fingerprint ridge goes out very strong directionality in topical manifestations, along streakline direction finger print data show it is irregular with
Machine feature, document [17] studies have shown that qualified pressing fingerprint Dou Youyige stable image quality area.From 2002 fingerprint of FVC
Take out 8 pieces of homologous fingerprint images in library and carry out binary conversion treatments, centered on certain point (such as central point or triangulation point), fixed half
The image of interception unit pixel width in diameter, as shown in Fig. 1 (c), it is seen then that under same position coordinate in the same direction on section
Although figure quality is not quite similar, strong self-similarity is shown.Document [18] explains the texture image of labyrinth
Fractal characteristic, and demonstrate with fractal image code fingerprint image with deformation insensitivity, therefore fingerprint classification feature can be used as.
3. the fractal image code of several picture
Fractal dimension is the important parameter for describing several picture fractal characteristic.Different definition that there are many fractal dimensions,
Middle box counting dimension is widely used because being easy to carry out mathematical computations and experiment measurement[19]。
Box counting dimension is defined as follows:
Let f be n dimension real number space RnIn any non-empty bounded subset, note N (A, δ) indicates that maximum gauge is that δ and can cover
The minimum number of the set of F is covered, then the upper box dimension definition of F are as follows:
The lower box dimension definition of F are as follows:
If upper and lower box ties up equal, the box counting dimension of F is defined as:
When calculating divides the box counting dimension of shape F, this point of shape can be placed on the grid of an even partition, be calculated needed for covering F
The minimum grid number wanted.The variation of required covering grid number, can calculate box counting dimension when by checking that grid gradually refines.If net
When lattice side length is ε, F is divided into N number of grid, then box counting dimension are as follows:
4. the fractal dimension calculation of fingerprint image
Steps are as follows for the fractal dimension calculation of fingerprint image:
(1) with the coordinate of fingerprint singularity (when no singular point, then the particle of fetching print image) for the center of circle, after binarization
Fingerprint image on using given threshold as radius, interception Fig. 2 (a) shown in border circular areas；
(2) spectrum analysis is carried out to the border circular areas intercepted, acquires the principal direction of fingerprint image in the region；
(3) image wide along fingerprint principal direction interception unit pixel, draws position-grayscale image as shown in Fig. 2 (b), presses
Formula (4) calculates its box counting dimension FDp, as shown in Fig. 2 (c)；
(4) image wide along the vertical direction interception unit pixel of fingerprint principal direction, draws the position-as shown in Fig. 2 (b)
Grayscale image calculates its box counting dimension FD by formula (4)v, as shown in Fig. 2 (c).
Two, multiclass classification retrieval frame and implementation process
In order to improve the recognition speed, accuracy and the robustness that carry out 1:N Automatic FingerprintVerification in large capacity fingerprint base,
The present invention takes crestal line number and the fractal dimension of fingerprint image quality stability region between Finger print, fingerprint singularity special for classification
Sign carries out three-level classification to fingerprint.In view of low quality fingerprint can seriously affect the accuracy and rapidity of identification, before classification, & apos
Low-quality fingerprint is rejected by quality evaluation, to improve the performance of Automated Fingerprint Identification System.
The progressive assessment of 2.1 fingerprint qualities substep
The present invention carries out the progressive assessment of substep to fingerprint image quality using the method that document [20,21] propose.
