CN102663372A - Iris texture normalization method based on dual-spring model - Google Patents

Abstract

The invention relates to an iris texture normalization method based on a dual-spring model. The method includes that ration of pupil radius to iris radius is set under an iris standard state, and ratio of the width of a near-pupil to the width of (near-pupil area + far-pupil area) is set under an iris standard state; the number of required sampling lines of the near-pupil area and the far-pupil area is set; the near-pupil area and the far-pupil area are modeled into series connected springs with different elastic coefficients; lower-pupil radius, near-pupil area width and far-pupil area width under a standard state are calculated according to iris actual radius; inner circumference and outer circumference of an iris are divided equally, correspondingly and respectively, coordinates of a certain point on the edge of a pupil and coordinates of a point arranged on the edge of the iris and corresponding to the point on the edge of the pupil are calculated, and distance between the two points is calculated; when the iris is under the non-standard state, the width of the near-pupil area and the far-pupil area is calculated, and linear sampling is conducted on the near-pupil area and the far-pupil area; and sampling points are normalized into a rectangular image. Compared with the existing non-linear normalization method, the method is simple and practical, can reduce errors caused by pupil contracting and expanding, and can increase recognition rate of iris recognition.

Description

A kind of iris texture method for normalizing based on the dual spring model

Technical field

The present invention relates to the iris distinguishment technical field, particularly a kind of iris texture method for normalizing.

Background technology

Along with the development of network and infotech, personal identification is differentiated and is obtained unprecedented attention, also is faced with more and more serious test.Living things feature recognition (Biometrics) be with the intrinsic various physiology of human body and morphological feature as the identification medium, thereby reach unique identification personal identification, carry out the emerging research subject of personal identification.Compare with traditional identity authentication means, based on the identity authentication technology of living things feature recognition have be difficult for forgeing or lose, advantage such as anti-counterfeiting performance is good, carry.

Iris, as important biological characteristic, be used for identity differentiate have natural in advantages such as protection feature, high complexity, high stability, high antifalsifications.Compare with the other biological feature identification technique, iris recognition is one of the highest method of accuracy rate.Therefore the identity authentication technique based on iris obtains academia and the increasing attention of business circles.The annulus of iris between human eye black pupil and white sclera; Wherein present a kind of radial structure from inside to outside; Comprise many interlaced fine features that are similar to shapes such as spot, microgroove, crown, crypts, be called the texture information of iris.

As shown in Figure 1, existing iris recognition identity identifying method comprises following step: iris image acquiring step, image pre-treatment step, Iris Location step, iris normalization step, characteristic extraction step and character matching step.Wherein, the iris image acquiring step is used to gather and includes the iris image of supply discerning that enriches detailed information; The image pre-treatment step is used for judging whether the iris image that collects has iris, and whether whether image is clear and be the live body collection; The Iris Location step is used to locate the inside and outside circle and the eyelid of iris; Iris normalization step is used for iris image is normalized into the rectangular image of fixed resolution, wherein, has comprised whole available iris texture information in the rectangular image; Characteristic extraction step is responsible for iris texture information is encoded to the suitable recognized patterns information that can be used to; Character matching step is used for two iris feature codings are compared, to determine whether being to come from same eyes.

At present, the iris method for normalizing in most of iris recognition identity identifying methods all is the linear normalization method of iris texture, promptly all is the rectangle that iris is mapped to fixed size from annular with the mode of linear mapping.These class methods are carried out linear five equilibrium to the corresponding point on the iris internal and external circumference; Yet when taking iris image; Because the variation that ambient light is shone etc., can cause the convergent-divergent of pupil, the convergent-divergent of pupil can cause that iris texture corresponding convergent-divergent takes place changes; This convergent-divergent is uneven, is non-linear.Therefore, the linear normalization method of iris texture can't be eliminated this inhomogeneous convergent-divergent, thereby can reduce the accuracy rate of follow-up iris recognition.Wherein, the inhomogeneous convergent-divergent of iris texture is by the decision of the physilogical characteristics of iris.Ophthalmology research shows that the contraction of pupil and amplification mainly are the circular muscle fiber and the coefficient result of radial muscle fibre of iris.The circular muscle Fiber Distribution is near the inward flange of iris, and radial muscle fibre is distributed near the iris outward flange.Wherein, the circular muscle fiber is the main part that constitutes iris texture.When pupil receives that light stimulation shrinks and expands; Circular muscle fiber near the iris inward flange is significantly flexible; And the flexible amplitude of peripheral radial muscle fibre is very little; The texture of iris inward flange can significantly compress thereupon or uphold this moment, and what variation the outer peripheral texture of iris does not almost have.So the contraction of iris texture and expansion are not uniform linearity, but nonlinear, do not have affine unchangeability.As shown in Figure 2, can know and see nearly pupil district and the different flexible situation in pupil district far away.

