CN100447806C - Method, device and use for matching two-stage mixed-fingerprint characteristics - Google Patents
Method, device and use for matching two-stage mixed-fingerprint characteristics Download PDFInfo
- Publication number
- CN100447806C CN100447806C CNB2006100122502A CN200610012250A CN100447806C CN 100447806 C CN100447806 C CN 100447806C CN B2006100122502 A CNB2006100122502 A CN B2006100122502A CN 200610012250 A CN200610012250 A CN 200610012250A CN 100447806 C CN100447806 C CN 100447806C
- Authority
- CN
- China
- Prior art keywords
- matching
- minutiae
- point
- detail
- mtd
- 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.)
- Expired - Fee Related
Links
- 238000000034 method Methods 0.000 title claims abstract description 123
- 238000001514 detection method Methods 0.000 claims abstract description 15
- 230000008569 process Effects 0.000 claims description 52
- 238000004422 calculation algorithm Methods 0.000 claims description 21
- 230000003044 adaptive effect Effects 0.000 claims description 14
- 238000004364 calculation method Methods 0.000 claims description 7
- 230000001131 transforming effect Effects 0.000 claims description 6
- 238000006243 chemical reaction Methods 0.000 claims description 5
- 230000009466 transformation Effects 0.000 claims description 5
- 238000013459 approach Methods 0.000 claims description 4
- 238000004891 communication Methods 0.000 claims description 4
- 238000000605 extraction Methods 0.000 claims description 3
- 239000000203 mixture Substances 0.000 abstract 1
- 238000013519 translation Methods 0.000 description 9
- 230000014616 translation Effects 0.000 description 9
- 238000005516 engineering process Methods 0.000 description 8
- 230000008859 change Effects 0.000 description 7
- 238000010586 diagram Methods 0.000 description 6
- 230000006870 function Effects 0.000 description 3
- 230000002068 genetic effect Effects 0.000 description 3
- 238000004458 analytical method Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 2
- 230000000694 effects Effects 0.000 description 2
- 238000002474 experimental method Methods 0.000 description 2
- 239000004576 sand Substances 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- 238000009825 accumulation Methods 0.000 description 1
- PMGQWSIVQFOFOQ-YKVZVUFRSA-N clemastine fumarate Chemical compound OC(=O)\C=C\C(O)=O.CN1CCC[C@@H]1CCO[C@@](C)(C=1C=CC(Cl)=CC=1)C1=CC=CC=C1 PMGQWSIVQFOFOQ-YKVZVUFRSA-N 0.000 description 1
- 230000001186 cumulative effect Effects 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 238000012545 processing Methods 0.000 description 1
- 210000001525 retina Anatomy 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Landscapes
- Collating Specific Patterns (AREA)
- Image Analysis (AREA)
Abstract
A method for matching fingerprint character by two-stage mixture means includes integrating detail character matching manner with triangle matching manner organically, using result outputted by detail character matching manner as reference and using said reference as standard to carry out triangle matching detection on unmatched detail point in order to greatly raise accuracy of system identification and robustness of system under nonlinear deformation condition at certain prerequisite.
Description
Technical Field
The invention relates to a fingerprint feature matching method, in particular to a two-stage mixed fingerprint feature matching method which organically combines a minutiae feature matching method with a triangle matching method and quickly obtains a matching result by utilizing local minutiae feature information of a fingerprint image, a matching device for realizing the two-stage mixed fingerprint feature matching method and an intelligent card for matching fingerprint features by using the fingerprint feature matching device, and belongs to the technical field of pattern recognition and intelligent cards.
Background
As an effective means for solving the problem of identity recognition, smart cards have their own security value. Currently, smart card technology has been developed to a stage of identity recognition using biometrics. The biometric technology is used for collecting and analyzing unique features of human body, such as fingerprints, retina and voice patterns, and performing identity verification. Smart cards employing biometric identification technology are recognized as the best solution for reliable personal authentication today.
Among various biometric techniques, fingerprint recognition technology was the earliest developed, and people began to use computers to process fingerprints in the sixties of the twentieth century. Through continuous development, fingerprint identification is a relatively mature technology at present, and is widely applied to various fields such as communication, insurance, medical health, computer control systems, access control systems, attendance systems, online transactions, identity documents and the like.
Fingerprint identification is a typical pattern identification problem and mainly comprises fingerprint feature extraction and fingerprint feature matching. The fingerprint feature matching is to judge whether two fingerprint images come from the same person according to the feature description of the fingerprint, and is a key point for evaluating whether a fingerprint identification method is excellent or not.
The fingerprint matching technology has been intensively studied by the predecessors. To date, a variety of fingerprint matching algorithms have been proposed. The algorithms are based on graphic images and ridge structures, and the matching algorithm based on the feature points (minutiae) has the advantages of simplicity, rapidness, strong robustness and the like. The most common method at present is to use a detail point coordinate model provided by the federal survey bureau (FBI) in the united states to perform detail matching, and the most common method is to use two key points, namely a ridge line tip and a ridge line bifurcation point of fingerprint lines to identify fingerprints. By representing minutiae as a pattern of points, an automatic fingerprint authentication problem can be converted into a point pattern matching (minutiae matching) problem.
In addition, Isenor and Zaky propose a method for matching two fingerprint images by graph matching; hrechak proposes fingerprint feature matching based on structural information; the method comprises the following steps that (1) the Vinod and the Ghose apply an asymmetric neural network to fingerprint matching, and an asymmetric neural network point-based pattern matching algorithm is provided; the Tianjie et al apply the genetic algorithm to fingerprint matching and provide a fingerprint map matching algorithm based on the genetic algorithm; stockman et al propose a Hough transform-based method to convert point pattern matching into detection of peaks in Hough space of transform parameters; zsolt Mikl Lous et al propose an algorithm based on triangle matching; xudong Jiang et al propose a matching algorithm based on local and global structures; jain et al propose an algorithm for point pattern matching in Fingerprint matching, which converts minutiae in rectangular coordinate system into polar coordinate system, and performs point matching by string matching algorithm (see specifically Anil Jain, Lin hong and Ruud Bolle, On-Line Fingerprint Verification, IEEE Trans On Pattern Analysis and Machine Analysis, vol.19, No.4.p 302-313, 1997).
Among the above methods, the pure triangle matching method and the method based on the genetic algorithm are rarely adopted by people due to the excessive calculation amount, and the string matching method adopted by Jain et al is widely concerned due to the characteristics of simplicity and practicality, and is widely applied in practice. However, since this method relies heavily on a single matching reference point, some obvious pairs of minutiae are often missed when the nonlinear deformation is large, resulting in a high FRR (false rejection rate) of the system. For this, see fig. 1, where black dots are already matched dot pairs and square dots are supposed to be matched dot pairs.
Disclosure of Invention
It is a first object of the present invention to provide a two-stage hybrid fingerprint feature matching method. The method organically combines a minutiae matching method with a triangle matching method, and quickly obtains a matching result by using local minutiae information (including bifurcation points and end points of fingerprint ridge lines) of a fingerprint image.
