CN102270297B - Fingerprint image enhancement method - Google Patents

Abstract

The invention discloses a fingerprint image enhancement method, which aims at downsizing a filter while eliminating a directional effect so as to reduce run time for enhancing a fingerprint image. The method is characterized in that a new filter is designed, the filter is used for enhancing a fingerprint image I, and the image I is enhanced for many times through a formula so as to obtain an enhanced image E. With the method, the fingerprint image can be effectively enhanced, the directional effect can be eliminated by relatively less amount of calculations, and the run time for enhancement is greatly reduced on the premise of eliminating the directional effect.

Description

A kind of enhancement method of fingerprint image

Technical field

The present invention relates to the enhancement method of fingerprint image in fingerprint recognition field in the computer science.

Background technology

Fingerprint recognition obtains application more and more widely as a kind of identity identifying technology based on biological characteristic.High performance fingerprint recognition system needs high accuracy and feature extraction fast and matching algorithm.Feature extraction is extracted and the process of screening through image segmentation, direction calculating, figure image intensifying, lines extraction and refinement, minutiae feature usually.The fingerprint image enhancing is used for strengthening the texture structure of texture structure in the image, particularly low-quality image, thereby can extract characteristics of image more accurately.

The most frequently used a kind of image enchancing method is (the L Hong of the Enhancement Method based on the Gabor wave filter of Hong at present; Y.F.Wang; A.K.Jain; Fingerprint Image Enhancement:Algorithm and Performance Evaluation, IEEE Transactions on Pattern Analysis and Machine Intelligence 20 (8) (1998) 777-789.).The technical scheme of this method is that image and Gabor wave filter are carried out convolution, and the Gabor wave filter is:

$g(x,y,{\mathrm{\δ}}_x,{\mathrm{\δ}}_y,d,\mathrm{\θ})=\exp (-\frac{1}{2}(\frac{X^2}{{\mathrm{\δ}}_x^2}+\frac{Y^2}{{\mathrm{\δ}}_y^2}))\cos \left(\frac{2\mathrm{\πY}}{d}\right)---\left(1\right)$

X＝xcosθ+ysinθ

Y＝xsinθ-ycosθ

Wherein x and y are coordinates, δ _xAnd δ _yBe envelope constant (envelope constant), d and θ are respectively local inter-ridge distance and ridge orientation.Suppose that (x, y) gray scale is that (x y), strengthens the back gray scale and becomes E (x y), gets during enhancing that (x be that center, size are the regional W of w * w y), and the size of regional W is w * w, and title w is the size of Gabor wave filter, then with pixel I for the pixel of the capable x row of the y of image I

$E(x,y)=\underset{(x^{\′},y^{\′})\∈W}{\mathrm{\Σ}}I(x^{\′},y^{\′})g(x^{\′}-x,y^{\′}-y,{\mathrm{\δ}}_x,{\mathrm{\δ}}_y,d,\mathrm{\θ})---\left(\text{2}\right)$

In the Hong method with δ _xAnd δ _yBe set to 4, and the size of regional W is set to w * w=11 * 11, this set makes the image that strengthens can have tangible directive effect, and the different enhancing degree of promptly local ridge orientation are obviously different, particularly for the image of low contrast.Experiment shows: with δ _xAnd δ _yBe set to d/2, d is local inter-ridge distance, strengthens the size w of wave filter simultaneously, and w is set to 3 times of inter-ridge distance, and generally getting w is odd number,

Can eliminate the directive effect that strengthens the result.For the image of 500dpi, inter-ridge distance generally is distributed in 5-15 pixel, and most of inter-ridge distances are between 8-10.For inter-ridge distance 10, w * w should be set to 31 * 31, could eliminate directive effect preferably.It is big more that w is provided with, and the enhancing time that needs is many more, and therefore strengthening w can increase and strengthen needed working time.How can both eliminate directive effect, dwindling filter size w again is the technical matters that those skilled in the art very pay close attention to.

Summary of the invention

The technical matters that the present invention will solve is: when eliminating directive effect, dwindle filter size, strengthen the working time that fingerprint image needs to reduce.

In order to solve the problems of the technologies described above, the technical scheme that the present invention proposes is:

The first step designs a new wave filter, and this wave filter is:

$c(x,y,d,\mathrm{\θ})=h\left(\mathrm{\π}\sqrt{\frac{x^2+y^2}{d}}\right)\cos \left(\frac{\mathrm{\λ\πY}}{d}\right)---\left(3\right)$

Y＝xsinθ-ycosθ

Wherein x and y are coordinates, and d and θ are respectively local inter-ridge distance and the ridge orientation of image (for example, if strengthen image pixel (x with this wave filter ₀, y ₀) gray-scale value, then d and θ are respectively (x ₀, y ₀) inter-ridge distance and the ridge orientation located),

It is function

Independent variable, (value of λ makes ∑ in λ=2.181268 _-d≤x≤d∑ _-d≤y≤dC (x, y, d, θ)=0).The pixel of the capable x of the y of image I to be strengthened row (x, y) gray scale be I (x, y), 0≤x≤width-1,0≤y≤height-1, width are the width of image I, height is the height of image I.

