JP2003337949A - Fingerprint matching device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a fingerprint matching device which can identify a person with high accuracy and also securely collate a fingerprint through a re-input operation in case of a failure in identification and has a small computation quantity. <P>SOLUTION: The fingerprint matching device comprises an image input part 1 for inputting a fingerprint image, a picture quality decision part 2 which divides the fingerprint image into a plurality of small areas, calculates ridge directions by the small areas, and decides the picture quality of the fingerprint image, a storage part 8 which stores the ridge directions, an image cutting-out part 3 which outputs image information obtained by cutting out a specified area around a center point according to the ridge directions, an image compensation part 4 which performs luminance corrections and smoothing corrections based upon the ridge directions and outputs results, an arithmetic part 5 which calculates and outputs identification information to be used for collation, and an identification information storage part 7 which stores the identification information. <P>COPYRIGHT: (C)2004,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fingerprint collation device that authenticates an individual by using a fingerprint image of a user.

[0002]

2. Description of the Related Art In recent years, in information devices such as mobile phone devices and personal computers, with the progress of networks, user authentication technology has become important for the purpose of protecting personal information when the network is utilized. .

As this user authentication method, an authentication method using biometrics has been widely used in place of the conventionally used method for each user password. Among them, the most widely used is a fingerprint collation device that collects a user's fingerprint image and collates it with a pre-registered fingerprint image to perform personal authentication.

In the case of an authentication method using such a fingerprint collation device, the state of the fingertip varies with time,
It was always difficult to collect good fingerprint images. For example, if the finger is wet immediately after washing the hand, or if it is abnormally dry, a good fingerprint image cannot be obtained. In addition, the pressing force when pressing the finger against the detection means such as the optical prism or the electrostatic capacitance element is insufficient or excessive,
Even if the finger shakes when acquiring the fingerprint image, a good fingerprint image cannot be acquired, and stable acquisition of the fingerprint image has been a major problem of the fingerprint collation device.

For this reason, in the conventional fingerprint collation device, the quality of the input fingerprint image is judged to be good or bad prior to the collation or registration. If the image quality is poor, the collation or registration is not performed and the user is informed. It is configured to request the re-input and input the fingerprint image again. As a result, the fingerprint image determined to have poor image quality is not used for collation, registration, etc., so that it is possible to improve the collation accuracy of personal authentication.

[0006]

In order to realize such a fingerprint collation device, as a conventional technique for determining the image quality of a fingerprint image, for example, identification information such as a branch, a dot, or a break from the fingerprint image is used. A method has been proposed in which the characteristic points are detected and the number of the characteristic points is used to determine whether the fingerprint is crushed or blurred (eg, Japanese Patent Laid-Open No. 8-129644). According to such a method, it becomes extremely difficult to extract a feature point when, for example, a ridge is completely crushed and there is almost no difference in luminance, and it is often the case that the crushed fingerprint image is accurately determined. There were things I couldn't do. Further, when such a method is used in a fingerprint collation apparatus that uses information other than feature points or feature amounts such as frequency distribution as identification information, a large amount of calculation is required to calculate feature points. This had to be done separately, which increased the load on the device and eventually led to an increase in the size of the device.

Further, there has been proposed a method in which a fingerprint image is divided into small areas, and when the brightness of the small areas is biased to a certain value, it is determined that the ridge lines are broken in the area (eg, Japanese Patent Laid-Open No. Hei 10-1999). 10-187983). According to this method
Even if the ridge line in the small area is not completely collapsed in reality, it may be determined to be collapsed if the brightness distribution is biased, and the user is required to re-input more than necessary. There was a problem that the burden was increased. Further, this method has a problem that it is not possible to make a judgment when the fingerprint is faint.

Further, in these conventional techniques, when an appropriate fingerprint image cannot be obtained, the factor of image quality deterioration, which is why the appropriate image cannot be obtained, is not clarified, and it is appropriate for the user. Since the guidance is not performed, the user simply repeats the input as instructed by the fingerprint collation device without taking any effective measures, which eventually causes a problem that the burden on the user becomes heavy. It was

The present invention has been made in consideration of the above points, and in order to minimize the re-input request, a fingerprint collation device capable of obtaining a good fingerprint image capable of performing personal authentication with a high authentication rate. It is possible to identify the factor that deteriorates the image quality of the fingerprint image as well as to provide the user with a proper guide and instruction on how to input the next time when re-inputting. An object of the present invention is to provide a fingerprint collation device that can reduce the burden. It is also an object of the present invention to provide a fingerprint collation device with a small amount of calculation, which is suitable even in a configuration that uses information other than a characteristic point such as a frequency distribution or a characteristic amount as identification information.

[0010]

A fingerprint collating apparatus according to the present invention divides a fingerprint image into a plurality of small areas and an image input section into which a fingerprint image to be registered is input, and a small area containing valid information. The fingerprint area is detected and the ridge direction is calculated for each small area that constitutes the fingerprint area, and the image quality of the fingerprint image is determined. If the image is determined to be a good image, the image quality determination that outputs the fingerprint image is performed. Section, a ridge direction calculated by the image quality determination section, for each small area, and a center point of the fingerprint image output from the image quality determination section based on the ridge direction stored in the storage section. Is calculated, and the image cutout portion that outputs image information obtained by cutting out a predetermined area around the center point and the image information output from the image cutout portion are subjected to luminance correction and smoothing based on the ridge direction. From the image compensation section that performs the digitalization correction and outputs From the force image information, calculates the identification information to be used for verification, and an arithmetic unit for outputting, characterized in that an identification information storage unit that stores the identification information output from the operation unit.

With such a configuration, an image that is determined to be a good image by the image quality determination unit can be registered in the identification information storage unit using a fingerprint image that has been brightness-corrected or smoothed. A high fingerprint collation device can be obtained.