A second object of the present invention is to provide a matching device for implementing the two-stage hybrid fingerprint feature matching method.
It is a third object of the present invention to provide a smart card for fingerprint feature matching using the above two-stage hybrid fingerprint feature matching device.
In order to achieve the purpose, the invention adopts the following technical scheme:
a two-stage hybrid fingerprint feature matching method is characterized by comprising the following steps:
the fingerprint feature matching method comprises two stages of detail feature matching and triangle constraint matching;
in the minutiae matching stage, for a pair of given reference points in the feature point sets of the two fingerprints, matching the minutiae sets of the two fingerprints by using a self-adaptive bounding box minutiae matching method, recording matching scores of the minutiae sets, and taking the size of the matching scores as a basis for judging whether the matching is successful or whether triangular matching is carried out subsequently;
in the triangular constraint matching stage, if the matching score obtained in the detail feature matching stage is higher than a preset high threshold value, the input detail point set and the template detail point set are considered to come from the same finger, and the matching process is ended; if the matching score is between the preset high threshold and the preset low threshold, further searching other matching minutiae pairs by utilizing a triangular matching process, and recalculating the matching score. If the recalculated matching score is higher than a preset middle threshold value, the input minutiae set and the template minutiae set are considered to come from the same finger, and the matching process is ended; and if the matching score obtained in the detail feature matching stage is lower than a preset low threshold value, selecting the next pair of detail nodes as reference points to restart the detail feature matching process.
Wherein,
and the detail feature matching calculation is carried out under a polar coordinate system.
In the detail feature matching stage, a given reference point of the pair is determined by the following steps:
and calculating Euclidean distance and rotation angle difference between one point in the template point set and one point in the input point set under the condition that the two points are the same type of detail points, and performing detail matching by taking the two points as a pair of reference points if the two points meet a preset threshold condition.
The detail feature matching stage further comprises the steps of:
1) transforming the input point set and other minutiae in the template point set to a polar coordinate system by taking the template minutiae and the input minutiae as reference minutiae;
2) sorting template minutiae and input minutiae in polar coordinates in the increasing direction of polar angles and connecting the template minutiae and the input minutiae into strings;
3) the strings are matched using an adaptive bounding box approach, the minutiae pair information on the match is recorded, and a match score is calculated.
The size of the adaptive bounding box is represented by radius _ size and angle _ size, which are calculated for template minutiae with a polar radius r using the following formula:
radius_size＝(α1+β1r)
angle_size＝(α2-β2r)
wherein alpha is1，β1，α2，β2R is the polar radius of the template minutiae point, which is a preset empirical parameter and is all greater than zero.
In the triangular constraint matching stage, the mixed matching score Ms is obtained by:
wherein nb _ pair is the number of matching point pairs obtained in the detail feature matching stage, new _ pair is the number of matching detail point pairs newly added in the triangle constraint matching stage, and Mc，NcThe number of the characteristic points of the template and the input fingerprint image in the public area is respectively, match _ error is the weighted average of each matching minutia to the matching error, and penalty is the penalty score.
The triangular constraint matching stage further comprises the following steps:
1) acquiring registration parameters of two fingerprint minutiae sets according to the matched point pairs;
2) registering the minutiae sets of the two fingerprints by using the calculated registration parameters;
3) taking a plurality of reference detail point pairs as a reference, carrying out primary detection, and searching other possible matching point pairs to be selected;
4) and for the preliminarily detected matching point pairs to be selected, performing triangular matching on the matching point pairs and the matched point pairs to form a triangle, and checking whether the triangle can meet the geometric constraint of the triangle.
In step 2 of the triangular constraint matching stage, the registration of the detail point set specifically includes the following steps:
1) calculating the relative offset and the relative rotation angle between the matched template point set and the input point set and the matched reference point obtained in the detail feature matching process;
2) according to a transformation formula:
and transforming the input detail point set into a coordinate system where the template detail point set is located.
In the preliminary detection of the detail points in the step 3 of the triangular constraint matching stage, a self-adaptive limit box method is used for matching the feature points;
the adaptive bounding box threshold is:
radius_size＝2(α1+β1r)
angle_size＝2(α2-β2r)
wherein alpha is1，β1，α2，β2R is the polar radius of the template minutiae point, which is a preset empirical parameter and is all greater than zero.
Step 4 of the triangular constraint matching stage is specifically divided into the following steps:
1) finding a triangle matched reference line segment in the matched minutiae pairs according to a triangle constraint condition;
2) performing triangle matching on the matching point pairs to be selected and triangles formed by the reference line segments respectively;
3) and the matching point pairs to be selected meeting the triangular constraint condition are new matching point pairs found by the compensation algorithm.
The triangle constraint conditions are as follows:
1) the angle difference between the corresponding angles of each vertex is within a preset range;
2) the length difference between three corresponding sides of the triangle is within a preset range;
3) the central point position offset is within a preset range;
4) the direction difference of the minutiae points corresponding to the vertices of the triangle is within a predetermined range.
5) The hand shapes of the triangles are the same.
The triangle matched reference line segment is determined by the following steps:
1) initializing a matching error counter of all matching minutiae to the connecting line to be 0;
2) calculating the matching error between the matching triangle pairs formed by all the matching detail points, and accumulating the matching error into a matching error counter of three corresponding edges of the triangle;
3) and taking 5-10 connecting lines with the minimum matching error and the length larger than a preset threshold value as a reference line segment for carrying out triangular matching on the candidate matching point pair.
A two-stage hybrid fingerprint feature matching apparatus, comprising:
a polar coordinate conversion unit for converting the spatial positions of the relevant input detail points and the template detail points into polar coordinates;
the minutiae matching unit is used for matching minutiae sets of the two fingerprints by using a self-adaptive limit box minutiae matching method for a pair of given reference points in the two fingerprint feature point sets and recording matching scores of the minutiae sets;
the triangular constraint matching unit is used for considering that the input minutiae set and the template minutiae set come from the same finger if the matching score obtained in the minutiae matching stage is higher than a preset high threshold value, and ending the matching process; if the matching score is between the preset high threshold and the preset low threshold, further searching other matching minutiae pairs by utilizing a triangular matching process, and recalculating the matching score. If the recalculated matching score is higher than a preset middle threshold value, the input minutiae set and the template minutiae set are considered to come from the same finger, and the matching process is ended; and if the matching score obtained in the detail feature matching stage is lower than a preset low threshold value, selecting the next pair of detail nodes as reference points to restart the detail feature matching process.
A smart card having a microprocessor, a memory, and a communication circuit, characterized in that:
the intelligent card is also provided with the two-stage mixed fingerprint feature matching device.