2.2 multiclass classification processes
Fingerprint multistage consecutive sort process proposed by the present invention based on fractals is as shown in Figure 3:
(1) being rounded a fingerprint base is set to be sorted, is denoted as A；
(2) one piece of fingerprint image being taken out from A and carrying out quality evaluation, low quality fingerprint image is included into if off quality
Set B；Otherwise judge its line type classification, be included into set Ci, wherein i=1~6, indicate 5 kinds of basic line type classes and 1 miscellany；
(3) it if fingerprint classification result obtained by previous step is left-handed, dextrorotation, whirlpool thrin, calculates between singular point
Crestal line number, and set R is included into according to given classifying rulesi；
(4) the fingerprint image fractal dimension of fingerprint quality stability region is calculated, and set F is included into according to given rulei；
(5) operation for carrying out (2)-(4) to all fingerprints to be sorted has after completing all classification task:
And meet following relationship:
2.3 consecutive sorts and redundant search
1. consecutive sort
The inhomogeneities of the probability distribution of fingerprint classification in a natural environment will significantly reduce the fingerprint point of fixed classification number
The classification effectiveness of class method, but increase classification number again may because between class difference reduce and lead to accuracy, or even make to classify non-
Uniquely.In view of consecutive sort can be difficult to the disadvantages of determining excessive with fingerprint quantity in subclass to avoid fingerprint classification, and can
To come the precision and speed of balanced sort searching system by adjusting range of search, and fingerprint ridge line number and fractal dimension are number
Value type data, therefore the present invention proposes to divide shape using crestal line number between Finger print, fingerprint singularity and fingerprint quality stability region
Three kinds of features of dimension carry out fingerprint three-level consecutive sort.The specific method is as follows:
When carrying out secondary classification by fingerprint ridge line number, sort to the fingerprint in second level subclass by crestal line number, then by finger
Line fractal dimension carries out three-level classification, and sorts to the fingerprint in three-level subclass by its fractal dimension, convenient for carrying out fingerprint by class
Consecutive retrieval.
2. redundant search
When being retrieved using crestal line number between fingerprint singularity and the fractal dimension of fingerprint central area, the object that is retrieved is
Numerical value vector can use radii fixus search or incremental search[22].The present invention centered on the starting point searched for, respectively forwardly, after
Direction is searched for one by one, until searching for successful match or having retrieved the fingerprint in preset threshold.
As known from the above, with the increase of classification series, computation complexity is increase accordingly, and the fingerprint number of identification to be matched
Amount is that the fingerprint quantity in new first-level class substantially reduces, and therefore, this method can effectively improve oneself based on large capacity fingerprint base
The retrieval rate of dynamic fingerprint recognition system, reduces search match time.
Three, experimental result and classification performance analysis
In the NISTDB4 fingerprint base for being suitable for automatic fingerprint classification research[23]Upper test fingerprint classification proposed by the present invention
The performance of algorithm.The library contains 4000 width difference fingerprint images of 2000 pieces of fingerprints, and image size is 512 × 512 pixels, every width figure
As being all labelled with its generic by fingerprint expert, wherein about 17% fingerprint for being difficult to classify because smudgy is noted as
2 kinds of classifications, test result are that one of them thinks that classification is correct.
The classification of 3.1 line types
The present invention randomly selects 2000 width fingerprint images in NIST-4 database and carries out quality assessment, wherein 214 width refer to
Print image is unqualified, is tested with remaining 1786 width fingerprint image sorting algorithm of the invention, the classification of first order line type
Test result it is as shown in table 1.
The basic line type classification results of table 1
As shown in Table 1, after taking quality evaluation strategy and increasing miscellany in the classification of line type, the error rate of line type classification
Are as follows:
ER=(25+23+19+7+56)/1786=7.28% (9)
Classification accuracy are as follows:
AR=1-ER=92.72% (10)
The classification accuracy of method provided by the invention has compared with the 84.3% of 85.4% and the document [7] of document [6]
It significantly improves.
The classification of 3.2 crestal line numbers
It chooses left-handed, dextrorotation and 3 class of whirlpool in 1786 width fingerprint test samples and amounts to 1065 width fingerprint images and carry out the
Secondary classification test.Since swirl type fingerprint has 2 or more singular points, nearest that of distance is calculated in experiment to singular point
Between crestal line number.
