CN102262730A - Fingerprint matching method based on multiple reference point pairs - Google Patents

Abstract

The invention discloses a fingerprint matching method based on multiple reference point pairs, which comprises the following steps: determining template fingerprints and input fingerprints, acquiring the matching scores of every detailed point pair by a fingerprint matching algorithm based on three-value characteristic vector, and ranking all the detailed point pairs according to the matching scores in a descending order; selecting the first m detailed point pairs as the initial reference point pairs, and computing the rotation/translation parameters of the initial reference point pairs; selecting the reference point pairs according to the conformity, carrying out global matching by using each reference point pair as the reference to acquire a matched detailed point pair set, and fusing information of all the matched detailed point pair sets by using a voting fusion strategy; and acquiring the number of final matched detailed point pairs, computing the matching scores according to the number of final matched detailed point pairs, and judging whether the template fingerprints are matched with the input fingerprints according to the matching scores and the matching threshold. The invention overcomes and solves the problem of nonlinear deformation of fingerprints in the existing fingerprint matching algorithm, and has favorable fingerprint matching performance.

Description

A kind of based on the right finger print matching method of multiple reference points

Technical field

The present invention relates to automatic fingerprint recognition field, specifically proposed a kind of based on the right finger print matching method of multiple reference points.

Background technology

The last stage of (fingerprint collecting, fingerprint pre-service and the fingerprint matching) processing stage that fingerprint matching being three of automatic algorithm for recognizing fingerprint is one of committed step of decision algorithm performance.The problem that fingerprint matching will solve is that the characteristic information that extracts from two given fingerprint images is carried out measuring similarity, judges that finally whether these two pieces of fingerprints are from same finger.

The realization of fingerprint matching algorithm and performance all are closely related with the fingerprint characteristic of choosing.Fingerprint characteristic can be divided into global characteristics usually, local feature and fine feature.Global characteristics such as texture information, central point (being commonly called as the Core point) and trigpoint (being commonly called as the Delta point), local feature comprises end points, bifurcation (Bifurcation), isolated point, ring, island, burr and bridge etc., and fine feature is as the features such as pore on the fingerprint that extracts at the fingerprint image that obtains from high resolution sensor.According to the feature difference of choosing, fingerprint matching mainly is classified as figure coupling, texture coupling, streakline coupling and minutiae point coupling etc.Wherein minutiae point matching process expression way is simple, has made full use of the difference of fingerprint image on minutia, and matching accuracy is higher, thereby has obtained widespread use.However, to still have some problems to solve perfect for existing minutiae point matching algorithm.In existing fingerprint minutiae matching process, mostly utilize the information such as position, direction, type and minutiae point density of minutiae point during feature selecting, structure can characterize the proper vector or the neighbour structure of minutiae point characteristic, and then the matching problem of fingerprint is converted into the measuring similarity problem of proper vector or neighbour structure.Wherein, a key factor that influences the minutiae point matching performance is exactly the non-linear deformation problems of fingerprint image.

Be embodied in: because there is non-linear deformation in fingerprint, when be reference point with a pair of minutiae point to the time, big more from reference point to minutiae point far away more right position difference and direction difference, when difference acquires a certain degree, originally the minutiae point that can mate to may be because gap be excessive can't Satisfying Matching Conditions, refuse to know thereby may cause.Existing most of fingerprint minutiae matching algorithm has all been ignored this problem.Important process step as automatic algorithm for recognizing fingerprint system, the non-linear deformation problems of fingerprint image should fully be paid attention in the fingerprint matching, a good minutiae point matching algorithm, should be able to take into full account and handle the non-linear deformation problems of fingerprint, thereby eliminate the right erroneous matching of minutiae point that non-linear deformation causes, improve the minutiae point matching performance.

Summary of the invention

Technical matters to be solved by this invention is the non-linear deformation problems of fingerprint that exists in the existing fingerprint matching algorithm in order to overcome, provide a kind of based on the right minutiae point matching process of multiple reference points, this method has not only solved the non-linear deformation problems of fingerprint image to a certain extent, and has good fingerprint matching performance.