A method for using the above-mentioned intelligent card to carry on the authentication, users input his fingerprint through the fingerprint gathering equipment on the card reader, after fingerprint sensor gathers the fingerprint data, submit to the characteristic and withdraw the module, and transmit the fingerprint characteristic to the said intelligent card; the intelligent card matches the input fingerprint characteristics with the stored fingerprint characteristics, if the matching is successful, the identity authentication is completed, and the intelligent card is characterized in that:
the intelligent card adopts a two-stage mixed fingerprint characteristic matching method to carry out fingerprint characteristic matching.
The two-stage mixed fingerprint feature matching method and the matching device provided by the invention have the following advantages:
(1) the detail feature matching and the triangle matching are organically combined, so that the accuracy of system identification under the condition of nonlinear deformation is greatly improved on the premise of ensuring the real-time performance of the system and smaller size of the fingerprint feature template;
(2) in the method, the triangle matching uses some line segments with the minimum matching error in the point pairs matched in the detail feature matching stage as a reference to search other feature point pairs possibly matched, so that the computation amount in the matching process is greatly reduced;
(3) because the influence of the detail characteristic point matching error and the mismatch detail point on the matching result is considered in the calculation of the matching score, a more accurate matching result can be obtained.
The intelligent card adopting the hybrid fingerprint feature matching method can further improve the safety reliability and the use convenience of the intelligent card, thereby providing a new feasible way for the fusion of the biological feature recognition technology and the intelligent card application technology.
Drawings
The invention is described in further detail below with reference to the figures and specific examples.
FIG. 1 is a schematic diagram of unmatched pairs of points during detail feature matching.
Fig. 2 shows a bounding box of variable size.
FIG. 3 is a flow chart of the trigonometric compensation algorithm.
Fig. 4 is a schematic diagram illustrating a calculation process of the angular offset.
FIG. 5 is a diagram illustrating matching of a minutia with a plurality of minutiae.
Fig. 6 is a schematic diagram of a triangle formed by connecting three minutiae points.
Fig. 7 is a schematic diagram of the process of triangle matching.
FIG. 8 is a schematic overall flow chart of a two-stage hybrid fingerprint feature matching method.
FIG. 9 is a logical block diagram of a typical smart card.
Fig. 10 illustrates an exemplary process of fingerprint feature matching of a smart card with a built-in two-stage hybrid fingerprint feature matching device.
Detailed Description
The basic idea of the two-stage mixed fingerprint feature matching method is as follows: for a pair of given reference points in the two fingerprint feature point sets, matching the minutiae point sets of the two fingerprints by using a self-adaptive bounding box minutiae feature matching method, recording matching scores of the minutiae point sets, and taking the size of the matching score as a basis for whether the matching is successful or whether triangular matching is performed subsequently.
If the matching score obtained in the detail feature matching stage is higher than a higher threshold value T1If the input detail point set and the template detail point set come from the same finger, the matching process is successfully ended; if the matching score is between a lower threshold T2And a higher threshold value T1And further searching other possible matching minutiae pairs by utilizing a triangular matching process, and recalculating the matching score. If the recalculated match score is above a predetermined threshold Tr(T2＜Tr＜T1) If the input detail point set and the template detail point set come from the same finger, the matching process is successfully ended; if the matching score obtained in the detail feature matching stage is lower than a lower threshold value T2Then the next pair of minutiae points is selected as the reference point to restart the minutiae feature matching process. Threshold value T1，T2And TrDepending on the accuracy required of the system, this is generally obtained by a large number of experiments.
From the technical implementation perspective, the two-stage hybrid fingerprint feature matching method mainly includes the following technical contents: feature point set registration, detail feature matching, triangle matching compensation and matching score calculation. The details thereof will be described one by one.
1. Feature point set registration
For the template fingerprint to be matched and the input fingerprint, only coordinate position, direction and type information of minutiae points detected from a fingerprint image are obtained. Since the correspondence between the two fingerprint images is not known in advance, a suitable transformation is first found to correspond them, and this process is the registration of the fingerprints.
To this end, a similarity transformation is first constructed using a pair of feature points (one from the template fingerprint and one from the input fingerprint) as reference points, using the coordinate translation and direction rotation relationships between them, to transform the input fingerprint feature points into the coordinate space of the template fingerprint feature points. Since the non-linear deformation of the fingerprint image tends to be radial, the deformation in a certain area is large and then expands outwards non-linearly, so that the non-linear deformation can be better described in polar coordinates. In addition, in a polar coordinate system, there is no need to consider the translation between the reference points of the input image and the template image, because the translation between the input image and the template image is fixed, i.e., the translation between another pair of corresponding points is the same as the translation between the reference points, so that when the coordinates of another pair of corresponding points are converted into polar coordinates with respect to the reference points, the translations are cancelled out, and it is obviously easier to handle the rotation between the two images in the polar coordinate system than in the rectangular coordinate system. For the above reasons, the present invention performs the detail feature matching process in the polar coordinate system.
Even if the input fingerprint and the template fingerprint come from the same finger, there will be deformations like translation, rotation, and dimension change between them. Before matching the two images, the deformation parameters between them are first estimated and the two images are registered. Since two fingerprint images are usually acquired with the same instrument, the coefficient of scale change between them can be assumed to be 1. In addition, the translation between the two images may not be considered in polar coordinates. Thus only the rotation parameters between the input image and the template image need to be estimated.
Order to
Representing M minutiae points in the template fingerprint,
representing N minutiae points in the input fingerprint. Wherein x and y are coordinate positions of the minutiae points,
the direction of the minutiae point.
In order to convert the minutiae into the polar coordinate system, first, a reference point is selected from the template minutiae set and the input minutiae set as an origin in the corresponding polar coordinate system, and polar coordinates of other minutiae with respect to the reference point are calculated.
For each point P in the template point seti(1. ltoreq. i. ltoreq.M) and each point Q in the input point setj(1. ltoreq. j. ltoreq.N), if they are the same type of minutiae, calculating the Euclidean distance and P between themiDirection relative to QjRotation angle of direction rotate [ i ]][j]. If the two satisfy certain threshold conditions (such as the distance does not exceed 100, the angle difference does not exceed 45, and the threshold is determined by applying the degrees of freedom for image acquisition), they are used as a pair of reference points to perform the following detail matching process. Otherwise, other possible reference point pairs are then examined.