Further, the once calculated ridge direction for each small area is stored in the storage unit, and the center point of the fingerprint area is calculated and smoothing correction is performed using the ridge direction, so that the total calculation amount can be reduced. .

Next, the fingerprint collation apparatus of the present invention divides the fingerprint image into a plurality of small areas and an image input section into which a fingerprint image to be collated is input, and detects a small area containing valid information. As a fingerprint area, the ridge direction is calculated for each small area that constitutes the fingerprint area, the image quality of the fingerprint image is determined, and an image quality determination unit that outputs a fingerprint image when determined to be a good image, The storage unit that stores the ridge direction calculated by the image quality determination unit for each small area, and the center point of the fingerprint image output from the image quality determination unit is calculated based on the ridge direction stored in the storage unit. , Luminance correction and smoothing correction based on the ridge direction are performed on the image cutout portion that outputs image information obtained by cutting out a predetermined area around the center point and the image information output from the image cutout portion. The image compensation unit that performs and outputs the The identification information storage unit that stores the identification information to be used for the calculation, and the calculation unit that calculates the identification information from the image information output from the image compensation unit, compares the identification information with the identification information stored in the identification information storage unit, and performs collation. It is characterized by having and.

With such a configuration, the image quality determination is performed before the authentication is performed using the fingerprint image, and the personal identification is performed using the fingerprint image that is determined to be a good image. Obtainable.

Furthermore, by performing image compensation such as brightness correction and smoothing prior to personal authentication, it is possible to obtain a fingerprint collation device having a high authentication rate.

Furthermore, since the ridge direction for each region once calculated is stored in the storage unit, and the center point of the fingerprint region is calculated and smoothing correction is performed using the same, the overall calculation amount can be reduced. .

Further, the image quality judgment unit further detects the small area in the fingerprint area in which the direction of the ridge cannot be obtained as the direction undetectable area, thereby making the image quality judgment more efficient. It becomes possible to quantitatively
When performing image quality adjustment such as brightness correction, processing can be selectively performed for each small area.

Further, the image quality determination unit further determines that the fingerprint image is defective information when the number of detected direction undetectable areas accounts for a predetermined ratio or more with respect to the number of small areas occupying the fingerprint area. By determining that there is, it is possible to perform the image quality determination more efficiently and with good reproducibility.

Further, the image quality judging section further detects a ridge collapsed area in which the ridge is collapsed in the detected direction undetectable area, and the number of the ridge collapsed areas occupies the fingerprint area. By deciding that the fingerprint image is defective information when occupying a predetermined ratio or more with respect to the number of small areas,
It is possible to efficiently detect a defective image when the fingerprint image is crushed when the finger is wet when the fingerprint is input and the pressing force is too large.

Further, the image quality judging section further detects a ridge fading area in which the ridge is faint in the detected direction undetectable area, and the number of the ridge fading areas occupies the fingerprint area. By deciding that the fingerprint image is defective information when occupying a predetermined ratio or more with respect to the number of small areas,
It is possible to efficiently detect a defective image when the fingerprint image is faint when the finger when inputting a fingerprint is abnormally dry or when the finger moves and shakes during imaging.

Further, the image of the direction undetectable area is binarized, and the value of each small area forming the direction undetectable area becomes "
When the ratio of the number of pixels of "0" to the number of pixels forming the small area is equal to or larger than a predetermined ratio, the small area is detected as the ridge line collapse area, and the ridge line collapse area is detected by a simple calculation. It is possible to

In addition, the image of the direction undetectable area is binarized, and the value is "0" for each small area forming the direction undetectable area.
When the ratio of the number of pixels of 1 "to the number of small areas forming the small area is equal to or larger than a predetermined ratio, the small area is detected as the ridge fading area, and the ridge fading area is calculated by a simple calculation. Can be detected.

Further, the image quality determining section further includes a display section connected to the arithmetic section and the image quality determining section for displaying a message to the user, and when the image quality determining section determines that the fingerprint image is defective information, the user can replay it. When a message prompting for input is displayed on the display unit, the image deterioration factor of the defective image is identified. Therefore, when a predetermined message is displayed for the image deterioration factor to prompt the user to re-input. In addition, effective guidance and instruction can be performed, and a fingerprint collation device with a high authentication rate can be realized.

Further, the image compensating unit divides the image information into a plurality of regions and performs the luminance correction based on the luminance histogram in each region, thereby performing the luminance correction for each of the plurality of divided regions. Even in an image in which a part of the image changes in brightness extremely, it is possible to perform effective brightness correction for the entire image.

Further, the image compensating unit divides the image information into a plurality of areas and corrects the brightness based on the brightness histogram in the areas determined to be the fingerprint areas in each area, thereby excluding the areas other than the fingerprint areas. It has nothing to do with fingerprint images such as backgrounds,
It is possible to perform the brightness correction by removing the influence of the extremely high brightness part.

Further, the image compensating unit divides the image information into a plurality of areas, and performs the brightness correction based on the brightness histogram in the areas excluding the areas determined to be the direction undetectable areas in each area. By doing so, it is possible to remove the adverse effects of the direction undetectable region such as crushing and blurring, and perform brightness correction of the image including the background.

Further, the image compensator divides the image information into a plurality of areas, and corrects the brightness based on the brightness histogram in the area where the ridge direction is detected in each area. It is possible to remove the adverse effects of the area in which is not detected, for example, the direction undetectable area and the background area, and perform the brightness correction.

Further, the image compensating unit divides the image information into a plurality of areas and corrects the brightness from the brightness histogram in each area and in a predetermined range of areas surrounding the area.
It is possible to obtain an image with high brightness continuity in the image after the brightness correction for each area.