The several classes of other statistical results of 2 fingerprint ridge line of table
Crestal line number seeks carrying out on the fingerprint image of refinement between singular point.In view of same in fingerprint to be identified and fingerprint base
There are crestal line number between when quality difference, may cause calculated singular point is not exactly the same between one fingerprint, influence of noise and
Image thinning process may also cause the part of crestal line to distort, and leading to the calculating of crestal line number, there may be errors；In addition, being extracted
The position of singular point be also likely to be present error, for the robustness and accuracy for improving sorting algorithm, the present invention is using following
Redundancy classification policy:
If crestal line number maximum value is n, the fingerprint of crestal line number=0~2 is divided into one kind, is denoted as R1；Crestal line number=2~4
Fingerprint is divided into one kind, is denoted as R2；The rest may be inferred, until crestal line number=(n-2)~n is divided into last one kind, it is clear that each adjacent
There is overlapping covering in class, being formed, there is the redundancy of higher robustness to classify.Experiment shows usually may be used according to crestal line number between singular point
Fingerprint is divided into 13 classes, the fingerprint positioned at redundancy boundary is classified as upper one kind, and the results are shown in Table 2 for statistic of classification.
The classification of 3.3 fractal dimensions
Third level characteristic of division uses the fractal dimension of fingerprint image.Fingerprint similar in fractal dimension its grain distribution also phase
Seemingly, therefore discriminant classification can be carried out by the difference of the fractal dimension of fingerprint image.1000 pieces of homologous fingerprints pair are randomly selected, it is calculated
The fractal dimension of fractal dimension and each pair of fingerprint is poor, as a result as shown in Figure 4.Fig. 4 (a) shows that the FRACTAL DIMENSION of fingerprint image mainly collects
In between 1.2-1.6, fraction FRACTAL DIMENSION lower than 1.2 fingerprint image quality it is obviously relatively low.Fig. 4 (b) indicates each pair of same
The difference of the dimension of source fingerprint image, wherein the dimension difference of 82% homologous fingerprint image is lower than 0.2.The present invention will by fractal dimension
Fingerprint is divided into 24 classes as shown in table 3.
3 fingerprint image fractal dimension classification statistical conditions of table
If the consecutive sort of second and third grade can be achieved by crestal line number, fractal dimension sequence between fingerprint singularity.When into
When row fingerprint recognition, retrieval matching can be carried out in preset maximum search radius, if search radius is sufficiently large, matching is quasi-
True rate can reach 100%, but with the increase of search radius, while reject rate reduces, retrieval rate is decreased.It is practical
In application, can be according to the requirement appropriate adjustment search radius to matching accuracy rate.
3.4 performance evaluation
Penetrating coefficient P (system penetration coefficient) can be used for measuring Automated Fingerprint Identification System
Multiclass classification performance [24]:
P=C/N (11)
Wherein N is the fingerprint sum in system fingerprint database, C be find required for one piece of fingerprint than logarithm or
With number.It is scanned in the Automated Fingerprint Identification System of 1:N, if finding the best finger of matching effect without corresponding classification
The penetrating coefficient of line is P=1.If system uses linear search, classifying in the ideal situation, it is correct not refused and classified, then is
The penetrating coefficient of system are as follows:
Wherein p (wi) be every one kind distribution probability.The penetrating coefficient of first order classification can be obtained by table 1~3:
The classification penetrating coefficient of the second level:
The classification penetrating coefficient of the third level:
Therefore the penetrating coefficient of classification method proposed by the present invention are as follows:
P=P1×P2×P3=0.00078 (16)
Compared with three kinds of arithmetic results 0.147,0.028-0.14,0.0022-0.0447 that document [25] is mentioned, the present invention
The method of offer has better classification performance.
Classification effectiveness function based on classification information entropy can also be used for the classification performance of evaluation Automated Fingerprint Identification System.If
K is the characteristic that system classifies to fingerprint, then classification information entropy is defined as:
Classification effectiveness function is defined as:
Classification effectiveness function shows that sorting algorithm has the ability that fingerprint is divided into E class.Point of method provided by the invention
Class efficiency are as follows:
E=E1×E2×E3 (19)
Method i.e. provided by the invention has the ability of 1280 classes of being divided into fingerprint, with traditional 5 major class classification methods and
The 151-226 class of document [7] is compared, and classification effectiveness significantly improves, it is expected to be met requirement of the larger capacity fingerprint base to classification, be shown
It writes and improves retrieval rate and matching accuracy rate.
It is worth noting that due to crestal line number of the fingerprint central point in principal direction and perpendicular to the fingerprint in principal direction point
Both shape dimensions are completely mutually indepedent, therefore are expected to improve the accuracy of fingerprint classification simultaneously as characteristic of division using the two.