The input image and the template image are registered in a polar coordinate system, and all other input detail points and template detail points are only required to be respectively corresponding to the reference point PiAnd QjConverting into a polar coordinate system, and adding an angle rotate [ i ] to the polar angles of all the input minutiae points][j]. That is, the input minutiae and the template minutiae are both respectively referenced to the reference point PiAnd QjConversion into a polar coordinate system using the following equation
Wherein
Is the rectangular coordinate of the minutiae point to be converted,
is a rectangular coordinate of the reference minutiae point,
is a representation of the minutiae point in polar coordinates (r)iDenotes the polar radius, θiThe polar angle is shown to be,
indicating the angle between the minutiae direction and the reference minutiae direction). Then, we input theta for each minutiaiPlus an angle rotate [ i ]][j]。
2. Detail feature matching
The detail point matching algorithm used in the invention is a self-adaptive limit box detail feature matching method, which specifically comprises the following steps:
1) by template minutiae point PiAnd an input minutiae point QjAs reference minutiae points, the other minutiae points in the input point set and template point set are transformed to the polar coordinate system using the minutiae point set registration method described above:
2) the template minutiae and the input minutiae in polar coordinates are sorted in the increasing direction of the polar angle and connected into a string, which is expressed as follows:
wherein M and N are the number of the detail points in the template and the input feature vector respectively;
3) matching string P with adaptive bounding box approachi sAnd Qj sRecording the minutiae point pair information on the matching, calculating a matching score, and taking the size of the matching score as a basis for whether triangular matching is performed or not or whether the matching is successful or not:
if the matching score is higher than a higher threshold T1If the input detail point set and the template detail point set come from the same finger, the matching process is successfully ended;
if the matching score is between a lower threshold T2And a higher threshold value T1And further searching other possible matching minutiae pairs by utilizing a triangular matching compensation process, and recalculating a matching score. If the recalculated match score is above a predetermined threshold Tr(T2＜Tr＜T1) If the input detail point set and the template detail point set come from the same finger, the matching process is successfully ended;
if the matching score is below a lower threshold T2Then selecting the next pair of minutiae points as the reference point pair and returning to step 1) to restart the detail characteristic matching process.
4) If the matching process cannot be successfully ended in step 3), the two fingerprints are considered as fingerprints from different fingers.
The adaptive bounding box approach described above is described below. The adaptive bounding box and its size are shown in fig. 1, one bounding box is a box placed over a template minutia. The size of the bounding box is denoted radius _ size and angle _ size, the values of which will vary with the size of the minutiae polar radius. Radius _ size and angle _ size of template minutiae with polar radius r are calculated using the following equations.
radius_size＝(α1+β1r) (6)
angle_size＝(α2-β2r) (7)
Wherein alpha is1，β1，α2，β2R is the polar radius of the template minutiae point, which is a preset empirical parameter and is all greater than zero.
The adaptive sized bounding box is used instead of the fixed sized bounding box to make the algorithm more robust to nonlinear deformations. The non-linear deformation is generally large in a particular area and then expands non-linearly outward. When the pole radius of a minutia point is small, a small deformation can cause a large change in the pole angle, and a small change in the pole radius. So in this case the angle _ size of the bounding box should be larger and radius _ size smaller. On the other hand, when the pole radius of a minutia point is large, a small change in the pole angle results in a large variation in the position of the minutia point, and the deformation of the pole radius can be considered as the accumulation of the deformations of all regions between the minutia point and the reference minutia point. So in this case the angle _ size of the bounding box should be small and radius _ size should be large.
Matching Pi sAnd Qj sThe algorithm of (a) is described as follows:
1) the size of the bounding box for each template minutia is determined by equations (6) to (7). Set nb _ pair [ i][j]＝0.(nb_pair[i][j]Representing minutiae P in templatesiAnd an input minutiae point QjDetail point logarithm in matching when performing feature matching as a reference detail point)
2) The following cycle is performed:
While 1≤m≤M do
While 1≤n≤N do
if the template minutiae m and the input minutiae n satisfy conditionl, then
nb_pair[i][j]＝nb_pair[i][j]+1；
endif
Increase n；
End while
Increase m；
End while
In the above process, conditionl is defined as:
wherein:
and r _ low [ m ], r _ high [ m ], θ _ low [ m ], and θ _ high [ m ] are lower and upper limits of the bounding box polar radius and polar angle error, respectively, corresponding to the template minutiae m (their values are _ radius _ size, radius _ size, _ angle _ size, and angle _ size, respectively, calculated by equations (6) and (7)). Condition is a condition for considering template minutiae m and input minutiae n as a matching point pair. The meaning is that the input minutiae n should be inside the bounding box of the template minutiae m, and the difference in orientation of these two minutiae should be less than epsilon (e.g., epsilon-30).
3. Computation of minutiae matching scores
After the above minutiae matching process is completed with reference to two minutiae points (one from the template fingerprint and one from the input fingerprint), assuming that the number of minutiae points pairs on the match is nb _ pair, the cumulative match error for each matched point pair is match _ error. The matching score is calculated as follows:
wherein M isc，NcThe number of feature points of the template and the input fingerprint image in the common area, respectively, match _ error is the matching error of each matching minutiae pair (including Δ r, Δ θ,) A, β are predetermined weighting coefficients (with typical values of 2.0 and 3.0, respectively). Matching the above matching score with a preset matching threshold T1And comparing to judge whether to match or not.
4. Triangular matching compensation
The detail feature matching method can overcome a certain degree of nonlinear deformation and accurately find most matching point pairs, but because the method depends on a single datum point and the influence of larger nonlinear deformation too much, part of the point pairs which should be matched cannot become matching point pairs, so that the FRR (false rejection rate) is higher. In order to overcome the problem, the invention introduces a minutiae triangular matching compensation process by utilizing the topological invariance of a fingerprint minutiae set, and the basic idea is to use matched pairs as a reference to search other matched pairs of feature points. The specific flow is shown in fig. 3, and comprises the following steps:
s1: acquiring registration parameters of two fingerprint minutiae sets according to the matched point pairs;
s2: registering the minutiae sets of the two fingerprints by using the calculated registration parameters;
s3: taking a plurality of reference detail point pairs as a reference, carrying out primary detection, and searching other possible matching point pairs to be selected;
s4: and for the preliminarily detected matching point pairs to be selected, performing triangular matching on the matching point pairs and the matched point pairs to form a triangle, and checking whether the triangle can meet the geometric constraint of the triangle. The method comprises the following steps:
s41: finding a triangle matched reference line segment in the matched minutiae pairs according to the triangle matching constraint;
s42: performing triangle matching on the matching point pairs to be selected and triangles formed by the reference line segments respectively;
s43: the matching point pairs to be selected which meet the triangular constraint condition are new matching point pairs searched by a compensation algorithm;
s5: and (6) ending.
The input of the triangle matching compensation method comprises a template image detail point set, an input image detail point set, matching reference point pairs and matching point pair information obtained in a detail feature matching stage. Since the original positions of the two point sets may have some rotation and translation around the reference point pair, the point set registration should be performed first, transforming the input minutiae set into the coordinate system in which the template minutiae set is located.
The technical details of the triangle matching compensation method are described in detail below. First, the specific steps of the registration of the detail point set in step S2 will be described.
Matching reference points of the template point set and the input point set obtained in the detail feature matching process are Ref _ t and Ref _ i respectively.