Further, the image compensating section further filters the luminance-corrected image information in the direction along the ridge direction stored in the storage section for each small area to obtain the luminance value. By smoothing, it is possible to provide an image with less noise and suitable for fingerprint authentication. Further, since the ridge direction for each area calculated in advance is used, it is possible to obtain a fingerprint collation device that requires a small amount of calculation.

Further, the image compensating section further filters the direction-undetectable area of the brightness-corrected image information in the same direction to smooth the brightness value, thereby eliminating the direction-undetectable area and the surrounding area. It is possible to obtain an image with less noise that maintains continuity with the region, and it is possible to realize a higher authentication rate.

Furthermore, the fingerprint collation apparatus of the present invention divides the fingerprint image into a plurality of small areas, calculates the ridge direction for each small area, and for each small area in which the ridge direction cannot be calculated. The feature is that the image quality deterioration factor is specified based on the binarized image obtained by binarizing the image of the small area in which the ridge direction cannot be calculated.

According to such a configuration, it is possible to perform particularly suitable image quality determination and identification of image quality deterioration factors in a configuration in which information other than feature points such as frequency distribution or feature amount is used as identification information. It is possible to provide a fingerprint collation device with a small amount of calculation.

[0033]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. As an embodiment of the present invention, an example of the configuration of a fingerprint matching device will be described in detail.

FIG. 1 is a block diagram showing a configuration of a fingerprint collation device 10 according to the embodiment of the present invention.

The fingerprint collation device 10 of the present embodiment displays an image input unit 1 formed of a capacitance type or optical type device into which a user's fingerprint image is input, and a fingerprint image output from the image input unit 1. Information about the ridge direction for each small region calculated by the image quality determination unit 2 that determines the image quality of the fingerprint image while dividing the image into small regions and calculating the ridge direction for each small region. Of the storage unit 8 for storing the
Image cutout unit 3 that outputs image information obtained by cutting out a fingerprint area including valid information, and performs brightness correction and smoothing processing on the image information output from image cutout unit 3 for registration and matching processing. An image compensating unit 4 that appropriately performs preprocessing of image information, an identification information storage unit 7 that stores identification information calculated from a fingerprint image and to be used for matching, and an identification information storage unit 7 that is connected to store the identification information. An arithmetic unit 5, which performs identification authentication by collating the identification information stored in the unit 7 with the identification information calculated from the image information output from the image compensation unit 4,
In addition, it has a display unit 6 which is connected to the image quality determination unit 2 and the calculation unit 5 and is composed of a display device such as an LCD or an EL.

Next, a processing flow in the fingerprint collation device 10 of the present embodiment will be described. FIG. 2 is a flowchart showing a processing flow of the fingerprint collation device 10 according to the embodiment of the present invention.

First, the user touches the image input unit 1 composed of a capacitance type sensor or an optical type sensor to obtain a multi-gradation (multi-value) fingerprint image 41 as shown in FIG. 3 (step SP1). In the fingerprint image 41, a portion where the fingerprint touches the sensor, that is, a ridge portion is darker, and other portions where the fingerprint is not touched are captured as brighter images. In the fingerprint matching device of the present invention,
The input method of the image input unit 1 is not limited.

Next, the image quality judgment unit 2 divides the fingerprint image 41 obtained by the image input unit 1 into small areas, calculates the ridge direction for each small area, and stores the result in the storage unit 8. The image quality of the fingerprint image 41 is stored and stored (step SP2). If the image quality determination unit 2 determines that the image quality is good, the process proceeds to the image information extraction process (step SP3) by the image extraction unit 3. On the other hand, when the image quality determination unit 2 determines that the image quality is poor, the display unit 6 displays that the image quality is poor and prompts the user to re-input (step SP7). Details of the processing in the image quality determination unit 2 will be described later.

If the image quality determination unit 2 determines that the image quality is good, the image cutout unit 3 uses the information about the ridge direction for each small area stored in the storage unit 8 and the fingerprint image 41.
The center point of the image is detected, and the image information of a small area located in a predetermined range with the center point as the center is cut out from the fingerprint image 41 (step SP3). Even with the same finger of the same user, the same place is not obtained as an image each time the fingerprint image 41 is acquired, and it is inevitable that the fingertip position varies each time the fingerprint image 41 is used for authentication. . Therefore, by cutting out with the center point as the center, it is possible to always cut out only the image information in the same range for the same fingerprint, and the collation accuracy of the fingerprint is improved.

Subsequently, the image compensating unit 4 performs brightness correction, image smoothing correction, etc. on the image information cut out by the image cutting unit 3 in order to emphasize the pattern drawn by the ridge of the fingerprint. (Step SP4). Details of the processing contents in the image compensator 4 will be described later.

Next, the calculation section 5 calculates the identification information to be used for collation from the image information output from the image compensation section 4 (step SP5).

Although the present invention does not limit the type of identification information calculated in step SP5, for example, the position or number (feature amount) of characteristic points such as breaks, branches, points, etc., and frequency analysis are performed. It is possible to use known identification information such as a frequency spectrum distribution calculated as a result, or a result of performing a cepstrum analysis or the like.

Subsequently, when the user is registered, the identification information calculated based on the fingerprint image 41 of the user to be registered is transmitted from the arithmetic unit 5 to the identification information storage unit 7 to identify the user. It is stored in the information storage unit 7. On the other hand, when the user's identity verification (authentication) is performed, the identification information calculated by the calculation unit 5 based on the fingerprint image 41 of the user to be verified and the identification information stored in the identification information storage unit 7 in advance. The identification information is collated to determine whether the input fingerprint image 41 belongs to the registered user (step SP).
6). The determination result of the calculation unit 5 is displayed on the display unit 6 and transmitted to the user.

Next, the processing in the image quality judgment unit 2 will be described in detail.

FIG. 4 is a flow chart showing the processing in the image quality judging section 2. In FIG. 4, first, the image quality determination unit 2
Shows the fingerprint image 42 output from the image input unit 1 in FIG.
As shown in (a), it is divided into a plurality of small areas 43 (step SP8). In addition, here, for simplification of description, the fingerprint image 42 is simplified and conceptualized. Also, regarding division into small areas 43, in practice, an image of vertical × horizontal = 512 × 256 pixels is divided into vertical × horizontal = 64 × 32 small areas for each 8 × 8 pixel and processed. However, in order to simplify the description here, vertical x horizontal = 80
X64 pixel image, vertical x horizontal = 8 x 8 = 10 x 8
It shows an example of division into individual small areas 43.

Then, for each of the divided small areas 43, the small area 43 in which it is detected that the ridge is imaged is extracted (step SP9). As a result, as shown in FIG. 5B, the area surrounded by the diagonal lines is determined to be the fingerprint area 44 in which the ridge is imaged.

Next, the ridge direction is obtained for each of the small areas 43 constituting the fingerprint area 44 extracted in step SP9. This is done by creating eight directions of filters oriented in any direction and filtering the pixels forming each small area 43 using each filter for each pixel (step SP10). As a result, which of the eight directions the ridge direction is for each pixel is calculated, and the direction occupying the largest number of pixels in the area is determined as the direction of the small area 43 (step SP11).

This processing will be described with reference to FIGS. 6 and 7. FIG. 6 is a diagram showing an example of the configuration of the eight-direction filters 21 to 28 for determining the ridge direction for each small region 43.

The filters 21 to 28 shown in FIG. 6 are digital filters which perform filtering so as to take the absolute value of the difference in luminance value between the hatched pixels (two pixels) with respect to the pixel of interest in the center. .

Hereinafter, the case where the ridge direction is calculated for the small area S in FIG. 5A will be described as an example.

As shown in FIG. 7, assuming that the enlarged view of the small area S for each pixel is shown in FIG. 7A, the luminance information of all the pixels forming the small area S is obtained. Then, the ridge direction is determined by performing a filtering operation using the above filters 21 to 28 one pixel at a time.

For example, the luminance values of a target pixel T and peripheral pixels (region of vertical × horizontal = 3 pixels × 3 pixels) are shown in FIG.
In the case of the relationship as shown in (b), the pixel of interest T is first filtered by sequentially using the eight types of filters 21 to 28 shown in FIG. 6, and a value corresponding to each direction is provided for each filter. To calculate.

For example, when the filter 21 is used,
Since the filter 21 is a filter that takes the absolute value of the difference between the brightness values of the pixels on both sides of the pixel of interest T, the result of the filtering is | 141-68 | = 73, and this value is in the direction D1, that is, in the horizontal direction. It becomes a calculated value.

Direction D1 (horizontal direction: 0 °): 141-68 = 73 (filter 21) Similarly, when calculation is performed using filters 22 to 28 in sequence, direction D2 (22.5 °): 68-20 = 48 (filter 22) direction D3 (45 °): 22-20 = 2 (filter 23) direction D4 (67.5 °): 22-13 = 9 (filter 24) direction D5 (vertical direction: 90 °): 130-13 = 117 (filter 25) direction D6 (112.5 °): 130-123 = 7 (filter 26) direction D7 (135 °): 156-123 = 33 (filter 27) direction D8 (157.5 °) ): 156-68 = 88 (filter 28), and the calculated value in the direction D3, that is, the value filtered using the filter 23 (in this case, "2"), is the smallest. Direction D the direction of the T
3 (45 °). If the change in luminance is small in any direction, the direction is not calculated, and the ridge direction cannot be calculated for that pixel, and the process ends.

In this way, the ridge direction is calculated for all the pixels forming the small area S. FIG. 7C is a plot of the calculated ridge direction for all pixels. Next, of the calculated ridge directions, the ridge direction that is the most in the small area S is used to determine the ridge direction of the small area S.

For example, in the case of the result shown in FIG. 7C, the relationship between the calculated ridge direction and the number of pixels is: direction D1 (0 °) 10 pixel direction D2 (22.5 °) 10 The pixel direction D3 (45 °) is 16 pixels, and as a result, the ridge direction of this small region S is changed to the direction D3 (45 °).
Can be determined (step SP1)
1).

However, even in the direction component obtained most,
When the absolute number of pixels serving as the direction component is less than or equal to a predetermined number, the small area 43 is set as the direction undetectable area,
Output (step SP11).

An example of the direction component of each small area 43 and the direction undetectable area 45 obtained as described above is shown in FIG.
Shown in. Here, the ridge direction is classified into eight directions as described above, and is horizontal (0 °) with respect to the paper surface, 22.5 °, 45 °, 67.5 °, 90 °, 112.
Any one of 5 °, 135 °, and 157.5 ° is shown. Further, the direction undetectable region 45 is shown by hatching.

The ridge direction data for each small area 43 calculated in this way is sent to the storage unit 8 in the fingerprint collation apparatus 10 and stored in the storage unit 8 and used for the subsequent processing at any time. As a result, the amount of calculation of the fingerprint collation device 10 as a whole can be reduced.

Returning to FIG. 4 again, the image quality judging unit 2 judges whether or not the ratio of the direction undetectable region 45 to the fingerprint region 44 is equal to or higher than a predetermined ratio (step SP12), and is equal to or higher than the predetermined ratio. If there is, it is determined as a defective image and the user is prompted to re-input the fingerprint image (step SP17). On the other hand, if the ratio is less than or equal to the predetermined ratio, it is determined that the image is a good image, the image quality determination unit 2 ends the process, and the next image cutout unit 3 performs the process (step SP15).

In step SP12, the predetermined ratio of the direction undetectable area 45 used for the image quality determination in the fingerprint area 44 varies depending on the type of the sensor used in the image input unit 1, the humidity, etc. 10% to 30
It was found that the good image and the bad image can be satisfactorily discriminated by setting the range up to%.

Further, FIG. 9 shows another processing flow in the image quality judging section 2. In this processing flow, after detecting the direction undetectable area 45 (step SP11), the image quality determination unit 2 determines the fingerprint state for each small area of the direction undetectable area 45 (step SP13) and determines the direction. The reason why the direction cannot be detected is specified for each undetectable area 45 (step SP14).

Specifically, it is detected whether the image in the direction undetectable area 45 is crushed in black or faint in white.

Therefore, for example, as shown in FIG.
A binarized image 46 of the fingerprint image 42 is created, and for each direction undetectable region 45, a small region 4 at the same position of the binarized image 46.
3, the black pixels (output: “0”) of the binarized image 46 with respect to the total number of pixels in the same small area 43.
Is calculated, and if the ratio is larger than a predetermined ratio, the direction undetectable region 45 is determined to be a blackout region,
On the contrary, if the ratio of white pixels (output: “1”) to the total number of pixels is larger than a predetermined ratio, the direction undetectable area 45 is determined to be a blurred area, and each direction undetectable area 45 is determined. Then, the state is detected.

In some cases, it may be determined that it is not included in either the blackout area or the blurred area. This is because the direction of the ridge is extremely curved in the direction undetectable area 45. The factors can be determined to be in case of a point.

In the case of the present embodiment, for example, in the case of the present embodiment, the number of pixels forming each small area 43 is 8 × 8.
There are 64 pixels in total, but if 80% or more of these are black, it is determined to be a blackout area, and if 80% or more is white, it is determined to be a blurred area.
In addition, it is determined that the direction of the ridge is extremely curved or a point within the small area 43.
By setting this predetermined ratio to a value of 70% to 90% depending on the sensor that constitutes the image input unit 1, the external environment, and the like, it is possible to excellently detect a blurred image and a blurred image.

Next, by performing these procedures for the small areas forming all the direction undetectable areas 45,
The number of small areas that are shaded in black and small areas that are shaded in white is counted to identify the factor of image quality deterioration (step SP14).

That is, these small areas which are blacked out,
Alternatively, when the ratio of the number of small areas that are blurred in white to the number of small areas that form the fingerprint area 44 is equal to or greater than a predetermined value, the fingerprint image can be determined to be a blurred image or a blurred image. According to experiments, this predetermined ratio is determined to be a defective image as a blurred image or a blurred image when the ratio of the blackened region or the blurred region to the fingerprint region 44 is approximately 10% or more. It is preferable to make a determination (step SP16). This ratio is also set to a value of 5% to 20% depending on the characteristics of the sensor that constitutes the image input unit 1 and factors such as the environment, so that a good image and a bad image can be discriminated well. Become.

If it is determined in step SP16 that the image is a crushed image, a message is displayed on the display unit 6 to warn the user that the fingerprint image has been crushed, and how to input the fingerprint image when prompting again. By displaying a guide ("Wipe your finger again, enter again", "Please put your finger on it a little more", etc.), it becomes possible for the user to make appropriate input, and to perform reliable authentication. (Step S17).

Similarly, when it is determined that the image is a blurred image, a message for warning the user that the image is blurred is displayed on the display unit 6 and a guide of how to input when prompting for another input (" Please put your finger on it a little more ”)) can be displayed.

On the other hand, when the image quality determination unit 2 determines that the image is a good image (step SP16), the process proceeds to the image cutout unit 3 process (step S15) in FIG. finish. The image cutout unit 3 cuts out a fingerprint image in the following steps.

In the image cutout unit 3, first, as shown in FIG. 11, the center position of the fingerprint image 42 is obtained. Figure 1
In the example shown in FIG. 1, the ridge direction information for each small area 43, which has been calculated by the image quality determination unit 2 and stored in the storage unit 8 is used. The vertical component and the horizontal component in the ridge direction of each small region 43 are calculated, and the small regions 43 adjacent to each other in the vertical direction and the small regions 43 adjacent to each other in the horizontal direction are calculated.
The distributions are calculated for each column and each row by adding together. Next, the peak points of each distribution are calculated, and the point connecting the respective peak points is determined as the center position PP of the fingerprint image.

Next, in the image cutting section 3, as shown in FIG. 12, the image information 47 in a predetermined range is cut out with the determined center position PP as the center. Here, the fingerprint image 42
An example will be shown in which 300 × 200 pixels are cut out from the number of pixels forming the vertical × horizontal = 512 × 256 pixels. By performing the cutout in this way, it is possible to make the sizes of the image information 47 used for matching uniform for the same fingerprint image 42 regardless of the conditions when the information was acquired.

Returning to FIG. 2 again, when the image cutout by the image cutout unit 3 (step SP3) is completed, the process shifts to the brightness correction (ridge enhancement) and smoothing correction processing by the image compensation unit 4 (step SP4). Next, the processing in the image compensator 4 will be described in detail.

In the image compensator 4, for example, the image information 47 shown in FIG. 12 is divided into a plurality of middle areas 48 as shown in FIG. 13 (for convenience, this is called the middle area. The size correlation is not limited and may be divided into regions of any size). Here, the division is made only in the vertical direction, but any division method may be used. A luminance histogram is taken for each of the middle regions 48, and the luminance is corrected so that the distribution has a constant distribution.

That is, the relationship between the number of pixels and the luminance in a certain middle area 48, that is, the luminance histogram is shown in FIG.
In the case of the relationship as shown in FIG. 14A, the gradation of the luminance is lowered (for example, 256 gradations → 64 gradations), and the relationship shown in FIG.
Luminance correction (contrast adjustment) is performed by correcting the relationship shown in (b), that is, by correcting the luminance histogram to be substantially constant.

A more specific description will be given. When the luminance histogram shown in FIG. 14A is totally integrated, it becomes equal to the number of pixels forming the middle region 48. For example, when the number of pixels forming the middle region 48 is N, the total integration value of the brightness histogram shown in FIG. 14A is N.

In the brightness correction, the brightness histogram (256 gradations) of FIG. 14A is integrated in order from gradation 0 to calculate gradation T having an area of (N / 64), and gradation 0 is calculated. For the pixels up to the gradation T, the gradation is corrected to 1. Thereafter, this is repeated for each area (N / 64), and the luminance correction is performed so that the luminance histogram shown in FIG.

By performing the brightness correction as described above, the image information 49 as shown in FIG.
The brightness can be corrected so that the image information 50 has a good contrast in the entire area as shown in (b).

Such a brightness correction may be performed for all the small areas forming the middle area 48, or for each of the middle areas 48, the brightness may be calculated only for the small area judged by the image quality judging section 2 to be the fingerprint area 44. Correction may be performed. in this case,
It is possible to perform brightness correction that is less affected by the brightness value of the area that is not determined as the fingerprint area 44, that is, the background or the like.

Further, the brightness correction may be carried out for each small area in each middle area 48, excluding the small area which is judged as the direction undetectable area 45 by the image quality judging section 2. In this case, it is possible to perform the brightness correction that is less affected by the blackout or blurring of the direction undetectable area 45.

Further, this brightness correction is performed for each middle area 48.
Then, the image quality determination unit 2 may be configured to perform only on the small area in which the ridge direction is detected. In this case, it is possible to perform the brightness correction that is less affected by both the area such as the background that is not determined to be the fingerprint area 44 and the direction undetectable area 45 such as a crush or a blur.

Furthermore, when the brightness correction is performed on a certain middle area 48, a brightness histogram including the surrounding middle areas 48 is also taken, and the result is reflected in the target brightness correction of the middle area 48. May be. That is, FIG.
In the case of performing the brightness correction of the area P, the brightness histogram of the area including the areas O and Q is taken in addition to the area P, and the result is reflected in the brightness correction of the area P.

According to this structure, it is possible to perform the luminance correction with less variation in the luminance of each middle area 48 and with higher continuity of the luminance of the entire image.

These brightness correction processes are applied to all middle areas 48.
Is performed, the unevenness of the luminance of the entire image information is corrected, and the image in which the shade of the ridge is clear is obtained as a whole.

Next, the smoothing process of the image information will be described. Similarly, returning to step SP4 in FIG. 2, the image compensating unit 4 performs the smoothing process on the luminance-corrected image information 50. This smoothing process is most effective in a fingerprint matching device that uses a matching method that does not use feature points as identification information, such as a frequency analysis method or a cepstrum analysis method, when performing personal identification using a fingerprint pattern. .

When smoothing the image information, normally,
In many cases, the method of blurring vertically and horizontally is used, but in the present invention, the method of blurring in the direction parallel to the direction is used by using the information of the ridge direction of each small area 43 calculated by the image quality determination unit 2. . However, in the direction non-detectable area 45, the blurring is fixed vertically and horizontally. By doing so, it becomes possible to remove noise without losing the structure of the ridge.

The smoothing process is performed in accordance with the ridge direction of each small area 43, as shown in FIG.
From 8 types of filters 31 to 38 of × 5 pixels,
This is performed by selecting a filter according to the ridge direction of each small area 43 stored in the storage unit 8 and filtering all the pixels forming each small area 43 using the selected filter.

In the filters 31 to 38 shown in FIG. 16, the pixel at the center of each filter is the pixel of interest, and the luminance value of the pixel of interest is based on the coefficient given to each pixel including peripheral pixels. Weighting is performed to determine the value of the pixel of interest. In addition, in FIG. 16, the coefficient “0” is stored in the pixel in which the coefficient is not described.

Specifically, for example, in the filter 31 (direction D1), as shown in FIG. 16, the luminance value of the target pixel E in the center is multiplied by the coefficient "8", and the pixels in the peripheral pixels are also integrated. Coefficient "2", "8", or "0" for each
Is added up and the total number is added up, that is, (8 + 4 × 2 + 2 × 6 = 28) in this example.
And is determined as the brightness value of the pixel of interest E at the center.

Such smoothing processing is performed on all the pixels forming the small area 43.

Further, in the direction undetectable area 45, vertical x horizontal = 3 x 3 in which blur can be made uniformly in all directions.
A pixel filter 39 is used.

By performing such smoothing processing, it is possible to smooth the image information without abrupt luminance change and with less noise without losing the information in the ridge direction of the image information.

FIG. 17 shows an example of the image information that has been smoothed and corrected (shown for one small area). The image information 51 is the image information before the smoothing correction, and the image information 52 is the image information of the small area after the smoothing correction. In the example shown in FIG. 17, the ridge direction of the small area of the image information 51 is the direction D2.
(22.5 °), and smoothing correction was performed using the filter 32. As a result, as shown in the image information 52 of FIG. 17, it was possible to reduce the noise component without losing the ridge information of the image information 51.

As a result, at the time of collation or registration in the subsequent calculation unit 5 (step SP6), it is possible to provide the image information 52 with a small noise and the like which can realize a high authentication rate. This smoothing correction is particularly effective in a fingerprint collation apparatus that uses a collation method that does not use feature points as identification information, such as a frequency analysis method or a cepstrum analysis method.

Next, returning to FIG. 2, the image information 52 obtained as described above is input to the calculation unit 5, and the calculation of the identification information such as the feature amount and the frequency distribution to be used for the collation is publicly known. Is performed (step SP5).

Next, the calculation unit 5 compares the authentication information such as the frequency distribution and the feature amount registered in advance stored in the identification information storage unit 7 with the identification information calculated in step SP5, Authenticate yourself (step SP
6).

Since the preprocessing such as the brightness correction and the smoothing correction as described above is performed, the quality of the arithmetic unit 5 is high, and the image information from the same fingerprint always includes the image information of the same area. Since the image information is cut out and the image information subjected to appropriate brightness adjustment is input, it is possible to dramatically improve the accuracy of the collation in the calculation unit 5 as compared with the conventional case.

Further, the above-described pre-processing is performed not only at the time of authentication but also at the time of registering the fingerprint image of the user, and the identification information is stored in the identification information storage unit 7, whereby the fingerprint collation is performed. It is possible to make the identification information, which is the reference used for personal authentication, accurate. In this way, at the time of registration and authentication, as described in this embodiment,
It is possible to obtain a highly accurate fingerprint authentication device with a high authentication rate by specifying the factor of image quality deterioration, performing brightness correction, and performing smoothing processing.

The present invention is not limited to specific numerical values such as the coefficient of the filter, the number of gradations, the luminance value, and the number in the ridge direction shown in the embodiments.

Further, in the embodiment of the present invention, the ridge direction of the fingerprint pattern is calculated and the arithmetic processing such as smoothing is performed based on the result, but the present invention is not limited to this. However, for example, even a configuration in which the direction other than the ridges of the fingerprint pattern, that is, the direction of the valley formed by the valleys is calculated, and each arithmetic processing is performed based on the result is included in the fingerprint collation device of the present invention. Needless to say.

[0102]

As described above, in the fingerprint collation device of the present invention, an image judged to be a good image by the image quality judgment unit is registered in the identification information storage unit by using the fingerprint image that has been brightness-corrected or smoothed. Since it is possible to perform verification and verification, it is possible to obtain a fingerprint verification device having a high personal authentication rate. In addition, the ridge direction for each small area that has been calculated once is stored in the storage unit, and using this, calculation of the center point of the fingerprint area,
Since the smoothing correction is performed, the total calculation amount can be reduced.

[Brief description of drawings]

FIG. 1 is a block diagram showing a configuration of a fingerprint collation device according to an embodiment of the present invention.

FIG. 2 is a fingerprint collation device according to an embodiment of the present invention,
Flowchart showing the flow of overall processing

FIG. 3 is a diagram showing an example of an acquired fingerprint image in the embodiment of the present invention.

FIG. 4 is a flowchart showing a processing flow in an image quality determination unit according to the embodiment of the present invention.

FIG. 5 is a conceptual diagram illustrating a small area and a fingerprint area according to the embodiment of the present invention.

FIG. 6 is a diagram showing an example of a direction detection filter according to the embodiment of the present invention.

FIG. 7 is a conceptual diagram illustrating a ridge direction calculation step in the embodiment of the present invention.

FIG. 8 is a diagram for explaining a part of the processing flow according to the embodiment of the present invention.

FIG. 9 is a flowchart showing another processing flow in the image quality determination unit according to the embodiment of the present invention.

FIG. 10 is a conceptual diagram illustrating a binarized image according to the embodiment of the present invention.

FIG. 11 is a conceptual diagram illustrating a step of calculating a center point of a fingerprint image according to the embodiment of the present invention.

FIG. 12 is a conceptual diagram illustrating clipped image information according to the embodiment of the present invention.

FIG. 13 is a conceptual diagram for explaining division of a middle area during brightness correction according to the embodiment of the present invention.

FIG. 14 is a conceptual diagram illustrating brightness correction according to the embodiment of the present invention.

FIG. 15 is a conceptual diagram showing an example of brightness correction of image information in the embodiment of the present invention.

FIG. 16 is a diagram showing an example of a smoothing direction filter according to the embodiment of the present invention.

FIG. 17 is a conceptual diagram showing an example of smoothing correction of image information in the embodiment of the present invention.

[Explanation of symbols]

1 Image input section 2 Image quality judgment unit 3 Image cutout part 4 Image compensation section 5 computing section 6 Display 7 Identification information storage 8 memory 10 Fingerprint matching device 21-28, 31-39 Filter 41,42 fingerprint image 43 small areas 44 Fingerprint area 45 direction undetectable area 46 Binary image 47,49,50,51,52 Image information 48 Middle area

Continued front page    (72) Inventor Chihiro Ueki             3-1, Tsunashima-Higashi 4-chome, Kohoku-ku, Yokohama-shi, Kanagawa             Matsushita Communication Industry Co., Ltd. (72) Inventor Tomoyuki Tsutoribe             3-1, Tsunashima-Higashi 4-chome, Kohoku-ku, Yokohama-shi, Kanagawa             Matsushita Communication Industry Co., Ltd. (72) Inventor Misa Sano             3-1, Tsunashima-Higashi 4-chome, Kohoku-ku, Yokohama-shi, Kanagawa             Matsushita Communication Industry Co., Ltd. (72) Inventor Yoshiyuki Matsuyama             3-1, Tsunashima-Higashi 4-chome, Kohoku-ku, Yokohama-shi, Kanagawa             Matsushita Communication Industry Co., Ltd. (72) Inventor Chie Ide             3-1, Tsunashima-Higashi 4-chome, Kohoku-ku, Yokohama-shi, Kanagawa             Matsushita Communication Industry Co., Ltd. (72) Inventor Sachiko Shiina             3-1, Tsunashima-Higashi 4-chome, Kohoku-ku, Yokohama-shi, Kanagawa             Matsushita Communication Industry Co., Ltd. F-term (reference) 5B043 AA09 BA02 DA04 DA05 EA02                       EA03 EA04 EA06 EA07 EA08                       EA09 EA10 EA11 EA15 EA18                       FA03 GA02 HA02

Claims (17)

[Claims] 1. An image input unit into which a fingerprint image to be registered is input, the fingerprint image is divided into a plurality of small areas, and a small area containing valid information is detected as a fingerprint area. Along with calculating the ridge direction for each of the small areas constituting the image quality of the fingerprint image is determined, when it is determined that the image is a good image,
Based on the image quality determination unit that outputs the fingerprint image, the storage unit that stores the ridge direction calculated by the image quality determination unit for each of the small areas, and the ridge direction stored in the storage unit. An image output unit that calculates a center point of the fingerprint image output from the image quality determination unit and outputs image information obtained by cutting out a predetermined area around the center point, and an image output from the image cutout unit Brightness correction and smoothing correction based on the ridge direction are performed on the information, and the identification information to be used for matching is calculated from the image compensation unit that outputs the image compensation unit and the image information output from the image compensation unit. Then, the fingerprint collation device is provided with an arithmetic unit for outputting and an identification information storage unit for storing the identification information output from the arithmetic unit. 2. An image input unit into which a fingerprint image to be collated is input, the fingerprint image is divided into a plurality of small areas, a small area containing valid information is detected as a fingerprint area, and the fingerprint area is Along with calculating the ridge direction for each of the small areas constituting the image quality of the fingerprint image is determined, when it is determined that the image is a good image,
Based on the image quality determination unit that outputs the fingerprint image, the storage unit that stores the ridge direction calculated by the image quality determination unit for each of the small areas, and the ridge direction stored in the storage unit. An image output unit that calculates a center point of the fingerprint image output from the image quality determination unit and outputs image information obtained by cutting out a predetermined area around the center point, and an image output from the image cutout unit An image compensating unit that performs luminance correction and smoothing correction based on the ridge direction on the information and outputs the information, and an identification information storage unit that stores pre-registered identification information to be used for matching. A fingerprint collation device comprising: an arithmetic unit that calculates identification information from the image information output from the image compensation unit, compares the identification information with the identification information stored in the identification information storage unit, and performs collation. . 3. The image quality determination unit further detects a small area in the fingerprint area in which the direction of the ridge cannot be obtained, as a direction undetectable area. The fingerprint collation device according to claim 1 or claim 2. 4. The image quality determination unit, when the number of detected direction-undetectable regions occupies a predetermined ratio or more with respect to the number of small regions occupying the fingerprint region, the fingerprint image. 4. The fingerprint collation device according to claim 3, wherein is determined to be defective information. 5. The image quality determination unit further detects a ridge collapsed area in which a ridge is collapsed in the detected direction undetectable areas, and the number of the ridge collapsed areas is the number of the ridge collapsed areas. 4. The fingerprint collation apparatus according to claim 3, wherein the fingerprint image is determined to be defective information when the fingerprint image occupies a predetermined ratio or more with respect to the number of small areas occupying the fingerprint area. 6. The image quality determination unit further detects a ridge faint area in which a ridge is faint in the detected direction undetectable area, and the number of the ridge fading areas is the number of the ridge fading areas. 4. The fingerprint collation apparatus according to claim 3, wherein the fingerprint image is determined to be defective information when the fingerprint image occupies a predetermined ratio or more with respect to the number of small areas occupying the fingerprint area. 7. The image of the direction undetectable area is binarized, and the number of pixels with a value of 0 is set to the total number of pixels forming the small area for each small area forming the direction undetectable area. The fingerprint collation apparatus according to claim 5, wherein the small area is detected as the ridge-blurred area when the occupied ratio is equal to or larger than a predetermined ratio. 8. The image of the direction undetectable area is binarized, and the number of pixels having a value of 1 is set to the total number of pixels forming the small area for each small area forming the direction undetectable area. 7. The fingerprint collation apparatus according to claim 6, wherein the small area is detected as the ridge fading area when the occupied ratio is equal to or larger than a predetermined ratio. 9. A display unit, which is connected to the calculation unit and the image quality determination unit and displays a message to a user, when the image quality determination unit determines that the fingerprint image is defective information. The fingerprint collation device according to any one of claims 1 to 8, wherein a message prompting a user to re-input is displayed on the display unit. 10. The image compensating unit divides the image information into a plurality of regions, and performs the luminance correction based on a luminance histogram in each region. The fingerprint collation device according to any one of items. 11. The image compensating unit divides the image information into a plurality of areas, and performs brightness correction based on a brightness histogram in an area determined as a fingerprint area in each area. The fingerprint collation device according to any one of claims 1 to 9. 12. The image compensating unit divides the image information into a plurality of regions, and corrects the luminance based on a luminance histogram in the regions excluding the regions determined to be direction undetectable regions in each region. The fingerprint collation device according to claim 1, wherein the fingerprint collation device according to any one of claims 1 to 9 is performed. 13. The image compensating unit divides the image information into a plurality of areas, and performs brightness correction based on a brightness histogram in an area in which a ridge direction is detected in each area. The fingerprint collation device according to any one of claims 1 to 9. 14. The image compensating unit divides the image information into a plurality of areas, and performs a brightness correction from a brightness histogram in each area and a predetermined range of surrounding areas. 10. The fingerprint collation device according to any one of claims 1 to 9. 15. The image compensating unit further filters the luminance-corrected image information for each small region in a direction along a ridge direction stored in the storage unit, 2. The brightness value is smoothed.
The fingerprint collation device according to any one of claims 0 to 14. 16. The image compensating unit further filters the direction-undetectable region of the brightness-corrected image information in the same direction to smooth the brightness value. Fingerprint matching device. 17. A fingerprint image is divided into a plurality of small areas, a ridge direction is calculated for each small area, and the ridge direction is calculated for each small area in which the ridge direction cannot be calculated. A fingerprint collation device characterized by identifying a factor of image quality deterioration based on a binarized image obtained by binarizing an image of a small area that cannot be scanned.