4 conclusions
Fingerprint fast recognition technology in large capacity fingerprint base is a big difficulty in current fingerprint Recognition field, efficiently
Classification method be its key.Fingerprint multistage consecutive sort method proposed by the present invention based on fractals is to improve automatic finger
The recognition efficiency of line identifying system is target, carries out quality evaluation to fingerprint image first, then utilizes ridge between line type, singular point
The mutually independent characteristic of division of line number, fingerprint image stability region fractal dimension 3 realizes multiclass classification, forms three-level classification
Mode.Experiment shows that the classification of method provided by the invention is complete, penetration capacity is strong, recall precision is high, robustness is good, is big
The automatic identification of capacity fingerprint base provides a kind of retrieval matching mechanisms rapidly and efficiently, has very strong practicability.
Obviously, the above embodiment of the present invention be only to clearly illustrate example of the present invention, and not be pair
The restriction of embodiments of the present invention.For those of ordinary skill in the art, may be used also on the basis of the above description
To make other variations or changes in different ways.There is no necessity and possibility to exhaust all the enbodiments.It is all this
Made any modifications, equivalent replacements, and improvements etc., should be included in the claims in the present invention within the spirit and principle of invention
Protection scope within.
Bibliography
New development [J] the natural science of the fingerprint identification technology such as [1] Tian Jie, Chen Xinjian, Zhang Yangyang is in progress, and 2006,16
(4):400-408
TIAN Jie,CHEN Xin-jian,ZHANG Yang-yang.New Progress in Fingerprint
Recognition Technology[J].Progress in Natural Science,2006,16(4):400-408
[2] Dongmei SUN, Qiu Zhengding biometrics identification technology summarize [J] electronic letters, vol, 2001,29 (12A): 1744-
1748
SUN Dong-mei,QIU Zheng-ding.A Survey of the Emerging Biometric
Technology[J].Acta Electronica Sinica,2001,29(12A):1744-1748
[3]N Yager,A Admin.Fingerprint classification:A review[J].Pattern
Analysis and Application,2004,7(1):77-93
[4] Nie Guijun, Wu Chen, finger of the such as leaf tin monarch based on continuous affined nonlinear system and improved Poincar é Index
Line classification [J] electronic letters, vol, 2006,34 (5): 947-952
NIE Gui-jun,WU Chen,YE Xi-jun,XU Rong-qing.Fingerprint Classification
Based on Both Continuously Distributed Directional Image and Modified Version
of PoincaréIndex[J].Acta Electronica Sinica,2006,34(5):947-952
[5] a kind of the 12nd whole nation fingerprint classification algorithm [C] for big storehouse matching Ma Huimin, Su Guang great, Huang Ruke
The Qingdao neural computing academic meeting paper collection, China, 2003,368-373
MA Hui-ming,SU Guang-da,HUANG Ru-ke.A Fingerprint Classification
Algorithm for Large Database Matching[C].12th National Congress of neural
Network computing, 2003,368-373
[6] a kind of network-based large-scale finger print data search method [J] computer science of Wu Aiyan, Wei Shize,
2005,32(8):196-199
WU Ai-Yan,WEI Shi-Ze.A Search Way Based on Internet Fingerprint Large
Database[J].Computer Science,2005,32(8):196-199
[7] fingerprint classification algorithm [J] the computer work of Wang Wentao, Yin Yilong, Dai Hongjun, Wang Wenhui based on multiple index
Journey, 2010,36 (5): 188-190
WANG Wen-tao,YIN Yi-long,DAI Hong-jun,WANG Wen-hui.Fingerprint
Classification Algorithm Based on Multi-level Index[J].Computer Engineering,
2010,36(5):188-190
[8] fingerprint database searching algorithm research [J] East China of Chen Nan, Zhang Ming, Wu Chen, the Yang Honglin based on multiple index
Shipping industry institute journal (natural science edition), 2004,18 (4): 67-70
CHEN Nan,ZHANG Ming,WU Chen,YANG Hong-lin.Search Algorithm for
Multilevel Indexes of Fingerprint Database.Journal of East China Shipbuilding
Institute(Natural Science Edition),2004,18(4):67-70
[9]Nilsson K.and Bigun J.Localization of corresponding points in
fingerprints by complex filtering[J].Pattern Recognition Letters,2003,24(13):
2135–2144,.