For solving the problems of the technologies described above, technical scheme of the present invention is: a kind of based on the right finger print matching method of multiple reference points, it is characterized in that comprising step:

1) determines template fingerprint and input fingerprint;

2) adopt fingerprint matching algorithm to obtain template fingerprint and import all minutiae point of fingerprint to matching score based on three value tag vectors;

3) with all minutiae point to by the matching score descending sort, choose the forward m of score right as initial reference point;

4) calculate the right rotation translation parameters of initial reference point, the minutiae point of choosing correct coupling according to consistance is right to being reference point;

5) with every pair of reference point to being benchmark, carry out the global registration of template fingerprint and input fingerprint, obtain coupling minutiae point pair set;

6) adopt the ballot convergence strategy, merge all coupling minutiae point pair set information, obtain final coupling minutiae point logarithm;

7) calculate matching score according to final coupling minutiae point logarithm;

8), judge whether template fingerprint and input fingerprint mate according to matching score and matching threshold.

Of the present invention based on the right finger print matching method of multiple reference points when carrying out fingerprint matching, need obtain many feature selecting right earlier to correct reference point:

Of the present inventionly adopt fingerprint matching algorithm (the fingerprint matching algorithm [Zhang Liming of Liming Zhang based on three value tag vectors based on three value tag vectors based on the right finger print matching method of multiple reference points, Yin Yilong. based on fingerprint matching algorithm [C] the .Chinese Conference on Pattern Recognition of three value tag vectors, 2009,563-567.]) and rotation translation parameters consistance to choose many details reference points to correct coupling right, adopt the ballot convergence strategy according to a plurality of coupling minutiae point pair set information of a plurality of reference point afterwards, obtain the fingerprint matching result obtaining.Should solve the non-linear deformation problems of fingerprint based on the right minutiae point matching algorithm of multiple reference points.The every information definition of fingerprint minutiae that coupling is used is as follows:

A. minutiae point type

General minutiae point is divided into end points and bifurcation, and the minutiae point type refers to that minutiae point belongs to end points or bifurcation.The minutiae point type of correct coupling should be consistent.Minutiae point type ω defined formula is as follows:

Wherein, when ω=0, represent that this minutiae point is an end points; When ω=1, represent that this minutiae point is a bifurcation.

B. minutiae point density

Minutiae point density i.e. the interior number of minutiae point on every side of the certain neighborhood of this minutiae point, and selecting R herein is the neighborhood of radius.Because the minutiae point information and the fingerprint quality of fingerprint image are closely related, therefore density is defined as one three value vector, promptly sparse, general, intensive, use-1,0,1 expression respectively.The density σ defined formula of minutiae point is as follows:

$\mathrm{\&sigma;}=\left\{\begin{array}{cc}-1,& \mathrm{num}<t-\mathrm{\&Delta;}\\ 0,& t-\mathrm{\&Delta;}<\mathrm{num}<t+\mathrm{\&Delta;}\\ 1,& \mathrm{num}>t+\mathrm{\&Delta;}\end{array}\right.---\left(2\right)$

Wherein, num be in the minutiae point R radius circle territory around the number of minutiae point, the threshold value of the neighborhood minutiae point number that t determines for experiment, Δ is the error of permission.When σ=-1, represent that this minutiae point density is less, belong to the sparse details point; When σ=0, expression minutiae point density is general; When σ=1, expression minutiae point density is bigger, belongs to intensive minutiae point.

C. minutiae point dispersion

Minutiae point dispersion δ is the center with this minutiae point promptly, and R is that the interior minutiae point on every side of the neighborhood of radius arrives the distance of this minutiae point with average

$\mathrm{\&delta;}=\left\{\begin{array}{cc}-1,& \frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_i}{n}<\frac{R}{2}-d_0\\ 0,& \frac{R}{2}-d_0<\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_i}{n}<\frac{R}{2}+d_0\\ 1,& \frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_i}{n}>\frac{R}{2}+d_0\end{array}\right.---\left(3\right)$

Wherein, d _iBe the distance of i minutiae point to the center minutiae point, n is the number of minutiae point in the scope, and R is the radius in circle territory, d ₀Be constant threshold, determine by experiment.When δ=-1, represent this minutiae point around minutiae point nearer from the mean distance of this minutiae point, promptly the dispersion of this minutiae point is less; When δ=0, represent that this minutiae point dispersion is placed in the middle; When δ=1, represent this minutiae point around minutiae point far away from the mean distance of this minutiae point, promptly the dispersion of this minutiae point is bigger.

