CN100573554C - The direction filtering reinforcement method of fingerprint image - Google Patents

The direction filtering reinforcement method of fingerprint image Download PDF

Info

Publication number
CN100573554C
CN100573554C CNB2008100178757A CN200810017875A CN100573554C CN 100573554 C CN100573554 C CN 100573554C CN B2008100178757 A CNB2008100178757 A CN B2008100178757A CN 200810017875 A CN200810017875 A CN 200810017875A CN 100573554 C CN100573554 C CN 100573554C
Authority
CN
China
Prior art keywords
filtering
pixel
image
directions
fingerprint image
Prior art date
Application number
CNB2008100178757A
Other languages
Chinese (zh)
Other versions
CN101271516A (en
Inventor
范九伦
李利
何晶
Original Assignee
范九伦
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by 范九伦 filed Critical 范九伦
Priority to CNB2008100178757A priority Critical patent/CN100573554C/en
Publication of CN101271516A publication Critical patent/CN101271516A/en
Application granted granted Critical
Publication of CN100573554C publication Critical patent/CN100573554C/en

Links

Abstract

The present invention is a kind of direction filtering reinforcement method of fingerprint image.It comprises that mainly the normalized of fingerprint image, the field of direction estimation and the trend pass filtering of fingerprint ridge strengthen.Wherein the purpose of normalized is to eliminate the inhomogeneous influence to fingerprint image of finger presses, for the directional information of accurately extracting streakline is carried out adequate preparation.The present invention extracts the directional information of streakline with mask method, and the accuracy height has been established solid foundation for follow-up trend pass filtering strengthens.In the trend pass filtering of fingerprint ridge strengthens, the present invention carries out the level and smooth of streakline tangential direction with the smothing filtering template, carry out sharpening with the streakline normal direction of sharpening filtering template after level and smooth, can obviously strengthen the contrast of fingerprint ridge line and valley line, and can connect because the crestal line that the finger overdrying causes interrupts and removes and pointed the wet crestal line adhesion that causes, obtain good enhancing effect.

Description

The direction filtering reinforcement method of fingerprint image

Technical field

The present invention relates to the preceding Pre-processing Method for Fingerprint Image of a kind of fingerprint image identification, relate in particular to the direction filtering reinforcement method in a kind of fingerprint image preprocessing.

Background technology

Pre-process of Fingerprint Image is a most important parts in the fingerprint automatic recognition system, and pretreated result will directly influence the reliability and the accuracy of identification, and therefore, Pre-process of Fingerprint Image has become the focus of fingerprint recognition area research.The filtering enhancement process that all comprises a fingerprint image in the Pre-process of Fingerprint Image.In prior art, " electronic letters, vol " 2004 1 phases provide the fingerprint image increase method of a kind of " based on ridge to the fingerprint filtering algorithm ", article is pointed out: in the gatherer process of fingerprint, inevitably various noises can be introduced, when serious, even the wrong information that extracts may be caused, therefore before the feature that takes the fingerprint, need carry out Filtering Processing to fingerprint image,, strengthen useful information to remove garbage as far as possible.This method is at first removed fingerprint image with extreme value filtering Swashing type noise carries out smothing filtering to the filtered fingerprint image of extreme value then, to remove random noise.After twice filtering, the noise in the image is eliminated substantially, for the more effective finger print information that extracts, and obtain the fingerprint image of enhancing, and it has constructed a kind of ridge to wave filter, and ridge is to filtering, carry out filtering along the crestal line direction of fingerprint exactly, to strengthen the ridge information of fingerprint.At ridge before filtering, at first need to determine the field of direction of fingerprint image, this method utilization is optimized three ladder degree operators and is estimated the pixel direction, promptly calculate the deflection of each pixel earlier, and then according to deflection the direction of this pixel is quantized to eight possible streakline directions and get on, construct the smothing filtering template of corresponding eight directions again, with the template of above-mentioned eight directions respectively with the counterparty to fingerprint image carry out convolution algorithm, filtering strengthens the streakline that the counterparty makes progress, with the enhancing image overlay of eight directions, just can obtain the enhancing image of level and smooth streakline at last.Wherein, the smothing filtering template of horizontal direction is as follows in eight directions:

Tr = 1 1 1 1 1 5 5 5 5 5 8 8 8 8 8 5 5 5 5 5 1 1 1 1 1

Smothing filtering template to this horizontal direction rotates a certain angle, and according to the discrete characteristic of image slices vegetarian refreshments, just can obtain the smothing filtering template of other seven directions:

1 1 1 1 1 1 1 5 5 8 1 8 8 8 1 8 5 5 1 1 1 1 1 1 1 1 1 5 5 8 1 1 5 8 5 1 5 8 5 1 5 8 5 1 1 8 5 5 1 1 1 1 5 8 5 1 5 8 5 1 1 5 8 5 1 1 5 8 5 1 5 8 5 1 1

1 5 8 5 1 1 5 8 5 1 1 5 8 5 1 1 5 8 5 1 1 5 8 5 1 5 8 5 1 1 1 5 8 5 1 1 5 8 5 1 1 5 8 5 1 1 1 5 8 5 8 5 5 1 1 5 8 5 1 1 1 5 8 5 1 1 1 5 8 5 1 1 5 5 8 1 1 1 1 1 8 5 5 1 1 1 8 8 8 1 1 1 5 5 8 1 1 1 1 1

The problem that this method exists is as follows:

(1) before strengthening image, remove noise with extreme value filtering and smothing filtering, the detailed information that can lose a lot of fingerprint images, this is undesirable.

(2) before the trend pass filtering of image strengthens, the original fingerprint image is not done the normalized of gray scale, can't eliminate that because of finger presses is inhomogeneous fingerprint image is exerted an influence (situation that fingerprint center look heavy, peripheral look shallow promptly often occurring), this will produce harmful effect to the accuracy that follow-up directional information is extracted.

(3) algorithm that the field of direction is estimated in this method is the deflection that calculates each picture element earlier, and then according to deflection the direction of this picture element is quantized to eight directions and get on, that is to say, the process that changes into corresponding quantization direction value from the orientation angle value of obtaining exists the uncertain factor that quantizes selection, therefore, the direction value is difficult to accurately, and this will directly influence according to eight quantized directions and carry out the effect that streakline filtering strengthens.

(4) this method has only designed the smothing filtering template in trend pass filtering strengthens, do not design sharpening filtering template, and template is less, is difficult to reflect accurately the characteristic along the streakline trend pass filtering, changes violent or to exist the fingerprint image of streakline adhesion to strengthen effect not ideal to local direction.

Summary of the invention

Purpose of the present invention provides a kind of direction filtering reinforcement method of improved fingerprint image at the shortcoming in the prior art, in the hope of obtain outstanding effect on the fingerprint ridge line of level and smooth streakline, connection fracture.

For achieving the above object, the step of technical solution of the present invention is as follows:

One, the normalized of fingerprint image:

Adopt following formula that each gray values of pixel points in the fingerprint image is carried out normalized, obtain normalized fingerprint image:

(i j) is the Normalized Grey Level value of pixel in the image to g in the formula, and (i j) is the actual grey value of certain pixel to I, and m is the gray average of each pixel in the image, and v is a variance yields, m 0Be the gray average of each pixel expectation in the image, v 0Variance yields for expectation;

Two, the some direction is estimated

1, it is as follows to set up one 9 * 9 mask A:

??7 ??6 ??5 ??4 ??3 ??7 ??5 ??3 ??8 ??7 ??6 ??5 ??4 ??3 ??2 ??8 ??7 ??5 ??3 ??2 ??1 ??1 ??1 ??1 ??* ??1 ??1 ??1 ??1 ??2 ??3 ??5 ??7 ??8 ??2 ??3 ??4 ??5 ??6 ??7 ??8 ??3 ??5 ??7 ??3 ??4 ??5 ??6 ??7

* among this mask A is pending pixel, represents eight possible streakline directions of this pixel by eight rectilinear directions of 1~8 array,

2, with pixel *Be central point, calculate the average gray Gmean[i on described eight directions respectively] (i=1,8), then these eight directions are divided into 4 groups by direction perpendicular to each other, promptly 1 and 5 is one group, 2 and 6 is one group, 3 and 7 is one group, 4 and 8 is one group, calculate the difference G[i of two average gray in every group again]=| Gmean[i]-Gmean[i+4] |., get this absolute value differences G[i then] peaked both direction iMax, iMax+4 is as two of this pixel * possible streakline directions, if the gray-scale value at this pixel place is g (*), then the direction of this pixel is:

o ( * ) = iMax if | g ( * ) - Gmean ( iMax ) | < | g ( * ) - Gmean ( iMax + 4 ) | iMax + 4 otherwise - - - ( 8 )

Promptly get this both direction iMax, iMax+4 gray average near this pixel *The direction of gray-scale value as this pixel *The streakline direction;

3, this operation travels through all pixels in normalization fingerprint image, has just obtained the some direction field pattern of normalization fingerprint image;

Three, will put the direction field pattern and be converted into piece direction field pattern:

Two steps resulting some direction field pattern is divided into the overlapped image block of W * W, add up the pixel number on eight directions in each image block then, getting the direction that pixel number is maximum in eight directions is the streakline piece direction of this image block, so operation, travel through each image block of entire image, obtain the piece direction field pattern of fingerprint image at last;

Four, fingerprint image travel direction filtering after the normalization is strengthened: according to eight quantized directions of mask A defined, the smothing filtering template and the sharpening filtering template of constructing these eight directions with 7 * 7 window are as follows:

0 0 0 0 0 0 0 1 1 1 1 1 1 1 5 5 5 5 5 5 5 8 8 8 8 8 8 8 5 5 5 5 5 5 5 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 - 3 - 3 - 3 - 3 - 3 - 3 - 3 10 10 10 10 10 10 10 - 3 - 3 - 3 - 3 - 3 - 3 - 3 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 1 1 5 0 0 1 1 5 5 8 0 1 1 5 8 8 5 1 5 5 8 5 5 1 5 8 8 5 1 1 0 8 5 5 1 1 0 0 5 1 1 0 0 0 0 0 0 0 0 1 1 - 3 0 0 1 1 - 3 - 3 10 0 1 1 - 3 10 10 - 3 1 - 3 - 3 10 - 3 - 3 1 - 3 10 10 - 3 1 1 0 10 - 3 - 3 1 1 0 0 - 3 1 1 0 0 0 0

1 directional smoothing filtering template, 1 direction sharpening filtering template, 2 directional smoothing filtering templates, 2 direction sharpening filtering templates

0 0 0 0 1 5 8 0 1 1 1 1 8 5 0 1 5 5 8 5 1 0 1 5 8 5 1 0 1 5 8 5 5 1 0 5 8 5 1 1 1 0 8 5 1 0 0 0 0 0 0 0 0 1 - 3 10 0 1 1 1 1 10 - 3 0 1 - 3 - 3 10 - 3 1 0 1 - 3 10 - 3 1 0 1 - 3 10 - 3 - 3 1 0 - 3 10 - 3 1 1 1 0 10 - 3 1 0 0 0 0 0 0 0 1 5 8 5 0 0 1 5 8 5 1 0 1 1 5 8 5 1 0 1 5 8 5 1 0 1 5 8 5 1 1 0 1 5 8 5 1 0 0 5 8 5 1 0 0 0 0 0 0 1 - 3 10 - 3 0 0 1 - 3 10 - 3 1 0 1 1 - 3 10 - 3 1 0 1 - 3 10 - 3 1 0 1 - 3 10 - 3 1 1 0 1 - 3 10 - 3 1 0 0 - 3 10 - 3 1 0 0 0

3 directional smoothing filtering templates, 3 direction sharpening filtering templates, 4 directional smoothing filtering templates, 4 direction sharpening filtering templates

0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 5 8 5 1 0 0 0 1 5 8 5 1 0 0 1 5 8 5 1 1 0 0 1 5 8 5 1 0 0 1 1 5 8 5 1 0 0 1 5 8 5 1 0 0 0 1 5 8 5 - 3 10 - 3 1 0 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 1 0 0 1 - 3 10 - 3 1 0 0 1 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 0 1 - 3 10 - 3

5 directional smoothing filtering templates, 5 direction sharpening filtering templates, 6 directional smoothing filtering templates, 6 direction sharpening filtering templates

8 5 1 0 0 0 0 5 8 1 1 1 1 0 1 5 8 5 5 1 0 0 1 5 8 5 1 0 0 1 5 5 8 5 1 0 1 1 1 5 8 5 0 0 0 0 1 5 8 10 - 3 1 0 0 0 0 - 3 10 1 1 1 1 0 1 - 3 10 - 3 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 - 3 10 - 3 1 0 1 1 1 - 3 10 - 3 0 0 0 0 1 - 3 10 5 1 1 0 0 0 0 8 5 5 1 1 0 0 5 8 8 5 1 1 0 1 5 5 8 5 5 1 0 1 1 5 8 8 5 0 0 1 1 5 5 8 0 0 0 0 1 1 5 - 3 1 1 0 0 0 0 10 - 3 - 3 1 1 0 0 - 3 10 10 - 3 1 1 0 1 - 3 - 3 10 - 3 - 3 1 0 1 1 - 3 10 10 - 3 0 0 1 1 - 3 - 3 10 0 0 0 0 1 1 - 3

7 directional smoothing filtering templates, 7 direction sharpening filtering templates, 8 directional smoothing filtering templates, 8 direction sharpening filtering templates

Fingerprint image after the normalization is carried out filtering with the filtering template of eight directions successively, promptly earlier with the level and smooth streakline of smothing filtering template, with sharpening filtering template level and smooth back streakline is carried out sharpening again, thereby the filtering that obtains eight directions respectively strengthens image, again the image that eight filtering strengthen is superposeed the fingerprint image that is enhanced at last according to the piece field of direction that obtains then.

Further improved technical scheme of the present invention is as follows:

Described some direction field pattern is converted into that the overlapped image block size that adopts is 16 * 16 pixels in the step of piece direction field pattern, overlapping width ω=1~15 pixel.

The present invention's characteristics compared with the prior art are as follows:

One, the present invention has removed extreme value filtering and the smothing filtering in the prior art, has overcome its filtering and noise reduction and the defective of losing image detail information, makes the detailed information of original image keep more completely, is beneficial to follow-up fingerprint recognition.

Two, the present invention is before trend pass filtering strengthens, do relatively goodly to the pre-service of original image, at first it has carried out the normalized of pixel gray-scale value to original image, this normalized is very important, because most fingerprint images of being gathered all exist because of the inhomogeneous influence to fingerprint image of finger presses (center look heavy, peripheral look shallow).Normalized purpose is to make the gradation of image value reach predefined mean value and variance yields, elimination is because of the inhomogeneous influence to fingerprint image of finger presses, for the directional information of the streakline that correctly takes the fingerprint is carried out sufficient preparation, so it is correctly to extract streakline directional information important step.The secondth, it determines the streakline direction of pixel in the image with mask method, promptly according to the directivity characteristics of fingerprint ridge, mask A turns to eight possible directions with the streakline side vector of fingerprint image, determine the direction of this pixel then by the difference of calculating the average gray on the perpendicular direction of average gray on eight directions of average gray and calculating on its eight directions, this process does not have the conversion of intermediate parameters, therefore, the directional information accuracy height that is extracted.The correctness that directional information is extracted directly has influence on the effect that follow-up trend pass filtering strengthens, and it is to obtain the key point of good enhancing effect that the present invention can accurately extract directional information.

Three, the present invention is by being that the enhancing checking of a large amount of fingerprint images in the fingerprint database of FVC2000 obtains to code name: existing 5 * 5 template strengthens effective not as 7 * 7 template, simultaneously, directivity characteristics and experimental verification according to fingerprint image, on the tangential direction of streakline, carry out level and smooth, and on the direction of its normal, carry out sharpening, can obviously strengthen the contrast of fingerprint ridge line and valley line, and can connect because the crestal line that the finger overdrying causes interrupts and removes and pointed the wet crestal line adhesion that causes, therefore, the present invention is on the basis of existing eight directional smoothing filtering templates, constructed the sharpening filtering template of eight directions again, obtained the enhancing effect better than prior art.

Four, the present invention extracts that some low-quality fingerprint images have carried out the contrast experiment in the FVC2000 fingerprint database, the result is shown in Fig. 1~6, wherein Fig. 1, Fig. 4 are original image, above Fig. 1, there is the dark excessively problem of illumination, there is the unsharp problem of streakline in the below, has a large amount of crestal line broken strings among Fig. 4.Fig. 2, Fig. 4 adopt existing " based on ridge to the fingerprint filtering algorithm " to carry out the image enhanced results, Fig. 5, the 6th adopts the inventive method to carry out the image enhanced results, in order clearly to represent the enhancing effect, Fig. 3,4 and Fig. 5,6 provides is to strengthen the binary picture of image through simple binary conversion treatment.Both contrasts as can be seen, the present invention can strengthen the contrast of fingerprint image in the same old way well under the situation that does not adopt extreme value filtering and smothing filtering, and at level and smooth streakline be connected on the crestal line that ruptures and be far superior to prior art.Obtaining good like this enhancing effect is the accurate extraction that depends on fingerprint image streakline direction on the one hand, is the rationality that depends on smothing filtering template and sharpening filtering structure of transvers plate on the other hand.

Description of drawings

One of Fig. 1, original fingerprint image.

Fig. 2, the design sketch that adopts existing method Fig. 1 to be carried out the figure image intensifying

Fig. 3, the design sketch that adopts the present invention Fig. 1 to be carried out the figure image intensifying

Two of Fig. 4, original fingerprint image.

Fig. 5, the design sketch that adopts existing method Fig. 4 to be carried out the figure image intensifying.

Fig. 6, the design sketch that adopts the present invention Fig. 4 to be carried out the figure image intensifying.

Three of Fig. 7, original fingerprint image.

Fig. 8, Fig. 7 is carried out design sketch after the normalized.

One of fingerprint image after Fig. 9, the normalization.

The pixel direction field pattern of Figure 10, Fig. 9.

The piece direction field pattern of Figure 11, Figure 10.

Figure 12, trend pass filtering strengthen operational flowchart.

Embodiment

The present invention comprises that mainly normalized, the field of direction to fingerprint image estimated and trend pass filtering strengthens three big steps.Concrete branch is chatted as follows:

One, the normalized of fingerprint image:

Adopt following formula that each gray values of pixel points in the fingerprint image is carried out normalized,

(i j) is the Normalized Grey Level value of pixel in the image to g in the formula, and (i j) is the actual grey value of certain pixel to I, and m is the gray average of each pixel in the image, and v is variance (being the average of the difference square of each pixel gray-scale value and gray average in the image), m 0Be the gray average of each pixel expectation in the image, v 0Variance yields for expectation.

The normalized purpose of fingerprint image is to make the gray-scale value of image reach predefined mean value and variance yields.Average m when expectation 0With variance v 0When getting different value, normalized degree difference, average and variance value are big more, and the normalization degree is strong more.Push the fingerprint image of degree of irregularity for difference, by adjusting parameter m 0, v 0, just can reach and eliminate the inhomogeneous purpose of finger presses the fingerprint image influence.The present invention carries out normalized to the part fingerprint image in the FVC2000 fingerprint database, gets m in emulation experiment of the present invention 0=100, v 0=100, just can reach the purpose of normalized.Fig. 7,8 has shown the fingerprint image that normalized is forward and backward, both contrasts as can be seen, Fig. 7 pushes at the center heavier position and obtained gray scale adjustment preferably in Fig. 8, and can remove the fraction noise, obtain fingerprint image more clearly, this gray scale adjustment can not lost the detailed information of original image, is very favorable to going on foot the correct streakline direction of extracting pixel down.

Two, the some field of direction is estimated

1, determine after the normalization direction of each pixel in the fingerprint image, the mask A that sets up 9 * 9 earlier is as follows:

??7 ??6 ??5 ??4 ??3 ??7 ??5 ??3 ??8 ??7 ??6 ??5 ??4 ??3 ??2 ??8 ??7 ??5 ??3 ??2 ??1 ??1 ??1 ??1 ??* ??1 ??1 ??1 ??1 ??2 ??3 ??5 ??7 ??8 ??2 ??3 ??4 ??5 ??6 ??7 ??8 ??3 ??5 ??7 ??3 ??4 ??5 ??6 ??7

* among this mask A is pending central pixel point, represent eight possible streakline directions of this pixel by the rectilinear direction of 1~8 array, according to the relation of pixel gray-scale value on central pixel point * and its neighborhood all directions, can obtain the affiliated streakline direction of central pixel point *.

The gamma characteristic that the fingerprint image streakline distributes is: the gray-value variation along the streakline direction is slow, and along violent perpendicular to the gray-value variation of streakline direction; If point *On valley line, the mean value of the gray-scale value on the streakline direction must be the maximal value in the gray-scale value mean value (the whitest) on all directions, and must be the minimum value in the gray-scale value mean value (the most black) on all directions perpendicular to the mean value of the gray-scale value on the streakline direction; If point *On crestal line, the mean value of the gray-scale value on the streakline direction must be the minimum value in the gray-scale value mean value (the most black) on all directions, and must be the maximal value in the gray-scale value mean value (the whitest) on all directions perpendicular to the mean value of the gray-scale value on the streakline direction.According to this gamma characteristic, can judge central pixel point *Belong to which streakline direction, thereby obtain the some direction field pattern of image, specifically be calculated as follows:

With pixel *Be central point, calculate the average gray Gmean[i on described eight directions respectively] (i=1,8), then these eight directions are divided into 4 groups by direction perpendicular to each other, promptly 1 and 5 is one group, 2 and 6 is one group, 3 and 7 is one group, 4 and 8 is one group, calculate the difference G[i of two average gray in every group again]=| Gmean[i]-Gmean[i+4] |., get this absolute value differences G[i then] peaked both direction iMax, iMax+4 is as two possible streakline directions of this pixel *, if the gray-scale value at this pixel place is g (*), then the direction of this pixel is:

Promptly get this both direction iMax, iMax+4 gray average near this pixel *The direction of gray-scale value as this pixel *The streakline direction.

All pixels in this operation traversal normalization fingerprint image, just obtained the some direction field pattern of normalization fingerprint image, as shown in figure 10, this figure is the direction value i (i=1 with pixel, 8) be the fingerprint ridge direction field pattern of gray-scale value, can roughly find out the basic trend of streakline in the original image (Fig. 9) by this figure.

2, will put the direction field pattern and be converted into piece direction field pattern:

In order to suppress noise,, will put the field of direction and be converted into the piece field of direction according to the conforming characteristics of partial fingerprint image to the fingerprint image orientation estimation effect.Concrete grammar is:

Resulting some direction field pattern of 1 step is divided into the overlapped or nonoverlapping image block of W * W, this example adopts 16 * 16 overlapped image blocks, wherein overlapping width can be ω=1~15, this example is selected ω=8, like this, the lap that 128 pixels are just arranged between two adjacent image pieces, the ω value is more little in principle, the piece field of direction that obtains is accurate more, but can consume more time, at the image in the FVC2000 fingerprint database, get ω=8 and just can reach the purpose that suppresses noise, the level and smooth field of direction.The histogram of statistics block mid point direction then, promptly add up the pixel number on eight directions in each image block, getting the direction that pixel number is maximum in eight directions again is the piece direction of this image block, so operation, each image block of traversal entire image, obtain the piece direction field pattern of fingerprint image at last, as shown in figure 11.Contrast points field of direction Figure 10, as can be seen, resulting direction field pattern of above-mentioned conversion is more accurately, thereby carried out sufficient preparation for follow-up trend pass filtering strengthens.

Three, trend pass filtering strengthens

It is exactly to carry out filtering along the streakline direction of fingerprint to strengthen that trend pass filtering strengthens, and meets the directivity characteristics of fingerprint image.The present invention is level and smooth on the streakline tangential direction, and sharpening on normal direction when purpose is to strengthen fingerprint ridge line and valley line contrast, connects because the crestal line that the finger overdrying causes interrupts and removes and pointed the wet crestal line adhesion that causes.By to code name being the enhancing checking of a large amount of fingerprint images in the FVC2000 DB2 fingerprint database, find that 7 * 7 template can realize better strengthening effect, simultaneously, the present invention has constructed according to smothing filtering template and sharpening filtering template on eight directions of mask A defined as follows according to the experimental result that a large amount of fingerprint images strengthen:

0 0 0 0 0 0 0 1 1 1 1 1 1 1 5 5 5 5 5 5 5 8 8 8 8 8 8 8 5 5 5 5 5 5 5 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 1 - 3 - 3 - 3 - 3 - 3 - 3 - 3 10 10 10 10 10 10 10 - 3 - 3 - 3 - 3 - 3 - 3 - 3 1 1 1 1 1 1 1 0 0 0 0 0 0 0 0 0 0 0 1 1 5 0 0 1 1 5 5 8 0 1 1 5 8 8 5 1 5 5 8 5 5 1 5 8 8 5 1 1 0 8 5 5 1 1 0 0 5 1 1 0 0 0 0 0 0 0 0 1 1 - 3 0 0 1 1 - 3 - 3 10 0 1 1 - 3 10 10 - 3 1 - 3 - 3 10 - 3 - 3 1 - 3 10 10 - 3 1 1 0 10 - 3 - 3 1 1 0 0 - 3 1 1 0 0 0 0

1 directional smoothing filtering template, 1 direction sharpening filtering template, 2 directional smoothing filtering templates, 2 direction sharpening filtering templates

0 0 0 0 1 5 8 0 1 1 1 1 8 5 0 1 5 5 8 5 1 0 1 5 8 5 1 0 1 5 8 5 5 1 0 5 8 5 1 1 1 0 8 5 1 0 0 0 0 0 0 0 0 1 - 3 10 0 1 1 1 1 10 - 3 0 1 - 3 - 3 10 - 3 1 0 1 - 3 10 - 3 1 0 1 - 3 10 - 3 - 3 1 0 - 3 10 - 3 1 1 1 0 10 - 3 1 0 0 0 0 0 0 0 1 5 8 5 0 0 1 5 8 5 1 0 1 1 5 8 5 1 0 1 5 8 5 1 0 1 5 8 5 1 1 0 1 5 8 5 1 0 0 5 8 5 1 0 0 0 0 0 0 1 - 3 10 - 3 0 0 1 - 3 10 - 3 1 0 1 1 - 3 10 - 3 1 0 1 - 3 10 - 3 1 0 1 - 3 10 - 3 1 1 0 1 - 3 10 - 3 1 0 0 - 3 10 - 3 1 0 0 0

3 directional smoothing filtering templates, 3 direction sharpening filtering templates, 4 directional smoothing filtering templates, 4 direction sharpening filtering templates

0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 5 8 5 1 0 0 0 1 5 8 5 1 0 0 1 5 8 5 1 1 0 0 1 5 8 5 1 0 0 1 1 5 8 5 1 0 0 1 5 8 5 1 0 0 0 1 5 8 5 - 3 10 - 3 1 0 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 1 0 0 1 - 3 10 - 3 1 0 0 1 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 0 1 - 3 10 - 3

5 directional smoothing filtering templates, 5 direction sharpening filtering templates, 6 directional smoothing filtering templates, 6 direction sharpening filtering templates

8 5 1 0 0 0 0 5 8 1 1 1 1 0 1 5 8 5 5 1 0 0 1 5 8 5 1 0 0 1 5 5 8 5 1 0 1 1 1 5 8 5 0 0 0 0 1 5 8 10 - 3 1 0 0 0 0 - 3 10 1 1 1 1 0 1 - 3 10 - 3 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 - 3 10 - 3 1 0 1 1 1 - 3 10 - 3 0 0 0 0 1 - 3 10 5 1 1 0 0 0 0 8 5 5 1 1 0 0 5 8 8 5 1 1 0 1 5 5 8 5 5 1 0 1 1 5 8 8 5 0 0 1 1 5 5 8 0 0 0 0 1 1 5 - 3 1 1 0 0 0 0 10 - 3 - 3 1 1 0 0 - 3 10 10 - 3 1 1 0 1 - 3 - 3 10 - 3 - 3 1 0 1 1 - 3 10 10 - 3 0 0 1 1 - 3 - 3 10 0 0 0 0 1 1 - 3

7 directional smoothing filtering templates, 7 direction sharpening filtering templates, 8 directional smoothing filtering templates, 8 direction sharpening filtering templates

Wherein the smothing filtering template of 1 direction and sharpening filtering template are the filtering template of horizontal direction, and the template of all the other directions obtains by rotating 22.5 ° successively.

Respectively the filtering template on fingerprint image and eight directions is carried out convolution algorithm, promptly on all directions, use the level and smooth streakline of smothing filtering template earlier, with sharpening filtering template level and smooth back streakline is carried out sharpening again, strengthen image thereby obtain eight filtering on the direction respectively.During concrete operations, can be earlier the filtration module of Fig. 9 and direction 1 be carried out convolution algorithm, earlier smoothly, sharpening again, carry out successively according to the ordering of direction then.Figure 12 illustrates the flow process of this operation, as example, to A figure (image after the normalization), what it was done earlier is the filtering enhancing of 3 directions, B that can be from Figure 12, C figure sees, after the filtering enhancing through 3 directions, 3 direction streaklines in the G portion of image left side are more clearly, and the streakline of other direction is all not clear, when eight directions after all computing is finished, according to the piece field of direction that obtains eight images are made overlap-add procedure again, be about to gray values of pixel points on each direction and be changed to gray-scale value after template with this direction strengthens, promptly get enhancing image to the end, i.e. the figure of D among Figure 12.

Claims (2)

1, a kind of direction filtering reinforcement method of fingerprint image, its step is as follows:
1.1, the normalized of fingerprint image:
Adopt following formula that each gray values of pixel points in the fingerprint image is carried out normalized, obtain normalized fingerprint image:
(i j) is the Normalized Grey Level value of pixel in the image to g in the formula, and (i j) is the actual grey value of certain pixel to I, and m is the gray average of each pixel in the image, and v is a variance yields, m 0Be the gray average of each pixel expectation in the image, v 0Variance yields for expectation;
1.2, the some direction estimates
1.21, the mask A that sets up one 9 * 9 is as follows:
??7 ??6 ??5 ??4 ??3 ??7 ??5 ??3 ??8 ??7 ??6 ??5 ??4 ??3 ??2 ??8 ??7 ??5 ??3 ??2 ??1 ??1 ??1 ??1 ??* ??1 ??1 ??1 ??1 ??2 ??3 ??5 ??7 ??8 ??2 ??3 ??4 ??5 ??6 ??7 ??8 ??3 ??5 ??7 ??3 ??4 ??5 ??6 ??7
* among this mask A is pending pixel, represents eight possible streakline directions of this pixel by eight rectilinear directions of 1~8 array;
1.22, with pixel * is central point, calculate the average gray Gmean[i on described eight directions respectively] (i=1,8), then these eight directions are divided into 4 groups by direction perpendicular to each other, promptly 1 and 5 is one group, 2 and 6 is one group, 3 and 7 is one group, 4 and 8 is one group, calculate the difference G[i of two average gray in every group again]=| Gmean[i]-Gmean[i+4] |., get this absolute value differences G[i then] peaked both direction iMax, iMax+4 is as two of this pixel * possible streakline directions, if the gray-scale value at this pixel place is g (*), then the direction of this pixel is:
Promptly get this both direction iMax, iMax+4 gray average near the direction of the gray-scale value of this pixel * streakline direction as this pixel *,
1.23, all pixels in this operation traversal normalization fingerprint image, just obtained the some direction field pattern of normalization fingerprint image;
1.3, will put the direction field pattern and be converted into piece direction field pattern:
Resulting some direction field pattern of 1.2 steps is divided into the overlapped image block of W * W, add up the pixel number on eight directions in each image block then, getting the direction that pixel number is maximum in eight directions is the streakline piece direction of this image block, so operation, travel through each image block of entire image, obtain the piece direction field pattern of fingerprint image at last;
1.4, fingerprint image travel direction filtering after the normalization is strengthened: according to eight quantized directions of mask A defined, smothing filtering template and sharpening filtering template that the window with 7 * 7 is constructed eight directions are as follows:
0 0 0 0 0 0 0 1 1 1 1 1 1 1 5 5 5 5 5 5 5 8 8 8 8 8 8 8 5 5 5 5 5 5 5 1 1 1 1 1 1 1 0 0 0 0 0 0 0
1 directional smoothing filtering template
0 0 0 0 0 0 0 1 1 1 1 1 1 1 - 3 - 3 - 3 - 3 - 3 - 3 - 3 10 10 10 10 10 10 10 - 3 - 3 - 3 - 3 - 3 - 3 - 3 1 1 1 1 1 1 1 0 0 0 0 0 0 0
The sharp painstakingly filtering template of 1 direction
0 0 0 0 1 1 5 0 0 1 1 5 5 8 0 1 1 5 8 8 5 1 5 5 8 5 5 1 5 8 8 5 1 1 0 8 5 5 1 1 0 0 5 1 1 0 0 0 0
1 directional smoothing filtering template
0 0 0 0 1 1 - 3 0 0 1 1 - 3 - 3 10 0 1 1 - 3 10 10 - 3 1 - 3 - 3 10 - 3 - 3 1 - 3 10 10 - 3 1 1 0 10 - 3 - 3 1 1 0 0 - 3 1 1 0 0 0 0
2 direction sharpening filtering templates
0 0 0 0 1 5 8 0 1 1 1 1 8 5 0 1 5 5 8 5 1 0 1 5 8 5 1 0 1 5 8 5 5 1 0 5 8 5 1 1 1 0 8 5 1 0 0 0 0
3 directional smoothing filtering templates
0 0 0 0 1 - 3 10 0 1 1 1 1 10 - 3 0 1 - 3 - 3 10 - 3 1 0 1 - 3 10 - 3 1 0 1 - 3 10 - 3 - 3 1 0 - 3 10 - 3 1 1 1 0 10 - 3 1 0 0 0 0
3 direction sharpening filtering templates
0 0 0 1 5 8 5 0 0 1 5 8 5 1 0 1 1 5 8 5 1 0 1 5 8 5 1 0 1 5 8 5 1 1 0 1 5 8 5 1 0 0 5 8 5 1 0 0 0
4 directional smoothing filtering templates
0 0 0 1 - 3 10 - 3 0 0 1 - 3 10 - 3 1 0 1 1 - 3 10 - 3 1 0 1 - 3 10 - 3 1 0 1 - 3 10 - 3 1 1 0 1 - 3 10 - 3 1 0 0 - 3 10 - 3 1 0 0 0
4 direction sharpening filtering templates
0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0 0 1 5 8 5 1 0
5 directional smoothing filtering templates
0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0
5 direction sharpening filtering templates
5 8 5 1 0 0 0 1 5 8 5 1 0 0 1 5 8 5 1 1 0 0 1 5 8 5 1 0 0 1 1 5 8 5 1 0 0 1 5 8 5 1 0 0 0 1 5 8 5
6 directional smoothing filtering templates
- 3 10 - 3 1 0 0 0 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 1 0 0 1 - 3 10 - 3 1 0 0 1 1 - 3 10 - 3 1 0 0 1 - 3 10 - 3 1 0 0 0 1 - 3 10 - 3
6 direction sharpening filtering templates
8 5 1 0 0 0 0 5 8 1 1 1 1 0 1 5 8 5 5 1 0 0 1 5 8 5 1 0 0 1 5 5 8 5 1 0 1 1 1 5 8 5 0 0 0 0 1 5 8
7 directional smoothing filtering templates
10 - 3 1 0 0 0 0 - 3 10 1 1 1 1 0 1 - 3 10 - 3 - 3 1 0 0 1 - 3 10 - 3 1 0 0 1 - 3 - 3 10 - 3 1 0 1 1 1 - 3 10 - 3 0 0 0 0 1 - 3 10
7 direction sharpening filtering templates
5 1 1 0 0 0 0 8 5 5 1 1 0 0 5 8 8 5 1 1 0 1 5 5 8 5 5 1 0 1 1 5 8 8 5 0 0 1 1 5 5 8 0 0 0 0 1 1 5
8 directional smoothing filtering templates
- 3 1 1 0 0 0 0 10 - 3 - 3 1 1 0 0 - 3 10 10 - 3 1 1 0 1 - 3 - 3 10 - 3 - 3 1 0 1 1 - 3 10 10 - 3 0 0 1 1 - 3 - 3 10 0 0 0 0 1 1 - 3
8 direction sharpening filtering templates
Fingerprint image after the normalization is carried out filtering with the filtering template of eight directions successively, promptly earlier with the level and smooth streakline of smothing filtering template, with sharpening filtering template level and smooth back streakline is carried out sharpening again, thereby the filtering that obtains eight directions respectively strengthens image, again the image that eight filtering strengthen is superposeed the fingerprint image that is enhanced at last according to the piece field of direction that obtains then.
2, the direction filtering reinforcement method of fingerprint image according to claim 1, it is characterized in that: described some direction field pattern is converted into that the overlapped image block size that adopts is 16 * 16 pixels in the step of piece direction field pattern, overlapping width ω=1~15 pixel.
CNB2008100178757A 2008-04-02 2008-04-02 The direction filtering reinforcement method of fingerprint image CN100573554C (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CNB2008100178757A CN100573554C (en) 2008-04-02 2008-04-02 The direction filtering reinforcement method of fingerprint image

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CNB2008100178757A CN100573554C (en) 2008-04-02 2008-04-02 The direction filtering reinforcement method of fingerprint image

Publications (2)

Publication Number Publication Date
CN101271516A CN101271516A (en) 2008-09-24
CN100573554C true CN100573554C (en) 2009-12-23

Family

ID=40005481

Family Applications (1)

Application Number Title Priority Date Filing Date
CNB2008100178757A CN100573554C (en) 2008-04-02 2008-04-02 The direction filtering reinforcement method of fingerprint image

Country Status (1)

Country Link
CN (1) CN100573554C (en)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2011097935A1 (en) * 2010-02-11 2011-08-18 上海点佰趣信息科技有限公司 Binarization processing method for fingerprint image
CN104484679A (en) * 2014-09-17 2015-04-01 北京邮电大学 Non-standard gun shooting bullet trace image automatic identification method

Families Citing this family (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101770570B (en) * 2009-01-04 2013-09-25 茂晖科技股份有限公司 High-efficiency fingerprint image processing method
CN102368333A (en) * 2011-09-07 2012-03-07 常州蓝城信息科技有限公司 Fingerprint enhancement method based on time domain and frequency domain filtering
CN102930241B (en) * 2012-08-03 2015-07-22 北京天诚盛业科技有限公司 Fingerprint image processing method and processing device
CN102800066B (en) * 2012-08-03 2015-01-07 中山大学 Local extremum and pixel value gradient based improved image enhancement method
CN105205802B (en) * 2015-02-13 2017-04-12 比亚迪股份有限公司 Method and device for calculating ridge distance
CN104809464A (en) * 2015-05-19 2015-07-29 成都英力拓信息技术有限公司 Fingerprint information processing method
CN105740902B (en) * 2016-01-29 2019-06-11 广西大学 Fingerprint image Block direction accuracy estimation method
CN105844265B (en) * 2016-06-07 2019-07-16 Oppo广东移动通信有限公司 A kind of fingerprint image treating method and apparatus
CN106250887A (en) * 2016-09-09 2016-12-21 深圳市金立通信设备有限公司 A kind of fingerprint identification method and terminal
CN107451549B (en) * 2017-07-24 2020-05-12 云南大学 Non-contact fingerprint enhancement method and curvature-driven adaptive filter
CN109426795A (en) * 2017-09-05 2019-03-05 比亚迪股份有限公司 Fingerprint identification method and device
TWI650712B (en) * 2017-09-30 2019-02-11 北京集創北方科技股份有限公司 Fingerprint capture method and fingerprint capture module
CN109815772A (en) * 2017-11-20 2019-05-28 方正国际软件(北京)有限公司 Fingerprint enhancement, recognition methods, device and Fingerprint enhancement identifying system

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1480896A (en) * 2002-09-04 2004-03-10 长春鸿达光电子与生物统计识别技术有 Fingerprint identification method as well as fingerprint controlling method and system
CN1632823A (en) * 2003-12-24 2005-06-29 中国科学院自动化研究所 Automatic fingerprint classification system and method

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN1480896A (en) * 2002-09-04 2004-03-10 长春鸿达光电子与生物统计识别技术有 Fingerprint identification method as well as fingerprint controlling method and system
CN1632823A (en) * 2003-12-24 2005-06-29 中国科学院自动化研究所 Automatic fingerprint classification system and method

Non-Patent Citations (4)

* Cited by examiner, † Cited by third party
Title
. 司马海峰.基于指纹曲线特征的预处理算法研究. 2007
. 司马海峰.基于指纹曲线特征的预处理算法研究. 2007 *
一种改进的指纹图像增强算法. 陈静等.电脑知识与技术(学术交流),第2007 2期. 2007
一种改进的指纹图像增强算法. 陈静等.电脑知识与技术(学术交流),第2007 2期. 2007 *

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2011097935A1 (en) * 2010-02-11 2011-08-18 上海点佰趣信息科技有限公司 Binarization processing method for fingerprint image
CN104484679A (en) * 2014-09-17 2015-04-01 北京邮电大学 Non-standard gun shooting bullet trace image automatic identification method

Also Published As

Publication number Publication date
CN101271516A (en) 2008-09-24

Similar Documents

Publication Publication Date Title
US10726260B2 (en) Feature extraction and matching for biometric authentication
Yu et al. Image denoising using trivariate shrinkage filter in the wavelet domain and joint bilateral filter in the spatial domain
Salembier Morphological multiscale segmentation for image coding
Xu et al. Image inpainting by patch propagation using patch sparsity
CN101980284B (en) Two-scale sparse representation-based color image noise reduction method
US6072895A (en) System and method using minutiae pruning for fingerprint image processing
US7020591B1 (en) Partial differential equation model for image feature extraction and identification
JP4340553B2 (en) Biometric information verification device
O'Gorman et al. An approach to fingerprint filter design
CN101339607B (en) Human face recognition method and system, human face recognition model training method and system
JP5047005B2 (en) Image processing method, pattern detection method, pattern recognition method, and image processing apparatus
CN100412883C (en) Fingerprint identifying method and system
US6314197B1 (en) Determining an alignment estimation between two (fingerprint) images
Raja Fingerprint recognition using minutia score matching
EP1724725B1 (en) Finger/palm print image processing system and finger/palm print image processing method
Wu et al. Fingerprint image enhancement method using directional median filter
US7142699B2 (en) Fingerprint matching using ridge feature maps
US20140133711A1 (en) Biometric information correction apparatus, biometric information correction method and computer-readable recording medium for biometric information correction
US7496214B2 (en) Method of palm print identification
Rizon et al. Object detection using circular Hough transform
CN102096795B (en) Method for recognizing worn two-dimensional barcode image
Karnati et al. Fast non-local algorithm for image denoising
US20070274570A1 (en) Iris recognition system having image quality metrics
US6631212B1 (en) Twostage scheme for texture segmentation based on clustering using a first set of features and refinement using a second set of features
CN102184533B (en) Non-local-restriction-based total variation image deblurring method

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
C17 Cessation of patent right
CF01 Termination of patent right due to non-payment of annual fee

Granted publication date: 20091223

Termination date: 20100402