[10]Liu M H,Fingerprint classification based on singularities,in
Proceedings of Chinese Conferen ce on Pattern Recognition(2009),pp.1-5.
[11]Lin W.and Dubes R.A review of ridge counting in dermatoglyphics
[J].Pattern Recognition,1983,16(1):1–8.
[12]Kenneth J.Falconer.Fractal Geometry:mathematical foundations and
applications[M].Wiley,1990.
[13] Wei Gang, Lu Yiqin, Ou Yangjingzheng chaos, fractal theory and Speech processing [J] electronic letters, vol,
1996,24(1):34-38
WEI Gang,LU Yi-qing,OUYANG Jing-Zheng.Chaos and Fractal Theories for
Speech Signal Processing[J].Acta Electronica Sinica,1996,24(1):34-38
[14]Mandelbrot.The Fractal Geometry of Nature[M].New York,NY:
W.H.Freeman and Company,1982.
[15]Polikarpova,N.On the Fractal Features in Fingerprint Analysis[C]
.Proceedings of the 13th International Conference on Pattern
Recognition.1996,3:591-595.
[16]Mahdi Jampour,Hamid Shojaei.Compressing of Fingerprint Images by
Means of Fractals Feature[C].The Second International Conference on Machine
Vision,2009,18-21.
[17] deformation fingerprint matching algorithm [J] the Journal of Image and Graphics of Chen little Guang, Feng Jufu based on stability region,
2010,15 (8): 1220-1229
CHEN Xiao-guang,FENG Ju-fu.A Novel Algorithm for Distorted
Fingerprint Matching Using Stable Regions[J].Journal of Image and Graphics,
2010,15 (8): 1220-1229
[18]Fei Su,Jing-ao Sun,Anni Cai.Fingerprint classification based on
fractal analysis[C].Signal Processing Proceedings,5th International
Conference on Signal Processing,2000,3:1471-1474.
[19]Jian Li,QianDu,CaixinSun.An improved box-counting method for
image fractal dimension estimation[J].Pattern Recognition,2009,42(11):2460-
2469.
[20] Guo Qiang, Peng little Qi, Zhong Yunfei, Zuo Long enhance algorithm based on the fingerprint image quality that information availability is evaluated
[J] computer engineering with apply .2012,48 (8): 207-210
GUO Qiang, PENG Xiao-qi, ZHONG Yun-fei, ZUO Long.Fingerprint Image
Quality Enhancement Algorithm Based on Information Availability[J].Computer
Engineering and Applications, 2012,48 (8): 207-210
[21] Quality evaluation for fingerprint image algorithm [J] of Guo Xinxing, Peng little Qi, Zhao Qi, Wang Jiankang based on Assessing parameters
.2009,11:4393-4394,4397 are studied in computer application
GUO Xin-xing,PENG Xiao-qi,ZHAO Qi,WANG Jian-kang.Fingerprint image
quality evaluation based on discriminant factors[J].Application Research of
Computers, 2009,11:4393-4394,4397
[22]Cappelli R.,Maio D.and Maltoni D.Indexing Fingerprint Databases
for Efficient 1:N Matching in Proc.Int.Conf.on Control Automation Robotics
and Vision(6th),2000.
[23]Watson C.I.and Wilson C.L.NIST Special Database 4,Fingerprint
Database.U.S.National Institute of Standards and Technology,1992.
[24]J.L.Wayman.Error rate equations for the general biometric system
[J].IEEE Robotics and Automation Magazine,1999,6(1):35–48.
[25] Wang Wei reach three-level fingerprint classification systematic research [D] Zhejiang Polytechnical University, 2010.
WANG Wei-da.Research on Three-Level Fingerprint Classification System
[D] .Zhejiang University of Technology, 2010.
