WO2011097936A1

WO2011097936A1 - Fingerprint image enhancement method

Info

Publication number: WO2011097936A1
Application number: PCT/CN2010/080604
Authority: WO
Inventors: 陈晓峰; 刘君
Original assignee: 上海点佰趣信息科技有限公司
Priority date: 2010-02-11
Filing date: 2010-12-31
Publication date: 2011-08-18
Also published as: CN101794380A; CN101794380B

Abstract

An enhancement method of a fingerprint image is disclosed, which comprises the following steps: analyzing from the perspective of a frequency domain and removing the high-frequency component of an original fingerprint image by the fuzzy processing of the image; then, subtracting the fuzzed fingerprint image by utilizing the original image to remove the low-frequency component of the image so as to obtain relatively pure image high-frequency information with the low-frequency information of the image being already removed substantially; and afterwards, superposing the high-frequency information image with the original image to obtain the image, the high-frequency part thereof is enhanced, i.e. the sharpened image. Further, the influence of the nonuniformity of an illumination field is eliminated by utilizing the difference of the sharpened image and the sharpened image processed by fuzzification. Thus, the obtained enhanced fingerprint image is relatively ideal and it is convenient to select a threshold value in a subsequent image processing process when binarization processing is carried out on the fingerprint image, thereby lowering the probability of the erroneous extraction of the characteristic of a fingerprint and improving the accuracy of fingerprint identification.

Description

Fingerprint image enhancement method

The present invention relates to a fingerprint image preprocessing method, and in particular to a fingerprint image enhancement method. Background technique

With the advancement of society, the security of identity has received increasing attention. Traditional identification often uses documents, passwords, and so on. However, the certificate may be lost or copied; the password is easily forgotten or confused. Especially with the advent of the Internet age, more and more password settings are plaguing people: power-on password, email password, bank password, forum password... For these, if you set the same password, it will increase security risks. ; If you set a different password, it also brings trouble to password management. For this reason, identification technologies based on biometrics (eg, fingerprints, faces, irises, etc.) are gaining increasing attention. Among them, the recognition rate of fingerprint recognition is high, and the application is the most popular, and it is recognized as the "first of physical evidence."

In the process of fingerprint recognition, the image captured by the fingerprint acquisition device is a brightness map containing more noise, which must be preprocessed to remove a large amount of noise, and a dot line diagram with clear lines can be obtained. Extraction and matching of fingerprint features. Image enhancement is a very important part of the fingerprint image preprocessing process. The effect will directly affect the accuracy of feature extraction, and will also affect the threshold selection during subsequent binarization. Among them, the so-called gray image binarization is to change the image into a binary image that uses only two gray values to represent the foreground and background colors of the image, and the threshold value is related to the fingerprint feature extraction. The correctness.

Existing fingerprint image enhancement technologies mainly include histogram modification processing, image smoothing processing, low-pass filtering, and high-pass filtering, such as Laplace method, Wallis filtering method, median filtering method, and the like. However, these methods either increase the background noise while enhancing the fingerprint lines, or blur the fingerprint lines while reducing the background noise. For this reason, when the collected fingerprint image is not clear enough and the ridge line is more difficult to recognize, the existing fingerprint image enhancement technology has a poor processing effect, which often leads to the erroneous extraction of the fingerprint feature. In addition, when the fingerprint itself is severely scratched or worn, the existing fingerprint The processing effect of the enhanced technology is also difficult to meet the requirements. Summary of the invention

The object of the present invention is to provide a fingerprint image enhancement method, which can improve the effect of fingerprint image enhancement processing, reduce the probability of fingerprint feature error extraction, and improve the accuracy of fingerprint recognition.

To this end, the present invention provides a fingerprint image enhancement method, including: acquiring an original fingerprint image; performing blurring processing on the original fingerprint image to obtain a first blurred image; and subtracting the first blurred image from the original fingerprint image to obtain a a high-frequency information image; superimposing the high-frequency information image with the original fingerprint image to obtain a sharpened image; blurring the sharpened image to obtain a second blurred image; using the sharpened image minus Go to the second blurred image to get the enhanced fingerprint image.

Further, the process of superimposing the high frequency information image with the original fingerprint image comprises: adjusting the brightness value of the original fingerprint image as a whole; and superimposing the high frequency information image with the original fingerprint image of the adjusted brightness value.

Further, the adjustment coefficient of adjusting the brightness value of the original fingerprint image is greater than 0.

Further, the adjustment coefficient of adjusting the brightness value of the original fingerprint image is greater than or equal to 0.7.

Further, the process of performing fuzzification processing on the original fingerprint image adopts Gaussian blur processing

Further, the template of the Gaussian blur processing is a 7*7 template, that is,

Further, when the enhanced fingerprint image is binarized, the threshold is between -5 and +5.

In summary, in the above fingerprint image enhancement method, after the original image is blurred, the frequency is The domain angle analysis removes the high-frequency components of the image, and the obtained first blurred image mainly contains the low-frequency component of the image; then subtracting the first blurred image from the original image is equivalent to removing the low-frequency component of the original image. The obtained high-frequency information of the relatively pure image is obtained, and the low-frequency information of the image is basically removed; then the high-frequency information of the image is superimposed with the original image, and the image obtained by the high-frequency part is sharpened, that is, the sharpened image is obtained. Further, the influence of the unevenness of the illumination field is eliminated by using the difference between the sharpened image and the second blurred image (ie, the sharpened image after the blurring process). In this way, the obtained enhanced fingerprint image is ideal, which facilitates the selection of the threshold value when the binarization processing is performed in the subsequent image processing process, thereby reducing the probability of fingerprint feature error extraction and improving the accuracy of fingerprint recognition. DRAWINGS

1 is a front-back comparison diagram of a fingerprint image subjected to sharpening in an embodiment of the present invention; FIG. 2 is a gray line diagram of a fingerprint image and a blurred image of the image according to an embodiment of the present invention;

3 is a gray-scale difference line diagram of the fingerprint image and the blurred image thereof in FIG. 2;

4 is a schematic flowchart of a fingerprint image enhancement method according to an embodiment of the present invention; FIG. 5 is a schematic flowchart of a fingerprint image enhancement method according to another embodiment of the present invention; FIG. 6 is an image enhancement process according to an embodiment of the present invention; Schematic diagram of the results of the fingerprint image. detailed description

In order to make the technical features of the present invention more comprehensible, the present invention will be further described below in conjunction with the specific embodiments.

In view of the problems existing in the existing fingerprint image enhancement technology, the present invention adopts an improved diffusion mask sharpening (USM, Unsharp Mask) method for preprocessing the fingerprint image. The basic idea is to superimpose a clear fingerprint image positive image and a blurred fingerprint image negative image, and then superimpose it. A fingerprint image positive film whose brightness value is adjusted to obtain a sharpened fingerprint image positive film. Formulaize the process as follows:

g (x, y) = K - f(x, y) + (f(x, y) - f (x, y)) ( 1 ) where g J) is the sharpened image, f(x, y ) is the original image, f (x, y) is the blurred image of the original image. To adjust the adjustment factor of the brightness value of the original fingerprint image, it is greater than 0, usually greater than or equal to 0.7 to meet the requirements.

The principle of sharpening the enhancement effect of the fingerprint image is as follows:

The original image is blurred, and the analysis from the frequency domain angle removes the high-frequency components of the image; then subtracting the blurred fingerprint image from the original image is equivalent to removing the low-frequency components of the original image, and obtaining a relatively pure image. High-frequency information, and the low-frequency information of the image has been basically removed; and then the high-frequency information of the image is superimposed with the original image, and the image obtained by the high-frequency portion is sharpened, that is, the sharpened image. As shown in Fig. 1, it is the front-back comparison of the fingerprint image after the sharpening process, wherein the left side is the fingerprint image before processing, and the right side is the sharpened fingerprint image.

As can be seen from the figure, the image has a phenomenon of left dark and right, which is affected by uneven illumination field. In response to this processing defect, the present invention further improves the above sharpening method, and the basic implementation principle and process are as follows:

Assuming that the sharpened fingerprint image is represented as a 256*256 matrix, the data of the 127th row of the matrix is taken as an example analysis. Its gray line chart is shown in the solid line in Figure 2:

This polyline can be seen as a section of line 127 of the fingerprint image. Wherein, the ordinate value is gray value information, 0-255 means dark to light, and the abscissa represents the number of columns in the matrix. The undulations of the fold lines in the figure represent the stripes of the fingerprint, and the valleys and peaks of the fold lines represent the ridges (dark stripes) and valleys (light stripes) of the fingerprints in the fingerprint image, respectively. Because of the uneven illumination field, the fold line does not vibrate in the vicinity of a horizontal line as a whole, so it is difficult to take out a fixed threshold to binarize the fold line in the figure; that is, no horizontal line can be found, so that all the peaks are Above it, and all the valleys are below it. Thus, it is extremely disadvantageous for the subsequent fingerprint image binarization processing. The dotted line in the figure shows the gray line diagram of the fingerprint image after the blurring process. Due to the fuzzification process, the undulation of the dotted line at the peaks and valleys of the line is obviously weakened, but in other areas, the original is better. Real fold line.

Further consideration, if the difference between the real fold line and the dashed fold line is examined, the difference should be positive at the fingerprint valley (the fold line peak); at the fingerprint ridge (the fold line valley), the difference is negative; The value is near 0. Thereby, a gradation difference line graph as shown in Fig. 3 can be obtained. As can be seen from the figure, the polyline is generally close to vibrating near a horizontal line, so it is easier to take a fixed threshold to binarize the polyline in the graph; that is, you can find a horizontal line so that all the peaks are above it. And all the valleys are below it. In addition, it can be seen that the horizontal line is substantially between -5 and +5, so when the enhanced fingerprint image is binarized, the threshold is often taken between -5 and +5. As a result, the uneven effect of the illumination light field will be eliminated, which facilitates the selection of thresholds for binarization processing in subsequent image processing, thereby reducing the probability of fingerprint feature error extraction and improving the accuracy of fingerprint recognition.

Therefore, the fingerprint image enhancement method is summarized. One embodiment of the present invention provides a fingerprint image enhancement method as shown in FIG. 4, which includes the following steps:

S 1 : obtaining an original fingerprint image;

S2: Obscuring the original fingerprint image to obtain a first blurred image;

S3: subtracting the first blurred image from the original fingerprint image to obtain a high-frequency information image; S4: superimposing the high-frequency information image with the original fingerprint image to obtain a sharpened image; S5: sharpening Obscuring the image to obtain a second blurred image;

S6: Subtracting the second blurred image from the sharpened image to obtain an enhanced fingerprint image. It can be known from the above analysis that after the fingerprint image sharpening processing of steps S2 to S4, the low frequency information of the image is basically removed, and the enhanced image of the high frequency portion is obtained, thereby realizing the enhancement of the fingerprint image. However, if there is an uneven illumination field in the enhancement process, the subsequent fingerprint image binarization process is affected. For this reason, the present invention further utilizes the sharpened image and the sharpened image. The difference between the image after the image (ie the second blurred image) The effect of uneven field. In this way, the obtained enhanced fingerprint image is ideal, which facilitates the selection of the threshold value when the binarization processing is performed in the subsequent image processing process, thereby reducing the probability of fingerprint feature error extraction and improving the accuracy of fingerprint recognition.

The original image superimposed in the above step S4 may be the original image, but preferably the original fingerprint image whose overall brightness value is adjusted. Therefore, in another embodiment shown in FIG. 5, step S4 includes the following steps:

S41: adjusting the brightness value of the original fingerprint image as a whole;

S42: Superimpose the high frequency information image with the original fingerprint image with the adjusted brightness value.

Wherein, the adjustment coefficient of adjusting the brightness value of the original fingerprint image is K, and K>0, preferably, κ >

0.7.

In the above process, the blurring of the image can use Gaussian blur. Gaussian fuzzy is a fuzzy method with mean. In the experiment, the inventor used 3*3 template, 5*5 template, 7*7 template and other methods. After experiment comparison, the effect of using 7*7 template is better. it is good.

The Gaussian fuzzy template looks like this:

1 1 1

3*3 template: 1 1 1

1 1 1

5*5 template: 丄

7*7 template: 丄

49

Please refer to FIG. 6 , which is a fingerprint image result after the above image enhancement processing, wherein, left The fingerprint image after sharpening the fingerprint image through the above steps S2 to S4 is the fingerprint image further enhanced by the above steps S5 and S6. The comparison shows that the fingerprint image on the left is dark and the left and right, and the fingerprint image on the right has no such phenomenon, that is, the influence of the illumination field unevenness has been eliminated, which is beneficial to the subsequent binarization of the fingerprint image.

The above is only an example and is not intended to limit the invention, and the scope of the invention should be determined by the scope of the claims.

Claims

Rights request

A fingerprint image enhancement method, comprising:

Obtain the original fingerprint image;

Obscuring the original fingerprint image to obtain a first blurred image;

Subtracting the first blurred image from the original fingerprint image to obtain a high frequency information image;

Superimposing the high-frequency information image and the original fingerprint image to obtain a sharpened image; and blurring the sharpened image to obtain a second blurred image;

The second blurred image is subtracted from the sharpened image to obtain an enhanced fingerprint image.

2. The fingerprint image enhancement method according to claim 1, wherein the process of superimposing the high frequency information image with the original fingerprint image comprises:

Adjusting the brightness value of the original fingerprint image as a whole;

The high frequency information image is superimposed with the original fingerprint image whose adjusted brightness value is adjusted.

3. The fingerprint image enhancement method according to claim 2, wherein the adjustment coefficient of the brightness value of the original fingerprint image is adjusted to be greater than zero.

4. The fingerprint image enhancement method according to claim 3, wherein the adjustment coefficient of adjusting the brightness value of the original fingerprint image is greater than or equal to 0.7.

The fingerprint image enhancement method according to claim 1, wherein the process of blurring the original fingerprint image adopts Gaussian blur processing.

6. The fingerprint image enhancement method according to claim 5, wherein the Gaussian blur plate is a 7*7 mode, that is,

The fingerprint image enhancement method according to claim 1, wherein when the enhanced fingerprint image is binarized, the threshold is between -5 and +5.