The relative offset of the two sets of points is calculated as:
Δx＝Ref_t.x-Ref_i.x； (13)
Δy＝Ref_t.y-Ref_i.y； (14)
the calculation of the rotation angle is shown in fig. 4. And P' are corresponding points, the relative angles of the corresponding points and a reference point are respectively calculated, and the relative angle difference of each corresponding point pair is counted to obtain the rotation angle of the minutiae point. The average relative rotation angle is:
where Angle _ t is the Angle of the template minutiae with respect to the reference point Ref _ t, Angle _ i is the Angle of the input minutiae with respect to the reference point Ref _ i, and n is the logarithm of minutiae on the match.
Then, a transformation formula is adopted:
and transforming the input detail point set into a coordinate system where the template detail point set is located.
In step S3, the purpose of the preliminary minutiae detection is to find out possible pairs of matching points, and reduce the amount of computation of the later-stage triangle matching. This preliminary detection mainly uses the adaptive bounding box feature point matching technique described above.
To overcome the effects of non-linear distortion, a bounding box with a large range must be used to ensure that as many matching point pairs as possible are found. Considering the characteristic that the nonlinear deformation of the fingerprint image is diffused outwards in a small area, the closer the nonlinear deformation is to the central area of the deformation, the smaller the relative deformation is. Therefore, a plurality of (preferably 6-10) initial matching point pairs which are uniformly dispersed are used as the starting point of the initial detection, so that the effectiveness and the reliability of the initial detection are ensured.
Starting from a plurality of groups of matched pairs of minutiae points, the pairs of possibly matched points are searched in a larger bounding box. The result may be 1 to n, i.e. 1 template point (or input point) may have correspondence with a plurality of input points (or template points). As shown in FIG. 5, P may match P 'and P'.
When reference point pairs are searched from a plurality of reference points, the number of the found matched point pairs is large, and a plurality of obviously unmatched point pairs exist, so that further processing is required. Since there is no possibility of very large nonlinear deformations in a small local area of the fingerprint image, several (e.g., 3) already matched point pairs in the nearest neighbor are good reference objects for the candidate matching point pairs. If the position and angle difference between the candidate matching point pair and the nearest matching point pairs is large, the candidate matching point pair can be considered to be not matched, and the point pair is deleted.
In the preliminary detection process, the setting of the adaptive bounding box is critical, and a polar coordinate system with a certain point as a center is considered, and when the radius is smaller, the angular change is more sensitive, and conversely, the small angular change can cause larger deviation of the position of the detail point. The adaptive bounding box threshold is thus redesigned as follows:
radius_size＝2(α1+β1r) (17)
angle_size＝2(α2-β2r) (18)
wherein alpha is1，β1，α2，β2R is the polar radius of the template minutiae point, which is a preset empirical parameter and is all greater than zero.
Since the triangle formed by the three minutiae can better express the topological position relationship of the three minutiae, the triangle constraint in the step S4 is also an important loop in the compensation matching algorithm.
The main variables of the triangle include vertex position, angle and side length. The constraint characteristics are mainly converted from the variables, and comprise:
f1: minimum angle alphaminAnd intermediate angle alphamed
As shown in FIG. 6, let αi(i is 1, 2, 3) is three angles of a triangle formed by connecting three minutiae, and then the maximum angle alpha ismax＝max{αiH, minimum angle αmin＝min{αiH, intermediate angle αmed＝180°-αmax-αmin. Angle alphamaxThe corresponding vertex is denoted as P1Angle alphaminThe corresponding vertex is denoted as P2Angle alphamedThe corresponding vertex is denoted as P3。
F2: triangle hand shape phi
Definition of Zi＝xi+jyiIs a point (x)i，yi) In the form of a complex number. Note Z21＝Z2-Z1，Z32＝Z3-Z2，Z13＝Z1-Z3. The triangle hand is defined as: phi is sign (Z)21×Z32) I.e. the hand shape of the triangle is the sign function result of the cross product of the two complex forms.
F3: triangle three side length
L1＝|Z32|，L2＝|Z13|，L3＝|Z21|
F4: center of triangle
Zc＝(Z1+Z2+Z3)/3
Thus, the triangle constraint is defined as follows:
C1. the angle difference of the corresponding angle satisfies that | alpha-alpha' | < deltaαI.e. the angular difference between the corresponding angles of the vertices should be within a certain range. Wherein alpha and alpha' are respectively the alpha of a triangle formed by matched minutiae in two fingerprints of the same fingermaz，αmedOr alphamin。
C2. The side length difference of the triangle satisfies | L-L' | < deltaLI.e. the difference in length between the three corresponding sides of the matching triangle should be within a certain range. Wherein L and L' are respectively triangular L formed by matching minutiae in two fingerprints of the same finger1，L2Or L3。
C3. The central point position offset satisfies | ZC-Z′C|＜δCWherein Z isC，Z′CRespectively the central points of the triangles formed by the matched minutiae in the two fingerprints of the same finger.
C4. The direction difference of the minutiae corresponding to the vertexes of the triangle satisfies
I.e. the difference in direction between each vertex and the corresponding matching minutiae point should be within a certain range. Wherein
Respectively the direction of each vertex minutia of the triangle of two fingerprints of the same finger.
C5. The triangle hand shape satisfies that phi is phi, namely the hand shapes matching the triangle should be the same.
Assuming that the number of the matching point pairs found in the detail feature matching stage is n, the logarithm of the triangle formed by the matching point pairs and the n matching point pairs is C for one matching point pair to be selectedn 2. Obviously, all of C should be detectedn 2Matching between triangles is a very time consuming task. In order to reduce the amount of computation of triangle matching detection, we first connect the lines between the n matching point pairs to form Cn 2And selecting m (m is 5-10) connecting lines with the minimum matching error and the length larger than a certain threshold value from the line sides as reference line segments for triangle matching detection. As shown in fig. 7, a, B, C, D, P are minutiae points in the template image, a ', B ', C ', D ', P ' are minutiae points in the input image, and (a, a '), (B, B '), (C, C '), (D, D ') are matching corresponding points detected in the minutiae feature matching process. (P, P ') is a candidate matching corresponding point obtained after the preliminary detection, and in order to detect the candidate point pair (P, P') by using the triangle matching, it is required to determine whether the triangle constraint relationships of C1 to C5 are respectively satisfied between the triangle pair (Δ ABP, Δ a 'B' P '), (Δ ACP, Δ a' C 'P'), (Δ ADP, Δ a 'D' P '), (Δ BCP, Δ B' C 'P'), (Δ BDP, Δ B 'D' P '), (Δ CDP, Δ C' D 'P'). There are two problems with doing constraint matching in this way: 1) the calculated amount is large; 2) due to the influence of nonlinear deformation, mismatching point pairs may exist in matching point pairs obtained in the detail feature matching process, so that the accuracy of triangular matching may be influenced. For this reason, we only select the sets of line segments with the smallest matching error in the line segment pairs (AB, a ' B '), (AC, a ' C '), (AD, a ' D '), (BC, B ' C '), (BD, B ' D '), (CD, C ' D ') as the reference line segments for triangulating the candidate point pair (P, P '), e.g., (AC, a ' C '), (BD, B ' D ') as the matching references, and then only need to determine the triangle pair (Δ 4CP, Δ a ' C 'P '), (Δ BDP, Δ B' D 'P') are related. The advantages of this are: 1) the calculated amount of triangle matching is reduced; 2) more accurate matching reference can be obtained, and therefore the accuracy of triangle matching is improved.