D. the streakline bending direction at minutiae point place

Utilize the difference of the angle of minutiae point direction and field of direction direction to represent the bending direction of streakline, represent with λ, the minutiae point direction is meant the direction of the line of the starting point of streakline after the refinement and terminal point, field of direction direction is meant the direction of the field of direction of end points (being starting point), the positive and negative bending direction that reflects streakline of the angle of both direction, no matter how image rotates the minutiae point direction all the time in the inboard of crestal line, field of direction direction is all the time in the outside of crestal line.If field of direction direction deducts the value of minutiae point direction for just, represent that streakline is bent downwardly, for negative, the expression streakline is bent upwards, otherwise streakline convergence level.

Wherein, Ω is an angle threshold.When λ=-1, represent then that streakline is bent upwards and be last arc; Near 0 o'clock, the acquiescence streakline was a horizontal direction; When λ=1, represent that then it is arc down that streakline is bent downwardly.

In the described step 3), to all minutiae point to according to matching score according to descending sort, be expressed as χ ₁, χ ₂..., χ _m..., χ _I+1, χ _I+2..., χ _I+n, forward more expression is that correctly to mate the right possibility of minutiae point big more, coming a most forward m minutiae point to (being χ ₁, χ ₂..., χ _m) right as initial reference point, χ wherein _iRepresent i minutiae point, m is the minutiae point number;

In the described step 4), the rotation translation parameters be to the input fingerprint with respect to the template fingerprint registration after the quantification of translational movement and rotation amount.Because angle, the position not equal factor of finger when fingerprint capturer becomes the shadow zone to push, the input fingerprint has certain rotation translational movement with respect to template fingerprint, and the rotation translation parameters promptly is the measurement to this amount.Calculating the right rotation translation parameters of initial reference point is: initial reference point centering, at first establish a minutiae point T (X of template fingerprint _T, Y _T, θ _T), X wherein _TThe horizontal ordinate of representation template fingerprint minutiae on fingerprint image, Y _TThe ordinate of representation template fingerprint minutiae on fingerprint image, θ _TThe direction of representation template fingerprint minutiae on fingerprint image, and θ _T∈ (0, π), establish input fingerprint one minutiae point I (X again _I, Y _I, θ _I), X wherein _IThe horizontal ordinate of expression input fingerprint minutiae on fingerprint image, Y _IThe ordinate of expression input fingerprint minutiae on fingerprint image, θ _IThe direction of expression input fingerprint minutiae on fingerprint image, and θ _I∈ (0, π); Then minutiae point is to (T, I) rotation translation parameters is (Δ X, Δ Y, Δ θ), (Δ X wherein, Δ Y) expression input fingerprint minutiae is with respect to the offset of template fingerprint minutiae point on horizontal ordinate, and Δ θ represents to import fingerprint minutiae with respect to the skew of template fingerprint minutiae point direction, and every component computing formula of rotation translation parameters is as follows:

Δθ＝θ _I-θ _T (5)

ΔX＝X _I×cos(Δθ)+Y _I×sin(Δθ)-X _T (6)

ΔY＝-X _I×sin(Δθ)+Y _I×cos(Δθ)-Y _T (7)

Minutiae point corresponding on input fingerprint and the template fingerprint same position is to AB (Δ X _AB, Δ Y _AB, Δ θ _AB) and CD (Δ X _CD, Δ Y _CD, Δ θ _CD) between Euclidean distance Dis (AB, CD) formula is

$\mathrm{Dis}(\mathrm{AB},\mathrm{CD})=\sqrt{({({\mathrm{\&Delta;X}}_{\mathrm{AB}}-{\mathrm{\&Delta;X}}_{\mathrm{CD}})}^2+{({\mathrm{\&Delta;Y}}_{\mathrm{AB}}-{\mathrm{\&Delta;Y}}_{\mathrm{CD}})}^2+{({\mathrm{\&Delta;\&theta;}}_{\mathrm{AB}}-{\mathrm{\&Delta;\&theta;}}_{\mathrm{CD}})}^2)}---\left(8\right)$

These two pairs of minutiae point are to AB (Δ X _AB, Δ Y _AB, Δ θ _AB) and CD (Δ X _CD, Δ Y _CD, Δ θ _CD) have rotation translation parameters consistance and then should satisfy following formula

Dis(AB，CD)＜Ψ (9)

Wherein Ψ is apart from experimental threshold values.According to formula (5)-(8), calculate right rotation translation parameters of initial reference point and the Euclidean distance between parameter like this,, utilize rotation translation parameters consistance to choose the minutiae point of correct coupling right as the reference point according to formula (9).