Because the deficiency of the linear normalization method of iris texture, people have proposed the non-linear normalizing method of iris texture.Be called " non-linear normalizing method of body iris texture " like name; Publication number is that the patent of CN1776710A discloses a kind of nonlinear iris texture method for normalizing; This method at first utilizes arc structure to carry out radially non-linear normalizing annular iris region, makes all iris images have identical pupil zoom degree.To be called " iris texture normalization processing method " publication number be the patent of CN1445714A to name for another example; Its method that adopts is to adopt to correct function; With iris outer edge radius ratio is parameter, carries out non-linear sampling radially, thereby the inhomogeneous convergent-divergent of iris texture is proofreaied and correct.Above-mentioned two kinds of non-linear normalizing methods, more traditional linear normalization method improves to some extent, and has all adopted relevant mathematical model that the non-linear compression of iris is corrected; But above-mentioned two kinds of computing method are all comparatively complicated; Calculated amount is big, is difficult for through engineering approaches, and actual effect is accurate inadequately.

Summary of the invention

For solving one of problem of above-mentioned existing method; The invention provides a kind of iris texture method for normalizing based on spring model; It is more accurately reliable to compare existing linear normalization method, and it is more simple and practical to compare existing non-linear normalizing method, can be good at reducing because the error that convergent-divergent brought of pupil; Provide convenience for follow-up iris feature, go forward side by side to improve the overall discrimination of iris recognition.

The present invention is a kind of iris texture method for normalizing based on the dual spring model, comprises the steps:

Step 1 is set the ratio dRNToall of nearly pupil sector width under ratio dRPTol and the iris standard state of pupil radius and iris radius under the iris standard state and (nearly pupil district+far pupil district) width; Set the line number m that nearly pupil district and pupil district far away need sample ₁And m ₂, m wherein ₁, m ₂For greater than 0 integer, and m ₁+ m ₂=m; Nearly the pupil district is modeled as the spring series connection with unit elasticity coefficient k 1 and k2, wherein k respectively with pupil district far away ₁＜k ₂

Step 2 is obtained the iris radius R of the iris image of being gathered _I, according to said dRPTol, dRNToall and iris radius R _I, calculate the pupil radius R under the standard state _TO, the width dA0 in nearly pupil district and the width dB0 in pupil district far away;

Step 3, with the iris internal and external circumference of the iris image of being gathered corresponding respectively be divided into n part, calculate the coordinate (x of certain point on the pupil edge _Pi, y _Pi) and the iris edge on coordinate (x to putting _Ij, y _Ij), according to (x _Pi, y _Pi) and (x _Ij, y _Ij) calculate between said pupil edge point and the iris marginal point apart from dI, wherein, n is the integer greater than 0;

Step 4 when dI ≠ dA0+dB0, judges that the iris image of being gathered is in the iris off-rating, calculates the width dA1 in the nearly pupil district the iris off-rating under and the width dB1 in pupil district far away according to following formula:

$\frac{k1}{\mathrm{dA}0}(\mathrm{dA}0-\mathrm{dA}1)=\frac{k2}{\mathrm{dB}0}(\mathrm{dB}0-\mathrm{dB}1)$

dA1+dB1＝dI；

Step 5 according to dA1 that calculates and dB1, is carried out m respectively to nearly pupil district under the iris off-rating and pupil district far away ₁Row and m ₂Line linearity is sampled and is obtained sampled point;

Step 6 is normalized into the rectangular image that resolution is m * n with said sampled point through bilinear interpolation.

Preferably, in step 1, the line number m that said nearly pupil district need sample ₁=m * dRPTol, the line number m that said pupil far away district need sample ₂=m-m ₁

Preferably, in step 2, the width dA0=dRNToall in said nearly pupil district * (1-dRPTol) * R _I, width dB0=(the 1-dRPTol) * R in said pupil far away district _I-dA0.

Preferably, in step 3,

$x_{\mathrm{pi}}=x_{\mathrm{pI}}+R_{\mathrm{pupil}}*\cos (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

$y_{\mathrm{pi}}=y_{\mathrm{pI}}+R_{\mathrm{pupil}}*\sin (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

Wherein, (x _PI, y _PI) be pupil center's coordinate points, R _PupilBe the pupil radius.

$x_{\mathrm{Ij}}=x_{\mathrm{pI}}+R_I*\cos (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

$y_{\mathrm{Ij}}=y_{\mathrm{pI}}+R_I*\sin (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1]$

Wherein, (x _PI, y _PI) be pupil center's coordinate points, R _IBe iris radius.

Distance between said pupil edge point and the iris marginal point:

$\mathrm{dI}=\sqrt[2]{{(x_{\mathrm{Ij}}-x_{\mathrm{pi}})}^2+{(y_{\mathrm{Ij}}-y_{\mathrm{pi}})}^2}.$

Through the iris texture method for normalizing based on spring model provided by the invention; Can realize more accurately more reliable than existing linear normalization method; More simple and practical than existing non-linear normalizing method; Can be good at reducing because the error that convergent-divergent brought of pupil is provided convenience for follow-up iris feature, go forward side by side to improve the overall discrimination of iris recognition.

Description of drawings

Above-mentioned and/or additional aspect of the present invention and advantage are from obviously with easily understanding becoming the description of embodiment below in conjunction with accompanying drawing, wherein:

Fig. 1 is the system architecture synoptic diagram of iris identification method.

Fig. 2 is under different external conditions, the variation synoptic diagram in nearly pupil district and pupil district far away under the pupil different zoom degree.

Fig. 3 is the iris texture method for normalizing process flow diagram based on spring model of the present invention.

Fig. 4 A is the synoptic diagram that obtains the iris radius of the iris image of being gathered of the present invention.

Fig. 4 B is the synoptic diagram of the iris internal and external circumference of the iris image of gathering of the present invention.

Fig. 4 C is the synoptic diagram that is normalized into the rectangular image of m * n of the present invention.

Embodiment

Describe embodiments of the invention below in detail, the example of said embodiment is shown in the drawings, and wherein identical from start to finish or similar label is represented identical or similar elements or the element with identical or similar functions.Be exemplary through the embodiment that is described with reference to the drawings below, only be used to explain the present invention, and can not be interpreted as limitation of the present invention.

Fig. 3 is the iris texture method for normalizing process flow diagram based on spring model of the present invention.Referring to Fig. 3; The present invention is a kind of iris texture method for normalizing based on the dual spring model; Comprise the steps: step 1, set the ratio dRNToall of nearly pupil sector width under ratio dRPTol and the iris standard state of pupil radius and iris radius under the iris standard state and (nearly pupil district+far pupil district) width; Set the line number m that nearly pupil district and pupil district far away need sample ₁And m ₂, m wherein ₁, m ₂For greater than 0 integer, and m ₁+ m ₂=m; Nearly the pupil district is modeled as the spring series connection with unit elasticity coefficient k 1 and k2, wherein k respectively with pupil district far away ₁＜k ₂Step 2 is obtained the iris radius R of the iris image of being gathered _I, according to said dRPTol, dRNToall and iris radius R _I, calculate the pupil radius R under the standard state _TO, the width dA0 in nearly pupil district and the width dB0 in pupil district far away; Step 3, with the iris internal and external circumference of the iris image of being gathered corresponding respectively be divided into n part, calculate the coordinate (x of certain point on the pupil edge _Pi, y _Pi) and the iris edge on coordinate (x to putting _Ij, y _Ij), according to (x _Pi, y _Pi) and (x _Ij, y _Ij) calculate between said pupil edge point and the iris marginal point apart from dI, wherein, n is the integer greater than 0; Step 4 when dI ≠ dA0+dB0, judges that the iris image of being gathered is in the iris off-rating, calculates the width dA1 in the nearly pupil district the iris off-rating under and the width dB1 in pupil district far away according to following formula:

$\frac{k1}{\mathrm{dA}0}(\mathrm{dA}0-\mathrm{dA}1)=\frac{k2}{\mathrm{dB}0}(\mathrm{dB}0-\mathrm{dB}1)$

DA1+dB1=dI; Step 5 according to dA1 that calculates and dB1, is carried out m respectively to nearly pupil district under the iris off-rating and pupil district far away ₁Row and m ₂Line linearity is sampled and is obtained sampled point; Step 6 is normalized into the rectangular image that resolution is m * n with said sampled point through bilinear interpolation.

Through the iris texture method for normalizing based on the dual spring model of the present invention; Can still use linear method to sample; But more accurately more reliable than existing linear normalization method, and more simple and practical than non-linear normalizing method, can be good at reducing because the error that convergent-divergent brought of pupil; Provide convenience for follow-up iris feature, go forward side by side to improve the overall discrimination of iris recognition.

Wherein, in step 1, the line number m that said nearly pupil district need sample ₁=m * dRPTol, the line number m that said pupil far away district need sample ₂=m-m ₁

In step 2, referring to Fig. 4 A, the iris radius R of the iris image of being gathered _IObtain through the Iris Location link, the width dB0 in the width dA0 in said nearly pupil district and said pupil far away district can calculate through following formula: dA0=dRNToall * (1-dRPTol) * R _I, width dB0=(the 1-dRPTol) * R in said pupil far away district _I-dA0.

In step 3, referring to Fig. 4 B, with the iris internal and external circumference of the iris image of being gathered corresponding respectively be divided into n part, at first can calculate the coordinate (x of certain point on the pupil edge through following formula _Pi, y _Pi) and the iris edge on coordinate (x to putting _Ij, y _Ij):

$x_{\mathrm{pi}}=x_{\mathrm{pI}}+R_{\mathrm{pupil}}*\cos (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

$y_{\mathrm{pi}}=y_{\mathrm{pI}}+R_{\mathrm{pupil}}*\sin (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

Wherein, (x _PI, y _PI) be pupil center's coordinate points, R _PupilBe the pupil radius.

$x_{\mathrm{Ij}}=x_{\mathrm{pI}}+R_I*\cos (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

$y_{\mathrm{Ij}}=y_{\mathrm{pI}}+R_I*\sin (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1]$

Wherein, (x _PI, y _PI) be pupil center's coordinate points, R _IBe iris radius.

Again according to (x _Pi, y _Pi) and (x _Ij, y _Ij) calculate between said pupil edge point and the iris marginal point apart from dI,

$\mathrm{dI}=\sqrt[2]{{(x_{\mathrm{Ij}}-x_{\mathrm{pi}})}^2+{(y_{\mathrm{Ij}}-y_{\mathrm{pi}})}^2}.$

In step 4, can extrapolate the width dA1 in the nearly pupil district under the said off-rating and the width dB1 in pupil district far away and be respectively:

$\mathrm{dA}1=\frac{\mathrm{dB}0(\mathrm{\Δ}-1)+d1}{1+\mathrm{\Δ}*\frac{\mathrm{dB}0}{\mathrm{dA}0}},$ dB1＝dI-dA1，

Wherein $\mathrm{\Δ}=\frac{k_1}{k_2}.$

In a preferred embodiment of the invention, dRPTol is 0.37, and dRNToall is 0.375, this more best than the normalization effect under the state of value, and m=64, n=512.In this preferred embodiment, in step 1, the line number m that said nearly pupil district need sample ₁=24, the line number m that said pupil far away district need sample ₂=40.In step 3, with the iris internal and external circumference of the iris image of being gathered corresponding respectively be divided into n=512 part.In step 5, nearly pupil district under the iris off-rating and pupil district far away are carried out m respectively ₁=24 row and m ₂The sampling of=40 line linearities.In step 6, referring to Fig. 4 C, it is 512 * 64 rectangular image that said sampled point is normalized into resolution through bilinear interpolation.

Through the iris texture method for normalizing based on the dual spring model of the present invention; Can still use linear method to sample; But more accurately more reliable than existing linear normalization method, and more simple and practical than non-linear normalizing method, can be good at reducing because the error that convergent-divergent brought of pupil; Provide convenience for follow-up iris feature, go forward side by side to improve the overall discrimination of iris recognition.

The above only is a preferred implementation of the present invention; Should be pointed out that for those skilled in the art, under the prerequisite that does not break away from the principle of the invention; Can also make some improvement and retouching, these improvement and retouching also should be regarded as protection scope of the present invention.