Claims (4)
1. a kind of fingerprint classification method based on fractal dimension, it is characterised in that: the following steps are included:
(1) using the coordinate of fingerprint singularity as the center of circle, using the threshold value set as radius on fingerprint image after binarization, interception
One border circular areas；
(2) spectrum analysis is carried out to the border circular areas intercepted, acquires the principal direction of fingerprint image in the region；
(3) image wide along fingerprint image principal direction interception unit pixel, draws position-grayscale image, calculates its box counting dimension FDp；
(4) image wide along the vertical direction interception unit pixel of fingerprint image principal direction, draws position-grayscale image, calculates it
Box counting dimension FDv；
(5) according to box counting dimension FDp, box counting dimension FDvClassify to fingerprint.
2. a kind of fingerprint three-level classification method, it is characterised in that: respectively with crestal line number and right between Finger print, fingerprint singularity
It is required that fractal dimension described in 1 is characteristic of division, classify to fingerprint image.
3. fingerprint three-level classification method according to claim 2, it is characterised in that: the classification method specifically includes following
Step:
(1) being rounded a fingerprint base is set to be sorted, is denoted as A；
(2) one piece of fingerprint image is taken out from A and carries out quality evaluation using the progressive Evaluation Algorithm of Fingerprint Quality of substep, if quality is not
It is qualified then be included into low quality fingerprint image set B；Otherwise judge its line type classification, the line type classification is respectively left-handed, dextrorotation,
One of whirlpool, arch, account arch or miscellany；
(3) if fingerprint classification result obtained by previous step is left-handed, dextrorotation, whirlpool thrin, fingerprint image singular point is calculated
Between crestal line number, then classified according to crestal line number；
(4) the fingerprint image fractal dimension of fingerprint quality stability region is calculated, and is classified according to fractal dimension；
(5) all fingerprints to be sorted are carried out with the operation of (2)-(4).
4. fingerprint three-level classification method according to claim 3, it is characterised in that: in the step (5), completion is needed
Classify after the classification of fingerprint, obtained each fingerprint in gathering of classifying of classifying is carried out to according to crestal line number, according to its crestal line
Number is ranked up；Meanwhile obtained each fingerprint in gathering of classifying of classifying, root are carried out to according to fingerprint image fractal dimension
It is ranked up according to its fractal dimension.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610270838.1A CN105975909B (en) | 2016-04-27 | 2016-04-27 | A kind of fingerprint classification method and fingerprint three-level classification method based on fractal dimension |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610270838.1A CN105975909B (en) | 2016-04-27 | 2016-04-27 | A kind of fingerprint classification method and fingerprint three-level classification method based on fractal dimension |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105975909A CN105975909A (en) | 2016-09-28 |
| CN105975909B true CN105975909B (en) | 2019-11-29 |
Family
ID=56993791
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610270838.1A Active CN105975909B (en) | 2016-04-27 | 2016-04-27 | A kind of fingerprint classification method and fingerprint three-level classification method based on fractal dimension |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105975909B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2803001C1 (en) * | 2023-04-17 | 2023-09-05 | федеральное государственное бюджетное образовательное учреждение высшего образования "Национальный исследовательский университет "МЭИ" (ФГБОУ ВО "НИУ "МЭИ") | Method for identifying a person by the trace of the skin of a part of the human body using fractal dimension |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111062978B (en) * | 2019-11-27 | 2022-02-01 | 武汉大学 | Texture recognition method for spatio-temporal image flow measurement based on frequency domain filtering technology |
| CN111179265B (en) * | 2020-01-10 | 2023-04-18 | 敦泰电子（深圳）有限公司 | Image-based fingerprint quality evaluation method and device and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101145196A (en) * | 2006-09-13 | 2008-03-19 | 中国科学院自动化研究所 | Quick fingerprint identification method based on strange topology structure |