Second step, adopt wave filter to strengthen fingerprint image I, the image E after being enhanced, method is:

In image I, get with pixel (x; Y) (0≤x≤width-1; 0≤y≤height-1) is the center; Size is the regional W of w * w, and value is

to w (be about local inter-ridge distance 2 times).Adopt formula (4) that image I is strengthened, the image E after being enhanced, E (x, the gray scale of y) locating be E (x, y).

$E(x,y)=\underset{(x^{\′},y^{\′})\∈W}{\mathrm{\Σ}}I(x^{\′},y^{\′})c(x^{\′}-x,y^{\′}-y,d,\mathrm{\θ})---\left(4\right)$

D and θ are that (x, local inter-ridge distance and the ridge orientation y) located, x ', y ' are the coordinate of the pixel in the regional W in the formula (4).For the view picture fingerprint image, utilize formula (4) to calculate the gray scale after strengthening to each pixel (utilizing the method in the open source literature to calculate) of foreground area successively, just can obtain to strengthen image E.A fingerprint image I can strengthen repeatedly, and the number of times T of enhancing is by the decision of fingerprint image quality, and the process of the image E that is enhanced is:

2.1t=0 t is for strengthening the loop control variable of number of times.

2.2 (x y) is image I lower left corner pixel, x=0, y=0 to the initialization pixel.

2.3 if (x y) is background pixel to pixel, and then: (x y)=255, changes 2.9 to E; If (x y) is not background pixel to pixel, carries out 2.4.

2.4d=pixel (x; Y) inter-ridge distance of locating; (initialisation image E is at (x for filter size

for x, the ridge orientation of y) locating for θ=pixel; The gray-scale value E that y) locates (x, y)=0.Coordinate variable (the i of initialization wave filter; J),

2.5E (x, y)=E (x, y)+I ' (x+i, y+j) c (i, j, d, θ), wherein

$I^{\&prime;}(x+i,y+j)=\left\{\begin{array}{cc}I(x+i,y+j)& 0\&le;x+i<\mathrm{width},0\&le;y+j<\mathrm{height}\\ 0& \mathrm{else}\end{array}\right.$

2.6i＝i+1。

2.7 if changes 2.5; Otherwise carry out 2.7.1:

2.7.2 if

changes 2.5; Otherwise carry out 2.8;

2.8 if E (x, y)＞255, then E (x, y)=255; If E (x, y)＜0, then E (x, y)=0; (x, y)≤255, then (x's E y) remains unchanged as if 0≤E.

2.9x＝x+1。

2.10 if x＜width (width of image I) changes 2.3; Otherwise:

2.10.1x＝0，y＝y+1；

2.10.2 if y＜height (height of image I) changes 2.3; Otherwise carry out 2.11;

2.11t＝t+1。

2.12, change 2.2 if t＜T copies E to I; Otherwise finish, among this moment E in store I is strengthened T time after

The enhancing result.T is the number of times that strengthens, and is determined by the fingerprint image quality.

The 2.1 steps initialization control of enhancing process strengthens the loop variable t=0 of number of times; From 2.2 to 2.10.2 is once to strengthen;

2.11 the step increases 1 with t, expression has strengthened t time; 2.12 the step judges whether to have strengthened T time, and if also do not reach T time again execution once strengthen process.Compare with existing method, adopt the present invention can obtain following beneficial effect:

Can strengthen fingerprint image effectively; Can eliminate directive effect with few relatively calculated amount: need the wave filter size be set to 3 times of inter-ridge distance in order to eliminate directive effect when traditional Gabor wave filter strengthens fingerprint; And about 2 times (

) that wave filter of the present invention only needs wave filter size w to be set to inter-ridge distance just can eliminate directive effect; Thereby eliminating under the prerequisite of directive effect, adopt the present invention to strengthen and be 44% (promptly 4/9) needed working time based on the Enhancement Method of Gabor wave filter.

Description of drawings

Fig. 1 is an overview flow chart of the present invention.

Fig. 2 is the original image that is used to test comparison.

Fig. 3 is the present invention and background technology based on the enhancing result contrast to image shown in Figure 2 of the Enhancement Method of Gabor wave filter.

Embodiment

Fig. 1 is an overview flow chart of the present invention.

The first step designs a new wave filter, and this wave filter is shown in formula (3).

In second step, adopt wave filter to strengthen fingerprint image I, the image E after being enhanced.

Fig. 2 is the original image that is used to strengthen experiment, and this image is from FVC2000-DB1-100_1.