The process of finding the triangle matched reference line segment is as follows: initialize all matching minutiae pairs wires (common C)n 2Bar connecting line) has a matching error counter of 0; matching triangle pairs for all matching minutiae (common C)n 2A triangle pair), calculating the matching error (the sum of absolute values of all errors in C1-C4) between the triangle pair and accumulating the matching error into a matching error counter of three corresponding sides of the triangle; and taking m (m is 5-10) connecting lines with the minimum matching error and the length larger than a certain threshold (such as 50) as datum line segments for performing triangular matching on the candidate matching point pairs.
After triangulating the fiducial lines, the remaining work is to triangulate each candidate pair of matched points using the fiducial lines. And the matching point pairs to be selected meeting the triangular constraint conditions C1-C4 are new matching point pairs found by the compensation algorithm.
For the newly added matching point pairs obtained above, there may be some point pairs with wrong topological relation, so for each new matching point pair, consider the triangle formed by the new matching point pair and other nearby points, and if the hand shape is not in accordance with the situation, delete the new matching point pair.
After the triangle matching compensation, if a new matching point pair can be found, the matching score needs to be recalculated.
Considering the distribution of detail points in the template and input point sets, if there is a case where there is no match in the common area in both point sets, it should have some effect on the final result, thereby introducing a penalty score:
penalty＝tmplt_count+input_count (19)
where tmplt _ count and input _ count are the number of points on the template and input minutiae sets in the common region, respectively, but not matched. It should be noted that, for a point that may need to be punished, it must be considered whether a detail point appears in another point concentrated in a radius range of a certain length, if not, it is determined to be a point that needs to be punished, otherwise, it is a point that may not be matched, and no punishment should be given.
From this, the final match score can be derived as follows:
assume that the initial number of pairs of matching minutiae is nb _ pair and the newly added number of pairs of matching minutiae is new _ pair. The new match score is then:
wherein M isc，NcThe number of feature points of the template and the input fingerprint image in the common area is respectively, the values of alpha and beta are respectively about 2.0 and 3.0, and the meaning of match _ error is the same as that of the formula (12). And comparing the matching scores with a preset matching threshold value to judge whether the matching is performed or not.
Fig. 8 is a schematic overall flow chart of the two-stage hybrid fingerprint feature matching method according to the present invention. The fingerprint feature matching method can be divided into two stages, firstly, the minutiae matching stage disclosed in the steps 803-805 in fig. 8, and the work of this stage is to match the minutiae sets of two fingerprints by the adaptive bounding box minutiae matching method for a pair of given reference points in the two fingerprint feature sets, and record the matching scores thereof, and use the size of the matching scores as the basis for whether the matching is successful or whether the triangular matching is performed subsequently.
In the triangular constraint matching stages disclosed in steps 807-813, if the matching score obtained in the detail feature matching stage is higher than a higher threshold T1If the input detail point set and the template detail point set come from the same finger, the matching process is successfully ended; if the matching score is between a lower threshold T2And a higher threshold value T1And further searching other matching minutiae pairs by utilizing a triangular matching process, and recalculating the matching score. If the recalculated match score is above a predetermined threshold Tr(T2＜Tr＜T1) If the input detail point set and the template detail point set come from the same finger, the matching process is successfully ended; if the matching score obtained in the detail feature matching stage is lower than a lower threshold value T2Then the next pair of minutiae points is selected as the reference point to restart the minutiae feature matching process. Wherein the threshold value T1，T2And TrDepending on the accuracy required of the system, this is generally obtained by a large number of experiments.
The software for implementing the method can be made into a two-stage hybrid fingerprint feature matching device in the form of firmware, and is integrated in a smart card. The two-stage hybrid fingerprint feature matching device comprises a minutiae feature matching unit, a triangular constraint matching unit and a polar coordinate conversion unit. Wherein the polar coordinate conversion unit converts the spatial positions of the relevant input minutiae and template minutiae into polar coordinates. The detailed feature matching unit has the following functions: for a pair of given reference points in the two fingerprint feature point sets, matching the minutiae point sets of the two fingerprints by using a self-adaptive bounding box minutiae feature matching method, recording matching scores of the minutiae point sets, and taking the size of the matching score as a basis for whether the matching is successful or whether triangular matching is performed subsequently. The role of the triangular constraint matching unit is as follows: if the matching score obtained in the detail feature matching stage is higher than a high threshold value, the input detail point set and the template detail point set are considered to come from the same finger, and the matching process is ended; if the matching score is between the high threshold and the low threshold, further searching other matching minutiae pairs by utilizing a triangular matching process, and recalculating the matching score; if the recalculated matching score is higher than a preset middle threshold value, the input minutiae set and the template minutiae set are considered to come from the same finger, and the matching process is ended; and if the matching score obtained in the detail feature matching stage is lower than a low threshold value, selecting the next pair of detail nodes as reference points to restart the detail feature matching process.
The two-stage hybrid fingerprint feature matching device can be built in a microprocessor of the smart card or can be independently arranged. Fig. 9 illustrates an internal logic structure of a typical contactless smart card. The microprocessor chip used therein is an MPU having logic control, management, encryption/decryption functions, etc., and the memory includes ROM, RAM, EEPROM, etc. RFC is a radio frequency transceiver circuit, mainly solves the read-write communication and power supply of the card, CAU is an encryption operation coprocessor, SL is security logic. Of course, it is entirely possible that the device described above is used in a contact smart card.
Fig. 10 illustrates an exemplary process of fingerprint feature matching of a smart card with a built-in two-stage hybrid fingerprint feature matching device. A user inputs a fingerprint of the user through a fingerprint acquisition device on the card reader by using the card reader with fingerprint identification, and the fingerprint sensor submits the fingerprint data to a feature extraction module after acquiring the fingerprint data, extracts the fingerprint feature of the time and transmits the fingerprint feature to the smart card; the smart card matches the input fingerprint features with the stored fingerprint features, if the input fingerprint features are successful, the smart card allows the user to perform subsequent operations, and if the input fingerprint features are unsuccessful, the smart card prompts the user to input the fingerprint data again. The smart card may be automatically locked if unsuccessful multiple times in a general sequence.
The above description is only for the preferred embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (16)
1. A two-stage hybrid fingerprint feature matching method is characterized by comprising the following steps:
the fingerprint feature matching method comprises two stages of detail feature matching and triangle constraint matching;
in the minutiae matching stage, for a pair of given reference points in the feature point sets of the two fingerprints, matching the minutiae sets of the two fingerprints by using a self-adaptive bounding box minutiae matching method, recording matching scores of the minutiae sets, and taking the size of the matching scores as a basis for judging whether the matching is successful or whether triangular matching is carried out subsequently;
in the triangular constraint matching stage, if the matching score obtained in the detail feature matching stage is higher than a preset high threshold value, the input detail point set and the template detail point set are considered to come from the same finger, and the matching process is ended; if the matching score is between the preset high threshold and the preset low threshold, further searching other matching minutiae pairs by utilizing a triangular matching process, and recalculating the matching score; if the recalculated matching score is higher than a preset middle threshold value, the input minutiae set and the template minutiae set are considered to come from the same finger, and the matching process is ended; and if the matching score obtained in the detail feature matching stage is lower than a preset low threshold value, selecting the next pair of detail nodes as reference points to restart the detail feature matching process.
2. The two-stage hybrid fingerprint feature matching method of claim 1, wherein:
and the detail feature matching calculation is carried out under a polar coordinate system.
3. The two-stage hybrid fingerprint feature matching method of claim 1, wherein:
in the detail feature matching stage, a given reference point of the pair is determined by the following steps:
and calculating Euclidean distance and rotation angle difference between one point in the template point set and one point in the input point set under the condition that the two points are the same type of detail points, and performing detail matching by taking the two points as a pair of reference points if the two points meet a preset threshold condition.
4. The two-stage hybrid fingerprint feature matching method of claim 1, wherein:
the detail feature matching stage further comprises the steps of:
1) transforming the input point set and other minutiae in the template point set to a polar coordinate system by taking the template minutiae and the input minutiae as reference minutiae;
2) sorting template minutiae and input minutiae in polar coordinates in the increasing direction of polar angles and connecting the template minutiae and the input minutiae into strings;
3) the strings are matched using an adaptive bounding box approach, the minutiae pair information on the match is recorded, and a match score is calculated.
5. The two-stage hybrid fingerprint feature matching method of claim 1 or 4, wherein:
the size of the adaptive bounding box is represented by radius _ size and angle _ size, which are calculated for template minutiae with a polar radius r using the following formula:
radius_size＝(α1+β1r)
angle_size＝(α2-β2r)
wherein alpha is1，β1，α2，β2R is the polar radius of the template minutiae point, which is a preset empirical parameter and is all greater than zero.
6. The two-stage hybrid fingerprint feature matching method of claim 1, wherein:
in the triangular constraint matching stage, the mixed matching score Ms is obtained by:
wherein nb _ pair is the number of matching point pairs obtained in the detail feature matching stage, new _ pair is the number of matching detail point pairs newly added in the triangle constraint matching stage, and Mc，NcThe number of the characteristic points of the template and the input fingerprint image in the public area is respectively, match _ error is the weighted average of each matching minutia to the matching error, alpha and beta are preset weighting coefficients, and penalty is a penalty score.
7. The two-stage hybrid fingerprint feature matching method of claim 6, wherein:
the penalty score penalty is the sum of the number of unmatched template and input minutiae sets in the common area.
8. The two-stage hybrid fingerprint feature matching method of claim 1, wherein:
the triangular constraint matching stage further comprises the following steps:
1) acquiring registration parameters of two fingerprint minutiae sets according to the matched point pairs;
2) registering the minutiae sets of the two fingerprints by using the calculated registration parameters;
3) taking a plurality of reference detail point pairs as a reference, carrying out primary detection, and searching other possible matching point pairs to be selected;
4) and for the preliminarily detected matching point pairs to be selected, performing triangular matching on the matching point pairs and the matched point pairs to form a triangle, and checking whether the triangle can meet the geometric constraint of the triangle.
9. The two-stage hybrid fingerprint feature matching method of claim 8, wherein:
in the step 2, the step of registering the detail point set specifically includes the following steps:
1) calculating the relative offset and the relative rotation angle between the template point set and the input point set obtained in the detail feature matching process and the matching reference point;
2) according to a transformation formula:
and transforming the input detail point set into a coordinate system where the template detail point set is located.
10. The two-stage hybrid fingerprint feature matching method of claim 8, wherein:
in the step 3, in the preliminary detection of the detail points, a self-adaptive bounding box method is used for matching the feature points;
the adaptive bounding box threshold is:
radius_size＝2(α1+β1r)
angle_size＝2(α2-β2r)
wherein alpha is1，β1，α2，β2R is the polar radius of the template minutiae point, which is a preset empirical parameter and is all greater than zero.
11. The two-stage hybrid fingerprint feature matching method of claim 8, wherein:
the step 4 is specifically divided into the following steps:
1) finding a triangle matched reference line segment in the matched minutiae pairs according to a triangle constraint condition;
2) performing triangle matching on the matching point pairs to be selected and triangles formed by the reference line segments respectively;
3) and the matching point pairs to be selected meeting the triangular constraint condition are new matching point pairs found by the compensation algorithm.
12. The two-stage hybrid fingerprint feature matching method of claim 11, wherein:
the triangle constraint conditions are as follows:
1) the angle difference between the corresponding angles of each vertex is within a preset range;
2) the length difference between three corresponding sides of the triangle is within a preset range;
3) the central point position offset is within a preset range;
4) the direction difference of the minutiae corresponding to the vertex of the triangle is within a preset range;
5) the hand shapes of the triangles are the same.
13. The two-stage hybrid fingerprint feature matching method of claim 11, wherein:
the triangle matched reference line segment is determined by the following steps:
1) initializing a matching error counter of all matching minutiae to the connecting line to be 0;
2) calculating the matching error between the matching triangle pairs formed by all the matching detail points, and accumulating the matching error into a matching error counter of three corresponding edges of the triangle;
3) and taking 5-10 connecting lines with the minimum matching error and the length larger than a preset threshold value as a reference line segment for carrying out triangular matching on the candidate matching point pair.
14. A two-stage hybrid fingerprint feature matching apparatus, comprising:
a polar coordinate conversion unit for converting the spatial positions of the relevant input detail points and the template detail points into polar coordinates;
the minutiae matching unit is used for matching minutiae sets of the two fingerprints by using a self-adaptive limit box minutiae matching method for a pair of given reference points in the two fingerprint feature point sets and recording matching scores of the minutiae sets;
the triangular constraint matching unit is used for considering that the input minutiae set and the template minutiae set come from the same finger if the matching score obtained in the minutiae matching stage is higher than a preset high threshold value, and ending the matching process; if the matching score is between the preset high threshold and the preset low threshold, further searching other matching minutiae pairs by utilizing a triangular matching process, and recalculating the matching score; if the recalculated matching score is higher than a preset middle threshold value, the input minutiae set and the template minutiae set are considered to come from the same finger, and the matching process is ended; and if the matching score obtained in the detail feature matching stage is lower than a preset low threshold value, selecting the next pair of detail nodes as reference points to restart the detail feature matching process.
15. A smart card having a microprocessor, a memory, and a communication circuit, characterized in that:
the smart card further having therein a two-stage hybrid fingerprint feature matching device as recited in claim 14.
16. A method for authenticating identity using the smart card of claim 15, wherein a user inputs his fingerprint through a fingerprint acquisition device on a card reader, and a fingerprint sensor acquires fingerprint data and submits the fingerprint data to a feature extraction module and transmits fingerprint features to the smart card; the intelligent card matches the input fingerprint characteristics with the stored fingerprint characteristics, if the matching is successful, the identity authentication is completed, and the intelligent card is characterized in that:
the smart card performs fingerprint feature matching using the two-stage hybrid fingerprint feature matching method of claim 1.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100122502A CN100447806C (en) | 2006-06-14 | 2006-06-14 | Method, device and use for matching two-stage mixed-fingerprint characteristics |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CNB2006100122502A CN100447806C (en) | 2006-06-14 | 2006-06-14 | Method, device and use for matching two-stage mixed-fingerprint characteristics |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN1983301A CN1983301A (en) | 2007-06-20 |
| CN100447806C true CN100447806C (en) | 2008-12-31 |
Family
ID=38165820
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CNB2006100122502A Expired - Fee Related CN100447806C (en) | 2006-06-14 | 2006-06-14 | Method, device and use for matching two-stage mixed-fingerprint characteristics |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN100447806C (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2613116A (en) * | 2020-08-15 | 2023-05-24 | Ibm | Scalable operators for automatic management of workloads in hybrid cloud environments |
| US12169542B2 (en) | 2021-02-05 | 2024-12-17 | Samsung Electronics Co., Ltd. | Fingerprint authentication method and fingerprint authentication device |
Families Citing this family (11)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101122950B (en) * | 2007-09-04 | 2011-01-19 | 成都方程式电子有限公司 | Fingerprint elastic deformation correction method and device |
| CN101911115B (en) * | 2007-12-26 | 2012-08-08 | 日本电气株式会社 | Inter-pattern feature corresponding device, and inter-pattern feature corresponding method used for the same |
| CN102147817B (en) * | 2011-04-27 | 2014-01-15 | 杭州晟元芯片技术有限公司 | Fingerprint singular point quick searching method |
| CN102819729B (en) * | 2012-07-17 | 2014-07-16 | 内江师范学院 | Fingerprint recognizing method |
| CN104820983B (en) * | 2015-04-23 | 2018-11-23 | 清华大学 | A kind of image matching method |
| CN105335731B (en) * | 2015-11-13 | 2020-04-17 | Oppo广东移动通信有限公司 | Fingerprint identification method and device and terminal equipment |
| EP3232369B1 (en) * | 2016-04-15 | 2021-06-16 | Nxp B.V. | Fingerprint authentication system and method |
| CN106447839A (en) * | 2016-08-26 | 2017-02-22 | 合肥若涵信智能工程有限公司 | Intelligent fingerprint lock |
| JP2019121054A (en) * | 2017-12-28 | 2019-07-22 | 株式会社東海理化電機製作所 | Fingerprint authentication device |
| CN113255579B (en) * | 2021-06-18 | 2021-09-24 | 上海建工集团股份有限公司 | Method for automatically identifying and processing construction monitoring abnormal acquisition data |
| CN115240233B (en) * | 2022-08-04 | 2025-08-29 | 浙江中正智能科技有限公司 | A deformation fingerprint matching method based on spatial clustering |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20040105065A (en) * | 2003-06-04 | 2004-12-14 | 주식회사 우량정보기술 | Strong correlated or weakly correlated grouping fingerprint matching method |
| CN1581206A (en) * | 2003-08-14 | 2005-02-16 | 上海一维科技有限公司 | Fingerprint identification method |
-
2006
- 2006-06-14 CN CNB2006100122502A patent/CN100447806C/en not_active Expired - Fee Related
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR20040105065A (en) * | 2003-06-04 | 2004-12-14 | 주식회사 우량정보기술 | Strong correlated or weakly correlated grouping fingerprint matching method |
| CN1581206A (en) * | 2003-08-14 | 2005-02-16 | 上海一维科技有限公司 | Fingerprint identification method |
Non-Patent Citations (1)
| Title |
|---|
| 自动指纹识别中的图像增强和细节匹配算法. 罗希平，田捷.软件学报，第13卷第5期. 2002 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| GB2613116A (en) * | 2020-08-15 | 2023-05-24 | Ibm | Scalable operators for automatic management of workloads in hybrid cloud environments |
| US12169542B2 (en) | 2021-02-05 | 2024-12-17 | Samsung Electronics Co., Ltd. | Fingerprint authentication method and fingerprint authentication device |
Also Published As
| Publication number | Publication date |
|---|---|
| CN1983301A (en) | 2007-06-20 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN100447806C (en) | Method, device and use for matching two-stage mixed-fingerprint characteristics | |
| CN101609499B (en) | Rapid fingerprint identification method | |
| Kumar et al. | Personal authentication using hand images | |
| Kong et al. | Palmprint feature extraction using 2-D Gabor filters | |
| Jain et al. | Filterbank-based fingerprint matching | |
| Kang et al. | Contactless palm vein recognition using a mutual foreground-based local binary pattern | |
| CN100492395C (en) | Fingerprint feature fast matching method, device and application thereof | |
| US6778685B1 (en) | Two-stage local and global fingerprint matching technique for automated fingerprint verification/identification | |
| US10552661B2 (en) | Systems and methods for biometric identification | |
| Cao et al. | Fingerprint classification by a hierarchical classifier | |
| Zhang et al. | Combining global and minutia deep features for partial high-resolution fingerprint matching | |
| CN102332084B (en) | Identity identification method based on palm print and human face feature extraction | |
| CN104123537A (en) | Rapid authentication method based on handshape and palmprint recognition | |
| Ma et al. | Using b-spline curves for hand recognition | |
| Lee et al. | Dorsal hand vein recognition based on directional filter bank | |
| Xiao et al. | Extracting palmprint ROI from whole hand image using straight line clusters | |
| Doublet et al. | Contact less hand recognition using shape and texture features | |
| Chikkerur | Online fingerprint verification system | |
| US11893825B2 (en) | Method for determining a match between a candidate fingerprint and a reference fingerprint | |
| CN116704557A (en) | Low-quality fingerprint matching method based on texture information | |
| Doroz et al. | An accurate fingerprint reference point determination method based on curvature estimation of separated ridges | |
| Jiang et al. | A method using long digital straight segments for fingerprint recognition | |
| Surajkanta et al. | A digital geometry-based fingerprint matching technique | |
| Tong et al. | Local relative location error descriptor-based fingerprint minutiae matching | |
| Nachar et al. | Hybrid minutiae and edge corners feature points for increased fingerprint recognition performance |
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 | ||
| CF01 | Termination of patent right due to non-payment of annual fee | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20081231 |