| CN102708364A (en) * | 2012-05-31 | 2012-10-03 | 西安电子科技大学 | Cascade-classifier-based fingerprint image classification method |
| CN104636721A (en) * | 2015-01-16 | 2015-05-20 | 青岛大学 | Palm print identification method based on contour and edge texture feature fusion |
| CN105354560A (en) * | 2015-11-25 | 2016-02-24 | 小米科技有限责任公司 | Fingerprint identification method and device |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN100412883C (en) * | 2006-03-23 | 2008-08-20 | 北京中控科技发展有限公司 | Fingerprint identifying method and system |
| SE530514C2 (en) * | 2006-09-07 | 2008-06-24 | Innitor Ab | A method, apparatus and computer software product in fingerprint matching |
| US8983208B2 (en) * | 2006-11-14 | 2015-03-17 | West Virginia University | Pattern matching based on global quantitative characterization of patterns |
-
2016
- 2016-04-27 CN CN201610270838.1A patent/CN105975909B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101145196A (en) * | 2006-09-13 | 2008-03-19 | 中国科学院自动化研究所 | Quick fingerprint identification method based on strange topology structure |
| CN102708364A (en) * | 2012-05-31 | 2012-10-03 | 西安电子科技大学 | Cascade-classifier-based fingerprint image classification method |
| CN104636721A (en) * | 2015-01-16 | 2015-05-20 | 青岛大学 | Palm print identification method based on contour and edge texture feature fusion |
| CN105354560A (en) * | 2015-11-25 | 2016-02-24 | 小米科技有限责任公司 | Fingerprint identification method and device |
Non-Patent Citations (1)
| Title |
|---|
| Multi-level fingerprint continuous classification for large-scale fingerprint database using fractal analysis;Yunfei Zhong 等;《International Journal of Biometrics》;20160101;第8卷(第2期);第2.3节，第3.2节，第4.3节 * |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| RU2803001C1 (en) * | 2023-04-17 | 2023-09-05 | федеральное государственное бюджетное образовательное учреждение высшего образования "Национальный исследовательский университет "МЭИ" (ФГБОУ ВО "НИУ "МЭИ") | Method for identifying a person by the trace of the skin of a part of the human body using fractal dimension |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105975909A (en) | 2016-09-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108108657B (en) | Method for correcting locality sensitive Hash vehicle retrieval based on multitask deep learning | |
| Li et al. | SHREC’14 track: Extended large scale sketch-based 3D shape retrieval | |
| Yue et al. | Hashing based fast palmprint identification for large-scale databases | |
| CN109886284B (en) | Fraud detection method and system based on hierarchical clustering | |
| Iloanusi | Fusion of finger types for fingerprint indexing using minutiae quadruplets | |
| Cappelli et al. | A fingerprint retrieval system based on level-1 and level-2 features | |
| CN107239792A (en) | A kind of workpiece identification method and device based on binary descriptor | |
| CN110580510A (en) | clustering result evaluation method and system | |
| CN105975909B (en) | A kind of fingerprint classification method and fingerprint three-level classification method based on fractal dimension | |
| Xu et al. | Detection method of insulator based on single shot multibox detector | |
| CN114861760A (en) | Improved research based on density peak value clustering algorithm | |
| CN111914930A (en) | Density peak value clustering method based on self-adaptive micro-cluster fusion | |
| CN113705920B (en) | Method for generating water data sample set for thermal power plant and terminal equipment | |
| Zhong et al. | Multi-level fingerprint continuous classification for large-scale fingerprint database using fractal analysis | |
| Choi et al. | An improved fingerprint indexing algorithm based on the triplet approach | |
| Zaghdoudi et al. | Multiple classifier combination for steel surface inspection | |
| Khodadoust et al. | A novel indexing algorithm for latent palmprints leveraging minutiae and orientation field | |
| CN107563399A (en) | The characteristic weighing Spectral Clustering and system of a kind of knowledge based entropy | |
| Singh et al. | Segmentation techniques through machine based learning for latent fingerprint indexing and identification | |
| Tsai et al. | Fingerprint clustering for forensic data indexing | |
| Song et al. | Fingerprint indexing based on minutia-centred deep convolutional features | |
| Song et al. | Learning discriminative and invariant representation for fingerprint retrieval. | |
| CN113052082B (en) | Voltage sag event correlation method based on image recognition | |
| Boutella et al. | Fingerprint identification by wave atoms transform and SVM | |
| Omranpour et al. | Representation of fingerprint recognition system based on geometric and statistical features of distance and angle of minutiae points |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |