KR100795187B1 - Device for recognizing fingerprint and method thereof - Google Patents

Abstract

A device and a method for recognizing a fingerprint are provided to detect feature points after a normalization process, which corrects a ridge image with an average width of the ridge, without a preprocessing treatment for separating the ridge image and a background image. A ridge normalizer(40) generates a ridge image by correcting a ridge of an inputted fingerprint image with the advertiser width of the ridge. A feature point extractor(50) generates a feature point image by generating a mask with the average width calculated in the ridge normalizer and detecting the feature points of the ridge image by the generated mask. The normalizer comprises a scanner(42), an operator(44) calculating the average ridge width by measuring a distance between valleys or the ridges of the scanned fingerprint image, and a corrector(46). The feature point extractor comprises a mask generator(52), a searcher(54) searching the feature points of the ridge by checking a position and a direction of the ridge with the mask generated in the mask generator, and a filter(56).

Description

Fingerprint recognition device and method

1 is a flowchart illustrating a conventional fingerprint recognition process.

Figure 2 is a block diagram showing a fingerprint recognition device according to a preferred embodiment of the present invention.

Figure 3 is a sequence diagram showing a fingerprint recognition process according to a preferred embodiment of the present invention.

4 is a flowchart illustrating in detail the ridge normalization process constituting a preferred embodiment of the present invention.

Figure 5 is a sequence diagram showing in detail the feature point extraction process according to a preferred embodiment of the present invention.

6a to 6h are exemplary views showing a state in which a fingerprint image input by the preferred embodiment of the present invention is processed.

* Description of the symbols for the main parts of the drawings

10: input part 20: area selection part

30: normalization unit 40: ridge normalization unit

42: scanning unit 44: computing unit

46: correction unit 50: feature point extraction unit

52: mask generator 53: mask

54: search unit 56: filter

60: matching unit 70: database

The present invention relates to a fingerprint recognition device, and more particularly, to a fingerprint recognition device and method for extracting a feature point after the ridge normalization process of the input fingerprint without a preprocessing process.

1 is a sequence diagram illustrating a conventional fingerprint recognition process.

As shown in the drawing, the conventional fingerprint recognition method first starts from receiving a fingerprint from a user (S10). Input of the fingerprint is input by various methods, but generally, light emitted from the inside of the input unit is reflected on the ridges and valleys of the fingerprint, and the sensor receives the reflected light to form a fingerprint image.

The input fingerprint image is transformed into a form for extracting the feature point through the preprocessing process (S20). Hereinafter, the preprocessing process will be described in detail.

First, only the portions necessary for extracting the feature points are divided and separated (S21) and the segmented image is normalized (S22). In this case, the standardization means to make the density of the image constant. That is, the density of the input image varies depending on conditions such as humidity and pressure of the input portion, and the feature points are extracted as it is. In this case, since it is difficult to know the exact feature point location and direction, the density of the image is constantly adjusted so that one image has the same histogram distribution.

Next, in order to remove the noise included in the fingerprint image, the flow information of the fingerprint is required. For this, the direction of the fingerprint is extracted (S23). That is, the direction of the entire fingerprint is extracted and the fingerprint image is restored in a later step based on this.

After that, the frequency of the ridge is measured (S24). The frequency measurement of the ridges is different in size and magnification of the input image, so that the frequency of the ridges is matched to facilitate matching after feature point extraction.

Then, as the most important part of the preprocessing, an algorithm for separating the background region and the fingerprint region is performed. Typically, the fingerprint region is divided by reconstructing the fingerprint region using the maximum response using a Gabor filter. (S25).

Then, the reconstructed fingerprint image is binarized (S26). The binarization simplifies the complex shading of the fingerprint image into black and white.

Next, the binarized image is thinned (S27). The thinning refers to making an image of a ridge having various thicknesses into one line segment with reference to the direction of the ridge extracted in the directional extraction step.

As described above, when the thinning operation is completed, the image of the fingerprint is preprocessed for feature point extraction.

Then, the feature points of the ridges are extracted according to the thinned fingerprint ridges (S30). The feature point means a branch point where the ridge branches and a disadvantage in which the ridge ends.

The extracted feature point is matched with the feature point of the stored fingerprint image to determine whether the input feature point is the same as the stored fingerprint image (S40).

However, the prior art as described above has the following problems.

In other words, the preprocessing process is performed to recognize the fingerprint, and this preprocessing process has a problem that it takes a long time to recognize the fingerprint due to a large burden on the system.

In the prior art, when the fingerprint image is not clear, the Gabor filtering process for separating the ridge image and the background image is complicated, which takes a very long time, and the accuracy is also very low.

Accordingly, the present invention has been made to solve the above-mentioned conventional problems, and an object of the present invention is to normalize the ridge image with the average width of the ridge without preprocessing for separating the ridge image and the background image such as Gabor filtering. The present invention provides a fingerprint recognition apparatus and method for extracting feature points after going through a task only.

Another object of the present invention, in order to extract the feature point using the ridge image modified to the average width without thinning, generating a mask having a pixel having the average width of one side length and extracting the feature point using the same A fingerprint recognition device and method are provided.

 According to a feature of the present invention for achieving the above object, the present invention comprises a ridge normalization unit for generating a ridge image by modifying the ridge of the input fingerprint image to the average width of the input ridge; And a feature point extracting unit generating a mask using the average width calculated by the ridge normalization unit, and generating a feature point image by detecting feature points of the ridge image using the generated mask.

At this time, the ridge normalization unit, the scanning unit for scanning the input fingerprint image; A calculating unit calculating an average width of the ridges by measuring the distance between the bones or the distance between the ridges of the scanned fingerprint image; And a correction unit configured to generate a ridge image by correcting the ridge of the input fingerprint image to the calculated average width.

The calculation unit may calculate the area of the fingerprint ridge by measuring at least one of the distance between the centers of the valleys, the distance between the centers of the ridges, the area of the valleys, or the area of the ridges, and from the calculated average value of the ridge area, Calculate the average width of the ridge.

In this case, the correction unit corrects the ridge of the input fingerprint to have the same width as the average width calculated by the calculator to generate a ridge image.

The feature point extracting unit may include: a mask generating unit generating a mask for extracting feature points by inspecting a position and a direction of the ridges; A searcher for searching for a feature point of the ridge by inspecting the position and direction of the ridge using the mask generated by the mask generator; And it may be configured to include a filter for generating a feature point image by removing the ridge image except the feature point retrieved by the search unit.

The mask generating unit generates a rectangular mask made up of a plurality of pixels having the average width of one side.

In this case, the mask may be a matrix mask having n rows and n columns of pixels, and the variable n may be an integer of 3 ≦ n ≦ 32.

Meanwhile, the searcher moves the mask by one pixel in length to scan the fingerprint image to search for feature points of the fingerprint image.

At this time, the feature point, at least one of the disadvantages, branch points or inflection points of the ridge.

The search unit searches for the position and direction of the feature point.

Meanwhile, the filter removes a portion of the ridge image except for the feature point searched by the search unit and converts the image size into a set reference size to generate the feature point image.

On the other hand, the present invention may further comprise a region selection unit for selectively taking only the region containing the ridge of the fingerprint image input from the user.

The present invention may further include a normalization unit for generating a fingerprint image by adjusting the histogram distribution of the fingerprint image selected by the region selection unit to the same level.

The present invention may further include a matching unit for comparing and searching the feature point image extracted from the feature point extractor with reference information stored in a database.

On the other hand, the present invention (A) an input step of receiving a fingerprint image from the user; (B) a normalization step of generating a fingerprint image by matching the histogram distribution of the input fingerprint image to the same level; (C) a ridge normalization step of generating a ridge image by normalizing the ridges of the normalized fingerprint image to an average value; And (D) generating a mask with an average width of the ridge, and extracting a feature point by scanning the ridge image using the mask to generate a feature point image.

At this time, the step (C), (C1) measuring the thickness of the ridge while scanning the fingerprint image; (C2) calculating a mean width from the measured thickness of the ridges; And (C3) correcting the fingerprint image with the calculated average width to generate a ridge image.

And (D) comprises: (D1) generating a rectangular mask composed of a plurality of square pixels having the average width of one side; (D2) a search step of searching for a feature point of a fingerprint using the generated mask; And (D3) a filtering step of removing fingerprint images other than the searched feature points to extract only the image of the feature points; (D4) converting the size of the filtered image to a predetermined reference size to generate a feature point image.

In this case, the mask is a square matrix mask having pixels of three rows and three columns or more.

On the other hand, the feature point, any one or more of the disadvantages of the ridge, branch point or inflection point.

In this case, the method may further include a matching step of checking whether the fingerprint of the user matches the fingerprint of the stored database by comparing the feature point image with the fingerprint information stored in the database.

According to the present invention having the above-described configuration, the fingerprint recognition process can be simplified since the feature points of the fingerprint can be extracted without the process of separating the ridge image and the background image such as Gabor filtering and the preprocessing process such as thinning. There is this.

Hereinafter, with reference to the accompanying drawings, preferred embodiments of the fingerprint recognition device and method according to the present invention as described above will be described in detail.

2 is a block diagram showing a fingerprint recognition device according to an embodiment of the present invention.

As shown in the drawing, the fingerprint recognition device according to the present invention includes an input unit 10 for receiving a fingerprint image from a user. The input unit 10 generally consists of a scanning device.

On the other hand, the input unit 10 is connected to the area selection unit 20 for selecting a region of the input fingerprint image. The fingerprint image input to the input unit 10 is a part for selecting only the portion where the fingerprint is input and deleting the portion where the fingerprint is not input since most of the user does not input the fingerprint at the center of the input unit 10. to be.

The region selector 20 is provided with a normalizer 30 for constantly adjusting the overall density of the input fingerprint image. The image of the input fingerprint has a different concentration depending on the portion, which is caused by factors such as the degree of moisture content and the pressure applied when the user inputs the fingerprint. Therefore, the normalization unit 30 is a portion that constantly adjusts the distribution of unequal histograms of the fingerprint image selected by the region selection unit 20 to generate a fingerprint image having the same overall density.

In addition, the ridge normalization unit 40 is connected to the normalization unit 30 to correct the ridge of the fingerprint image adjusted to a constant density by the normalization unit 30 to a line of the same thickness having the average width of the ridges. .

The ridge normalizing unit 40 includes a scanning unit 42, a calculating unit 44, and a correcting unit 46. The scanning unit 42 includes a fingerprint image input from the normalizing unit 30. This is to calculate the width of each ridge by scanning it. The width of the ridge may be obtained by measuring the distance between the ridges, or may be obtained using the distance between the centers of the ridges, the distance between the centers of the valleys, and the width of the valleys. The calculation unit 44 connected to the scanning unit 42 calculates an average width of all the ridges of the fingerprint image using the width of the ridges measured by the scanning unit 42.

Alternatively, the ridge width may be calculated by sampling a plurality of portions of the fingerprint image, and the average width thereof may be used as the average width of the ridge.

The compensator 46 connected to the calculator 44 corrects the ridge of the fingerprint image using the average width calculated by the calculator 44 to generate a ridge image. The correction unit 46 corrects the thickness of the ridge to the average width while moving along each ridge.

That is, the ridge normalization unit 40 is a portion for uniformly correcting the thickness of the ridge of the input fingerprint image to generate the ridge image.

On the other hand, the ridge normalization unit 40 is provided with a feature point extraction unit 50 for extracting the feature point of the ridge image. The feature point includes a disadvantage indicating the start point and the end point of the ridge and a branching point branched into two or more branches. In addition, it is also possible to include the inflection point of the ridge in the feature point for a more precise comparison.

In order to extract the feature points of the ridge image, the feature point extractor 50 includes a mask generator 52, a searcher 54, and a filter 56. The mask generator 52 generates a mask for extracting feature points from the ridge image, and the mask is a mask having pixels having a forward matrix structure. In this case, one pixel constituting the mask has the same length as the average width calculated by the calculator 44.

The mask is preferably at least 3 × 3 or more mask. This is because the mask must be able to detect ridges traveling in different directions from the branched position in order to detect the branch point.

Of course, the larger the arrangement of the mask, the more detailed feature can be detected, but if the mask is excessively large, the feature extraction time can be long, it is preferable to use a mask arranged in an appropriate number Do.

Meanwhile, the searcher 54 is connected to the mask generator 52 to search for the ridge image using the mask generated by the mask generator 52.

The searcher 54 scans the entire image by moving the mask by one pixel unit. At this time, the searcher 54 detects a feature point including a disadvantage and a branch point of the ridge on the mask.

The filter 56 is connected to the search unit 54. The filter 56 deletes ridge image portions other than the feature points retrieved by the searcher 54 to generate an image in which only the feature points are displayed. Furthermore, the filter 56 preferably converts the image into a set reference size to generate a feature point image. This is a part for facilitating a comparison operation performed in the matching unit to be described later.

On the other hand, the filter 56 is connected to the matching unit for comparing the feature point image generated through the filter 56 and the fingerprint image stored in the data storage unit. A database 70 is connected to the matching unit, and the matching unit reads fingerprint information to be compared with the feature point image generated from the input fingerprint image from the database 70.

Hereinafter, the operation of the fingerprint recognition device according to the present invention having such a configuration will be described in detail according to a fingerprint recognition process.

3 is a flowchart illustrating a fingerprint recognition process according to a preferred embodiment of the present invention, Figure 4 is a sequential diagram showing in detail the ridge normalization process constituting a preferred embodiment of the present invention, Figure 5 is a Figure 6 is a sequential diagram illustrating the feature point extraction process according to a preferred embodiment, Figure 6a to Figure 6h is an exemplary view showing a state in which the fingerprint image input by the preferred embodiment of the present invention is processed.

In the following description, each step will be described with reference to FIG. 6, which is an example of an image that is actually processed in these steps.

As shown in these figures, in order to recognize a fingerprint by a preferred embodiment of the present invention, first receives a fingerprint image from the user through the input unit 10 (S100). In this case, the input fingerprint image is selected and stored only in the fingerprint input portion of the fingerprint image by the area selection unit 20 (S200). That is, since the fingerprint image input by the user is not exactly input at the center, as shown in FIG. 6A, only the portion where the fingerprint is input by the area selection unit 20 is cut out and the fingerprint image as shown in FIG. 6B. Create

The concentration of the fingerprint image is uniformly uniformed by the normalization unit 30 (S300). Accordingly, the nonuniform histogram of the fingerprint image is uniformly corrected to generate a fingerprint image. That is, the fingerprint image shown in FIG. 6B has a dark density at the center portion and a light density at the outer portion thereof. However, as shown in FIG. 6C through the normalization unit 30, the fingerprint image has a uniform density throughout. Is generated.

Thereafter, the fingerprint ridge width of the fingerprint image is normalized to the average width (S400). The ridges are normalized to generate ridges, which will be described in detail below.

First, in order to normalize the ridge, the average width of the ridge should be measured. For this purpose, the width of the ridge is measured while scanning the fingerprint image (S420). And the calculation unit 44 calculates the average width of the ridge using the measured width of the ridge (S440).

After calculating the average width of the ridges, the ridge width of the fingerprint image is corrected to the average width to generate a ridge image (S460).

As described above, in the fingerprint image input through the ridge normalization unit 40, the width of the ridge is constantly modified to generate the ridge image.

The ridge image is generated in the manner as described above is shown in Figure 6d. The width of the ridge of FIG. 6D is modified to the average width to have the same width.

The ridge image is input to the feature point extraction unit 50 to extract the feature points (S500). Here, the feature point refers to a feature of the fingerprint, and includes a disadvantage and a branch point of the ridge, and may also include an inflection point.

The processing procedure of the feature point extraction unit 50 for extracting the feature point is shown in detail in FIG. 5. As shown in FIG. 1, first, a mask is generated using the average width of the ridges in the mask provided in the feature point extractor 50 (S520). The mask is formed of pixels having the average width of one side, and the pixels are formed of many rows and many columns.

Preferably, the mask is formed in a matrix form in which the pixels are combined in a square of 3 × 3 or more, and one side is composed of three or more pixels to detect the direction at the feature point. 6E shows a state in which a 3 × 3 mask 53 is formed.

The mask 53 extracts the feature points included in the ridge image by moving the ridge image to a length of one pixel unit (S540). 6F illustrates a case where the mask 53 is located at a feature point. As shown in the drawing, when the mask 53 is located at a feature point (look for the case where the feature point is located on the pixels of two rows and two columns of the mask 53), the ridge image is detected from the ridge image detected at the peripheral pixels of the feature point. Identify the direction of feature points.

The mask 53 scans the entire ridge image to extract all the feature points of the ridge image. As such, the appearance of the ridge image from which the feature points are extracted is shown in FIG. 6G.

Next, the ridge image is erased by the filter 56 except for the extracted feature points, and an image in which only the feature points remain is generated (S560).

In addition, preferably the feature image is a magnification of the same size as the fingerprint information stored in the data storage in order to facilitate comparison with the fingerprint information stored in the data storage in the subsequent matching step to generate a feature point image ( S580). At this time, the conversion of the magnification is preferably converted based on the average width of the ridge. The feature point image generated as described above is illustrated in FIG. 6H.

As described above, when the feature point image is generated, the feature point image is compared with the fingerprint image stored in the database 70 to determine whether the input fingerprint is the same as the fingerprint information stored in the database 70. At this time, only the fingerprint information stored in the database 70 is also extracted and stored.

The rights of the present invention are not limited to the embodiments described above, but are defined by the claims, and those skilled in the art can make various modifications and adaptations within the scope of the claims. It is self-evident.

For example, the step of changing the magnification of the feature point image may have the same effect even if performed in another execution step of the fingerprint recognition process.

In the fingerprint recognition apparatus and method according to the present invention as described in detail above, the following effects can be expected.

That is, as in the conventional fingerprint recognition method, the step of separating the ridge image and the background image, such as the Gabor filtering step, is not included, and the processing speed is increased, and the pixel size is adjusted to the average width of the ridge without the thinning process. Since the same effect as that of the thinning process can be obtained, the overall fingerprint recognition process can be simplified, thereby improving the processing speed of the system.

And since the present invention does not include the Gabor filtering and thinning step, the directional extraction or ridge frequency extraction step of the ridge, which is a preprocessing step for the Gabor filtering or thinning process, is not necessary, thereby simplifying the fingerprint recognition process and thus error occurrence frequency. Has the advantage of being lowered.

Claims (21)

A ridge normalizer for generating a ridge image by correcting the ridge of the input fingerprint image to an average width of the input ridge; And a feature point extractor for generating a mask using the average width calculated by the ridge normalization unit and detecting a feature point of the ridge image using the generated mask. . The method of claim 1, The ridge normalization unit, A scanning unit scanning the input fingerprint image; A calculating unit calculating an average width of the ridges by measuring the distance between the bones or the distance between the ridges of the scanned fingerprint image; And And a correction unit configured to generate a ridge image by correcting the ridge of the input fingerprint image to the calculated average width. The method of claim 2, The calculation unit, The width of the fingerprint ridge is calculated by measuring at least one of the distance between the centers of the valleys, the distance between the centers of the ridges, the area of the valleys, or the area of the ridges, and the average width of the ridges is calculated from the average value of the calculated ridge areas. Fingerprint recognition device characterized in that. The method of claim 3, wherein The correction unit, And a ridge image is generated by correcting the ridge of the input fingerprint to have the same width as the average width calculated by the calculating unit. The method according to any one of claims 1 to 4, The feature point extraction unit, A mask generator for generating a mask for extracting feature points by inspecting the position and direction of the ridges; A searcher for searching for a feature point of the ridge by inspecting the position and direction of the ridge using the mask generated by the mask generator; And And a filter unit for generating a feature point image by removing the ridge image except for the feature point searched by the search unit. The method of claim 5, The mask generator, And a rectangular mask composed of a plurality of pixels having the average width of one side. The method of claim 6, The mask is, and a matrix mask having pixels of n rows and n columns. The method of claim 7, wherein The variable n, Fingerprint recognition device characterized in that an integer of 3≤n≤32. The method of claim 6, The search unit, And moving the mask in units of one pixel's length to scan the fingerprint image to search for a feature point of the fingerprint image. The method of claim 9, The feature point, Fingerprint implant device, characterized in that it comprises any one or more of the disadvantages, branch points or inflection points of the ridges. The method of claim 10, The search unit, Fingerprint recognition device characterized in that for searching the position and direction of the feature point. The method of claim 11, The filter, And removing a portion of the ridge image except for the feature point searched by the searching unit, and converting the image size into a predetermined reference size to generate a feature point image. The method of claim 1, And an area selector which selectively takes only an area including the ridges among the fingerprint images input from the user. The method of claim 13, And a normalization unit configured to generate a fingerprint image by adjusting a histogram distribution of the fingerprint image selected by the region selection unit to the same level. The method of claim 14, And a matching unit configured to compare and search the feature point image extracted from the feature point extractor with reference information stored in a database. (A) an input step of receiving a fingerprint image from the user; (B) a normalization step of generating a fingerprint image by matching the histogram distribution of the input fingerprint image to the same level; (C) a ridge normalization step of generating a ridge image by normalizing the ridges of the normalized fingerprint image to an average value; And (D) a fingerprint recognition method comprising generating a mask with an average width of the ridge, and extracting a feature point by scanning the ridge image using the mask to generate a feature point image. The method of claim 16, Step (C) is (C1) measuring the thickness of the ridge while scanning the fingerprint image; (C2) calculating a mean width from the measured thickness of the ridges; And (C3) a fingerprint recognition method comprising the step of correcting the fingerprint image to the calculated average width to generate a ridge image. The method according to claim 16 or 17, Step (D), (D1) generating a rectangular mask made up of a plurality of square pixels having the average width of one side; (D2) a search step of searching for a feature point of a fingerprint using the generated mask; And (D3) a filtering step of removing fingerprint images other than the searched feature points to extract only the image of the feature points; (D4) a fingerprint recognition method comprising converting the size of the filtered image into a set reference size to generate a feature point image. The method of claim 18, The mask is, Fingerprint recognition method characterized in that the square matrix mask having three rows or more pixels. The method of claim 19, The feature point, Fingerprint recognition method characterized in that it comprises any one or more of the disadvantages, branch points or inflection points of the ridge. The method of claim 20, And a matching step of comparing the feature point image with fingerprint information stored in a database and checking whether a user's fingerprint matches a fingerprint of a stored database.

Images