Fig. 3 is the present invention and background technology based on the enhancing result contrast to image shown in Figure 2 of the Enhancement Method of Gabor wave filter.Wherein (a) and (b) carry out respectively when to be background technology based on the Enhancement Method wave filter size of Gabor wave filter be set to 2 times of inter-ridge distance strengthening for 1 time and the enhancing image that strengthens acquisitions for 2 times; (a) can find out tangible directive effect in; Even in (b), directive effect is not eliminated fully yet.(c) and (d) be that Gabor filtering wave filter size carries out respectively when being set to 3 times of inter-ridge distance strengthening for 1 time and the enhancing image that strengthens acquisitions for 2 times; Can find out; (a) directive effect that occurs in is eliminated in (c) basically, and (d) that obtain through twice enhancing had texture structure clearly.(e) and (f) be that the inventive method is provided with and strengthens through 1 time respectively when parameter lambda=2.181268 and filter size w are 2 times of inter-ridge distances and strengthen the result who obtains 2 times.(e) eliminated the directive effect that occurs in (a), the linking of zones of different lines is level and smooth.Strengthen (f) that obtain through 2 times and slightly be better than (d), because the situation that lines is sticking mutually or break off appears in upper left quarter (d), and the enhancing of the texture structure (f) recovers better.(g) and (h) be the inventive method filter size w during for

(d is an inter-ridge distance) respectively through strengthening for 1 time and the result who strengthens acquisitions for 2 times.Tangible directive effect is arranged (g), and reason is filter size w 2 times less than inter-ridge distance.

In a word; There is obvious directive effect in background technology when being set to 2 times of inter-ridge distance based on the Enhancement Method wave filter size of Gabor wave filter; Need the wave filter size be increased to 3 times of inter-ridge distance in order to eliminate directive effect, thereby be original 2.25 times the working time that needs.Adopt the present invention, the wave filter size is 2 times of inter-ridge distance, not only eliminated directive effect, and do not increased working time, and the Enhancement Method that the enhancing result who obtains slightly is better than based on the Gabor wave filter increases the enhancing result who obtains after the wave filter size.

Claims (1)

1. A fingerprint image enhancement method is characterized in that comprising the following steps: The first step is to design a filter, which is: $\mathrm{<span\; class="notranslate">\; c</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; d</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; \&pi;</span>}\sqrt{\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; d</span>}}}<span\; class="notranslate">\; )</span>\mathrm{<span\; class="notranslate">\; cos</span>}<span\; class="notranslate">\; (</span>\frac{\mathrm{<span\; class="notranslate">\; \&lambda;\&pi;Y</span>}}{\mathrm{<span\; class="notranslate">\; d</span>}}<span\; class="notranslate">\; )</span>$

Y=xsinθ-ycosθ Where x and y are the coordinates, d and θ are the local grain spacing and grain direction, respectively,

is a function The independent variable of λ=2.181268; the pixel (x, y) grayscale of the yth row and xth column of the image I to be enhanced is I(x, y), 0≤x≤width-1, 0≤y≤height -1, width is the width of image I, height is the height of image I; The second step is to use a filter to enhance the fingerprint image I to obtain the enhanced image E, the method is: 2.1t=0, t is the loop control variable of the number of enhancements; 2.2 Initialize the pixel (x, y) as the pixel in the lower left corner of the image I, x=0, y=0; 2.3 If the pixel (x, y) is a background pixel, then: the gray value of E at (x, y) is E(x, y)=255, go to 2.9; if the pixel (x, y) is not a background pixel, Execute 2.4; 2.4d = grain pitch at pixel (x, y), θ = grain direction at pixel (x, y), filter size

Initialize the gray value E(x,y)=0 of the image E at (x,y); initialize the coordinate variable (i,j) of the filter, 2.5 E(x,y)=E(x,y)+I′(x+i,y+j)c(i,j,d,θ), where ${\mathrm{<span\; class="notranslate">\; I</span>}}^{<span\; class="notranslate">\; \&prime;</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\left\{\begin{array}{cc}\mathrm{<span\; class="notranslate">\; I</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; )</span>& \mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; width</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; 0</span>}<span\; class="notranslate">\; \&le;</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; j</span>}<span\; class="notranslate">\; <</span>\mathrm{<span\; class="notranslate">\; height</span>}\\ \mathrm{<span\; class="notranslate">\; 0</span>}& \mathrm{<span\; class="notranslate">\; else</span>}\end{array}\right.<span\; class="notranslate">\; ;</span>$ 2.6 i=i+1; 2.7 if Go to 2.5; otherwise execute 2.7.1: 2.7.1

j=j+1; 2.7.2 If

Go to 2.5; otherwise, go to 2.8; 2.8 If E(x,y)>255, then E(x,y)=255; if E(x,y)<0, then E(x,y)=0; if 0≤E(x,y) ≤255, then E(x,y) remains unchanged; 2.9 x=x+1; 2.10 If x<width, go to 2.3; otherwise: 2.10.1 x=0, y=y+1; 2.10.2 If y<height, go to 2.3; otherwise, go to 2.11; 2.11 t=t+1; 2.12 If t<T, copy E to I and go to 2.2; otherwise, end, at this time, E stores the enhancement result after enhancing I T times, and T is the number of enhancements, which is determined by the quality of the fingerprint image.

Images