Claims (10)

1. the iris texture method for normalizing based on the dual spring model is characterized in that, comprises the steps:

Step 1 is set the ratio dRNToall of nearly pupil sector width under ratio dRPTol and the iris standard state of pupil radius and iris radius under the iris standard state and (nearly pupil district+far pupil district) width; Set the line number m that nearly pupil district and pupil district far away need sample ₁And m ₂, m wherein ₁, m ₂For greater than 0 integer, and m ₁+ m ₂=m; Nearly the pupil district is modeled as the spring series connection with unit elasticity coefficient k 1 and k2, wherein k respectively with pupil district far away ₁＜k ₂

Step 2 is obtained the iris radius R of the iris image of being gathered _I, according to said dRPTol, dRNToall and iris radius R _I, calculate the pupil radius R under the standard state _TO, the width dA0 in nearly pupil district and the width dB0 in pupil district far away;

Step 3, with the iris internal and external circumference of the iris image of being gathered corresponding respectively be divided into n part, calculate the coordinate (x of certain point on the pupil edge _Pi, y _Pi) and the iris edge on coordinate (x to putting _Ij, y _Ij), according to (x _Pi, y _Pi) and (x _Ij, y _Ij) calculate between said pupil edge point and the iris marginal point apart from dI, wherein, n is the integer greater than 0;

Step 4 when dI ≠ dA0+dB0, judges that the iris image of being gathered is in the iris off-rating, calculates the width dA1 in the nearly pupil district the iris off-rating under and the width dB1 in pupil district far away according to following formula:

$\frac{k1}{\mathrm{dA}0}(\mathrm{dA}0-\mathrm{dA}1)=\frac{k2}{\mathrm{dB}0}(\mathrm{dB}0-\mathrm{dB}1)$

dA1+dB1＝dI；

Step 5 according to dA1 that calculates and dB1, is carried out m respectively to nearly pupil district under the iris off-rating and pupil district far away ₁Row and m ₂Line linearity is sampled and is obtained sampled point;

Step 6 is normalized into the rectangular image that resolution is m * n with said sampled point through bilinear interpolation.

2. the method for claim 1 is characterized in that, in step 1, and the line number m that said nearly pupil district need sample ₁=m * dRPTol, the line number m that said pupil far away district need sample ₂=m-m ₁

3. the method for claim 1 is characterized in that, in step 2,

The width dA0=dRNToall in said nearly pupil district * (1-dRPTol) * R _IWidth dB0=(the 1-dRPTol) * R in said pupil far away district _I-dA0.

4. in the method for claim 1, it is characterized in that, in step 3,

$x_{\mathrm{pi}}=x_{\mathrm{pI}}+R_{\mathrm{pupil}}*\cos (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

$y_{\mathrm{pi}}=y_{\mathrm{pI}}+R_{\mathrm{pupil}}*\sin (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

Wherein, (x _PI, y _PI) be pupil center's coordinate points, R _PupilBe the pupil radius.

5. in the method as claimed in claim 4, it is characterized in that, in step 3,

$x_{\mathrm{Ij}}=x_{\mathrm{pI}}+R_I*\cos (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1];$

$y_{\mathrm{Ij}}=y_{\mathrm{pI}}+R_I*\sin (i*\∂),\∂=\frac{\mathrm{\π}}{256},i\∈[0,n-1]$

Wherein, (x _PI, y _PI) be pupil center's coordinate points, R _IBe iris radius.

6. in the method as claimed in claim 5, it is characterized in that, in step 3, the distance between said pupil edge point and the iris marginal point:

$\mathrm{dI}=\sqrt[2]{{(x_{\mathrm{Ij}}-x_{\mathrm{pi}})}^2+{(y_{\mathrm{Ij}}-y_{\mathrm{pi}})}^2}.$

7. in the method for claim 1, it is characterized in that in step 4, the width dA1 in the nearly pupil district under the said off-rating and the width dB1 in pupil district far away are respectively:

$\mathrm{DA}1=\frac{\mathrm{DB}0(\mathrm{\Δ}-1)+d1}{1+\mathrm{\Δ}*\frac{\mathrm{DB}0}{\mathrm{DA}0}},$ DB1=dI-dA1, wherein $\mathrm{\Δ}=\frac{k_1}{k_2}.$

8. like each described method in the claim 1 to 7, it is characterized in that said dRPTol is 0.37, said dRNToall is 0.375.

9. like each described method in the claim 1 to 7, it is characterized in that said m is 64.

10. like each described method in the claim 1 to 7, it is characterized in that said n is 512.

Images