In the described step 5) template fingerprint and the input fingerprint global registration for respectively with every pair of reference point to being reference point, the global registration that utilization is carried out template fingerprint and input fingerprint based on the fingerprint matching algorithm and the rotation translation parameters consistance of three value tag vectors, wherein satisfy three value tags vector matching condition and the conforming minutiae point of rotation translation parameters simultaneously to being that correct coupling minutiae point is right, all obtain the coupling minutiae point pair set of a correspondence at every pair of reference point.

Afterwards, right coupling minutiae point set is merged to reference point:

In described step 6), adopt the ballot convergence strategy, to the fusion of voting, obtain the final minutiae point coupling logarithm PairNum of template fingerprint and input fingerprint, at all coupling minutiae point if the right votes T＞T of certain minutiae point ₀, then think this minutiae point to being that the minutiae point of correct coupling is right, i.e. PairNum=PairNum+1, wherein, T ₀Relevant with the number of coupling set, be experimental threshold values.

In the described step 7), according to final minutiae point coupling logarithm, the final score Socre of calculation template fingerprint and input fingerprint, Socre ∈ [0,100] wherein, computing formula is as follows:

$\mathrm{Socre}=\frac{\mathrm{PairNum}}{\sqrt{M*N}}\&times;100---\left(10\right)$

Wherein, M, N are respectively the minutiae point number of template fingerprint and input fingerprint.

In the described step 8), according to matching score and matching threshold, judge whether template fingerprint and input fingerprint mate, when Socre 〉=μ, think that then two width of cloth fingerprints successfully mate, do not match that wherein μ is matching threshold otherwise be considered as two width of cloth fingerprints, matching threshold can be adjusted the specific requirement of reject rate and misclassification rate according to practical application, generally gets 55.

This patent proposes a kind of based on the right minutiae point matching process of multiple reference points, it is right to putting as reference to choose many minutiae point to correct coupling, respectively with each to reference point to being benchmark, carry out the fingerprint global registration, obtain a plurality of coupling minutiae point pair sets, select the ballot convergence strategy, promptly to the minutiae point that occurred to combination, maximum polls of voting are identical with the set number, the ballot that certain a pair of minutiae point obtains is many more, illustrate that this minutiae point is big more to being the right possibility of correct minutiae point of mating, on the contrary more little, thus guarantee the right accuracy of coupling minutiae point.Calculating the matching score stage, set matching threshold, judge the fingerprint matching result, multiple reference points to coupling made full use of many groups based on different, correct match point to the mutual relationship between a plurality of coupling minutiae point set that obtain, overcome the non-linear deformation problems of the fingerprint that exists in the existing fingerprint matching algorithm, not only solve the non-linear deformation problems of fingerprint image to a certain extent, and had good fingerprint matching performance.

Description of drawings

The present invention is further described below in conjunction with drawings and Examples:

Accompanying drawing is that the present invention is a kind of based on the right finger print matching method schematic diagram of multiple reference points.

Embodiment

Below in conjunction with drawings and Examples, further set forth the present invention.Should be understood that these embodiment only to be used to the present invention is described and be not used in and limit the scope of the invention.Should be understood that in addition those skilled in the art can make various changes or modifications the present invention after the content of having read the present invention's instruction, these equivalent form of values fall within the application's appended claims institute restricted portion equally.

As shown in drawings, technical scheme of the present invention is: a kind of based on the right finger print matching method of multiple reference points, comprise step:

1) determines template fingerprint and input fingerprint;

2) adopt fingerprint matching algorithm to obtain template fingerprint and import all minutiae point of fingerprint to matching score based on three value tag vectors;

3) with all minutiae point to by the matching score descending sort, choose the forward m of score right as initial reference point;

4) calculate the right rotation translation parameters of initial reference point, the minutiae point of choosing correct coupling according to consistance is right to being reference point;

5) with every pair of reference point to being benchmark, carry out the global registration of template fingerprint and input fingerprint, obtain coupling minutiae point pair set;

6) adopt the ballot convergence strategy, merge all coupling minutiae point pair set information, obtain final coupling minutiae point logarithm;

7) calculate matching score according to final coupling minutiae point logarithm;

8), judge whether template fingerprint and input fingerprint mate according to matching score and matching threshold.

Wherein, of the present invention based on the right finger print matching method of multiple reference points when carrying out fingerprint matching,

At first, obtain many to the right feature selecting of correct minutiae point reference point:

The every information definition of the fingerprint minutiae of using is as follows

A. minutiae point type

General minutiae point is divided into end points and bifurcation,, minutiae point type ω defined formula is as follows:

Wherein, when ω=0, represent that this minutiae point is an end points; When ω=1, represent that this minutiae point is a bifurcation.

B. minutiae point density

Minutiae point density i.e. the interior number of minutiae point on every side of the certain neighborhood of this minutiae point, and selecting R (R generally gets 5 ridge distances, about 50 pixels) herein is the neighborhood of radius.Minutiae point density is defined as one three value vector, promptly sparse, general, intensive, use-1,0,1 expression respectively.The density σ defined formula of minutiae point is as follows:

$\mathrm{\&sigma;}=\left\{\begin{array}{cc}-1,& \mathrm{num}<t-\mathrm{\&Delta;}\\ 0,& t-\mathrm{\&Delta;}<\mathrm{num}<t+\mathrm{\&Delta;}\\ 1,& \mathrm{num}>t+\mathrm{\&Delta;}\end{array}\right.---\left(2\right)$

Wherein, num (value of num is between 0-8) be in the minutiae point R radius circle territory around the number of minutiae point, the threshold value of the neighborhood minutiae point number that t (t generally gets 3) determines for experiment, Δ (Δ generally gets 1) is the error of permission.When σ=-1, represent that this minutiae point density is less, belong to the sparse details point; When σ=0, expression minutiae point density is general; When σ=1, expression minutiae point density is bigger, belongs to intensive minutiae point.

C. minutiae point dispersion

Minutiae point dispersion δ is the center with this minutiae point promptly, and R is that the interior minutiae point on every side of the neighborhood of radius arrives the distance of this minutiae point with average.

$\mathrm{\&delta;}=\left\{\begin{array}{cc}-1,& \frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_i}{n}<\frac{R}{2}-d_0\\ 0,& \frac{R}{2}-d_0<\frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_i}{n}<\frac{R}{2}+d_0\\ 1,& \frac{\underset{i=1}{\overset{n}{\mathrm{\&Sigma;}}}{d}_i}{n}>\frac{R}{2}+d_0\end{array}\right.---\left(3\right)$

Wherein, d _iBe the distance of i minutiae point to the center minutiae point, n (n is generally 4) is the number of minutiae point in the scope, and R (R generally gets 5 ridge distances) is the radius in circle territory, d ₀(d ₀Generally get 2 ridge distances) be constant threshold, determine by experiment.When δ=-1, represent this minutiae point around minutiae point nearer from the mean distance of this minutiae point, promptly the dispersion of this minutiae point is less; When δ=0, represent that this minutiae point dispersion is placed in the middle; When δ=1, represent this minutiae point around minutiae point far away from the mean distance of this minutiae point, promptly the dispersion of this minutiae point is bigger.

D. the streakline bending direction at minutiae point place

Utilize the difference of the angle of minutiae point direction and field of direction direction to represent the bending direction of streakline, represent with λ, the value that field of direction direction deducts the minutiae point direction is for just, and the expression streakline is bent downwardly, and for negative, the expression streakline is bent upwards, otherwise streakline convergence level.

Wherein, Ω (Ω generally gets 5 degree) is angle threshold.When λ=-1, represent then that streakline is bent upwards and be last arc; Near 0 o'clock, the acquiescence streakline was a horizontal direction; When λ=1, represent that then it is arc down that streakline is bent downwardly.

E. rotate the translation parameters consistance

Calculating the right rotation translation parameters of initial reference point is: initial reference point centering, at first establish a minutiae point T (X of template fingerprint _T, Y _T, θ _T), X wherein _TThe horizontal ordinate of representation template fingerprint minutiae on fingerprint image, Y _TThe ordinate of representation template fingerprint minutiae on fingerprint image, θ _TThe direction of representation template fingerprint minutiae on fingerprint image, and θ _T∈ (0, π), establish input fingerprint one minutiae point I (X again _I, Y _I, θ _I), X _IWherein the horizontal ordinate of fingerprint minutiae on fingerprint image, Y are imported in expression _IThe ordinate of expression input fingerprint minutiae on fingerprint image, θ _IThe direction of expression input fingerprint minutiae on fingerprint image, and θ _I∈ (0, π); Then minutiae point is to (T, I) rotation translation parameters is (Δ X, Δ Y, Δ θ), (Δ X wherein, Δ Y) expression input fingerprint minutiae is with respect to the offset of template fingerprint minutiae point on horizontal ordinate, and Δ θ represents to import fingerprint minutiae with respect to the skew of template fingerprint minutiae point direction, and every component computing formula of rotation translation parameters is as follows:

Δθ＝θ _I-θ _T (5)

ΔX＝X _I×cos(Δθ)+Y _I×sin(Δθ)-X _T (6)

ΔY＝-X _I×sin(Δθ)+Y _I×cos(Δθ)-Y _T (7)

Minutiae point corresponding on input fingerprint and the template fingerprint same position is to AB (Δ X _AB, Δ Y _AB, Δ θ _AB) and CD (Δ X _CD, Δ Y _CD, Δ θ _CD) between Euclidean distance Dis (AB, CD) formula is

$\mathrm{Dis}(\mathrm{AB},\mathrm{CD})=\sqrt{({({\mathrm{\&Delta;X}}_{\mathrm{AB}}-{\mathrm{\&Delta;X}}_{\mathrm{CD}})}^2+{({\mathrm{\&Delta;Y}}_{\mathrm{AB}}-{\mathrm{\&Delta;Y}}_{\mathrm{CD}})}^2+{({\mathrm{\&Delta;\&theta;}}_{\mathrm{AB}}-{\mathrm{\&Delta;\&theta;}}_{\mathrm{CD}})}^2)}---\left(8\right)$

These two pairs of minutiae point are to AB (Δ X _AB, Δ Y _AB, Δ θ _AB) and CD (Δ X _CD, Δ Y _CD, Δ θ _CD) have rotation translation parameters consistance and then should satisfy following formula

Dis(AB，CD)＜Ψ (9)

Wherein Ψ (Ψ approximately gets 20) is apart from experimental threshold values.According to formula (5)-(8), calculate right rotation translation parameters of initial reference point and the Euclidean distance between parameter like this,, utilize rotation translation parameters consistance to choose the minutiae point of correct coupling right as the reference point according to formula (9).

Carry out fingerprint matching afterwards:

1) determines template fingerprint and input fingerprint;

2) employing obtains template fingerprint and imports all minutiae point of fingerprint to matching score based on the fingerprint matching algorithm of three value tag vectors, and wherein the characteristic formula of coupling use is shown in (1)-(4); Specific as follows: with the minutiae point type is example, if two minutiae point belong to same type, then the matching score of these two minutiae point adds 1, in like manner, for minutiae point density, if two minutiae point belong to the minutiae point of degree of the same race, promptly be both intensive minutiae point, be both general minutiae point or be both sparse details point then the matching score of two minutiae point add 1, otherwise score is constant, continues relatively next feature;

3) with all minutiae point to by the matching score descending sort, be expressed as χ ₁, χ ₂..., χ _m..., χ _I+1, χ _I+2..., χ _I+n, forward more expression is that correctly to mate the right possibility of minutiae point big more, choosing the forward m of score (m generally gets 30) (is χ ₁, χ ₂..., χ _m) right as initial reference point; χ wherein _iRepresent i minutiae point, m is the minutiae point number;

4) calculate the right rotation translation parameters of initial reference point, according to formula (5)-(8), calculate right rotation translation parameters of initial reference point and the Euclidean distance between parameter,, utilize the minutiae point of the correct coupling of rotation translation parameters consistance right being reference point according to formula (9);

5) with every pair of reference point to being benchmark, carry out the global registration of template fingerprint and input fingerprint, the global registration that utilization is carried out template fingerprint and input fingerprint based on the fingerprint matching algorithm and the rotation translation parameters consistance of three value tag vectors, wherein satisfy three value tags vector matching condition and the conforming minutiae point of rotation translation parameters simultaneously to being that correct coupling minutiae point is right, all obtain the coupling minutiae point pair set of a correspondence at every pair of reference point.

6) adopt the ballot convergence strategy, to the fusion of voting, obtain the final minutiae point coupling logarithm PairNum of template fingerprint and input fingerprint, if the right votes T＞T of certain minutiae point at all coupling minutiae point ₀, then think this minutiae point to being that the minutiae point of correct coupling is right, i.e. PairNum=PairNum+1, wherein, T is the actual votes that obtains, generally between 1-15; T ₀Relevant with the number of coupling set, be experimental threshold values, generally get the numerical value between the 6-15, according to knowing and miss the requirement of knowing and can dynamically adjust to refusing.

7) according to the final matching score Socre of final coupling minutiae point logarithm calculation template fingerprint and input fingerprint, Socre ∈ [0,100] wherein, computing formula is as follows:

$\mathrm{Socre}=\frac{\mathrm{PairNum}}{\sqrt{M*N}}\&times;100---\left(10\right)$

Wherein, M, N are respectively the minutiae point number of template fingerprint and input fingerprint, and the value of M, N is between 30-100;

8) according to matching score and matching threshold, judge whether template fingerprint and input fingerprint mate, when Socre 〉=μ, think that then two width of cloth fingerprints successfully mate, otherwise being considered as two width of cloth fingerprints does not match, wherein μ is matching threshold, and matching threshold can be adjusted the specific requirement of reject rate and misclassification rate according to practical application, generally gets 55.

More than show and described ultimate principle of the present invention and principal character and advantage of the present invention.The technician of the industry should understand; the present invention is not restricted to the described embodiments; that describes in the foregoing description and the instructions just illustrates principle of the present invention; without departing from the spirit and scope of the present invention; the present invention also has various changes and modifications, and these changes and improvements all fall in the claimed scope of the invention.The claimed scope of the present invention is defined by appending claims and equivalent thereof.

All from the present invention is to devise, the structure conversion of having done without creative work all drops within protection scope of the present invention.

Claims (6)

1. one kind based on the right finger print matching method of multiple reference points, it is characterized in that comprising step:

1) determines template fingerprint and input fingerprint;

2) adopt fingerprint matching algorithm to obtain template fingerprint and import all minutiae point of fingerprint to matching score based on three value tag vectors;

3) with all minutiae point to by the matching score descending sort, choose the forward m of score right as initial reference point;

4) calculate the right rotation translation parameters of initial reference point, the minutiae point of choosing correct coupling according to consistance is right to being reference point;

5) with every pair of reference point to being benchmark, carry out the global registration of template fingerprint and input fingerprint, obtain coupling minutiae point pair set;

6) adopt the ballot convergence strategy, merge all coupling minutiae point pair set information, obtain final coupling minutiae point logarithm;

7) calculate matching score according to final coupling minutiae point logarithm;

8), judge whether template fingerprint and input fingerprint mate according to matching score and matching threshold.

2. as claimed in claim 1 a kind ofly it is characterized in that, calculate the right rotation translation parameters of initial reference point in the described step 4) and be:, at first establish a minutiae point T (X of template fingerprint initial reference point centering based on the right finger print matching method of multiple reference points _T, Y _T, θ _T), X wherein _TThe horizontal ordinate of representation template fingerprint minutiae on fingerprint image, Y _TThe ordinate of representation template fingerprint minutiae on fingerprint image, θ _TThe direction of representation template fingerprint minutiae on fingerprint image, and θ _T∈ (0, π), establish input fingerprint one minutiae point I (X again _I, Y _I, θ _I), X wherein _IThe horizontal ordinate of expression input fingerprint minutiae on fingerprint image, Y _IThe ordinate of expression input fingerprint minutiae on fingerprint image, θ _IThe direction of expression input fingerprint minutiae on fingerprint image, and θ _I∈ (0, π); Then minutiae point is to (T, I) rotation translation parameters is (Δ X, Δ Y, Δ θ), (Δ X wherein, Δ Y) expression input fingerprint minutiae is with respect to the offset of template fingerprint minutiae point on horizontal ordinate, and Δ θ represents to import fingerprint minutiae with respect to the skew of template fingerprint minutiae point direction, and every component computing formula of rotation translation parameters is as follows:

Δθ＝θ _I-θ _T (5)

ΔX＝X _I×cos(Δθ)+Y _I×sin(Δθ)-X _T (6)

ΔY＝-X _I×sin(Δθ)+Y _I×cos(Δθ)-Y _T (7)

Minutiae point corresponding on input fingerprint and the template fingerprint same position is to AB (Δ X _AB, Δ Y _AB, Δ θ _AB) and CD (Δ X _CD, Δ Y _CD, Δ θ _CD) between Euclidean distance Dis (AB, CD) formula is

$\mathrm{Dis}(\mathrm{AB},\mathrm{CD})=\sqrt{({({\mathrm{\&Delta;X}}_{\mathrm{AB}}-{\mathrm{\&Delta;X}}_{\mathrm{CD}})}^2+{({\mathrm{\&Delta;Y}}_{\mathrm{AB}}-{\mathrm{\&Delta;Y}}_{\mathrm{CD}})}^2+{({\mathrm{\&Delta;\&theta;}}_{\mathrm{AB}}-{\mathrm{\&Delta;\&theta;}}_{\mathrm{CD}})}^2)}---\left(8\right)$

These two pairs of minutiae point are to AB (Δ X _AB, Δ Y _AB, Δ θ _AB) and CD (Δ X _CD, Δ Y _CD, Δ θ _CD) have rotation translation parameters consistance and then should satisfy following formula

Dis(AB，CD)＜Ψ (9)

It is right that the minutiae point of utilizing rotation translation parameters consistance to choose correct coupling is put the conduct reference, and wherein Ψ is apart from experimental threshold values.

3. as claimed in claim 1 a kind of based on the right finger print matching method of multiple reference points, it is characterized in that, in the described step 5) template fingerprint and the input fingerprint global registration for respectively with every pair of reference point to being reference point, the global registration that utilization is carried out template fingerprint and input fingerprint based on the fingerprint matching algorithm and the rotation translation parameters consistance of three value tag vectors, wherein satisfy three value tags vector matching condition and the conforming minutiae point of rotation translation parameters simultaneously to being that correct coupling minutiae point is right, all obtain the coupling minutiae point pair set of a correspondence at every pair of reference point.

4. as claimed in claim 1 a kind of based on the right finger print matching method of multiple reference points, it is characterized in that, in the described step 6), adopt the ballot convergence strategy, mate minutiae point to the fusion of voting at all, obtain the final minutiae point coupling logarithm PairNum of template fingerprint and input fingerprint, if the right votes T＞T of certain minutiae point ₀, then think this minutiae point to being that the minutiae point of correct coupling is right, i.e. PairNum=PairNum+1, wherein, T ₀Relevant with the number of coupling set, be experimental threshold values.

5. as claimed in claim 4 a kind ofly it is characterized in that based on the right finger print matching method of multiple reference points is in the described step 7), according to final minutiae point coupling logarithm, the final score Socre of calculation template fingerprint and input fingerprint, wherein Socre ∈ [0,100], computing formula is as follows:

$\mathrm{Socre}=\frac{\mathrm{PairNum}}{\sqrt{M*N}}\&times;100---\left(10\right)$

Wherein, M, N are respectively the minutiae point number of template fingerprint and input fingerprint.

6. as claimed in claim 5 a kind of based on the right finger print matching method of multiple reference points, it is characterized in that, in the described step 8), according to matching score and matching threshold, judge whether template fingerprint and input fingerprint mate, when Socre 〉=μ, think that then two width of cloth fingerprints successfully mate, do not match otherwise be considered as two width of cloth fingerprints, wherein μ is matching threshold, generally gets 55.

Images