JP2015088041A - Authentication device, authentication method, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve the authentication accuracy by suppressing the luminance variation of a finger vein image.SOLUTION: An infrared LED 4 emits infrared light to a finger. A camera 8 images a finger vein image based on a beam transmitted through the finger in a predetermined exposure time Te. An average luminance calculation unit 31 calculates an average luminance La1 of the finger vein image. An exposure time control unit 35 executes a control for varying the exposure time Te of the camera 8 in a range from a minimum time to a maximum time on the basis of the error of the average luminance La1 from a target luminance TLa of the finger vein image. An LED light quantity control unit 36 executes a control for varying the LED light quantity PI of the infrared LED in a range from zero light quantity to a maximum light quantity on the basis of the error.

Description

  The present invention relates to an authentication device, an authentication method, and a program.

  Conventionally, in the field of finger vein authentication, an individual is identified by collating it with feature point extraction data by the minutiae method using a near-infrared light image transmitted through the finger (hereinafter referred to as “finger vein image”). A feature point extraction method is known (for example, see Patent Document 1).

JP 2001-243465 A

However, in the conventional feature point extraction method including Patent Document 1, when rotation or misalignment occurs, the luminance of the finger vein image changes greatly and important feature points are lost, and authentication accuracy deteriorates. In addition, luminance changes in finger vein images due to external light such as sunlight scattered light and indirect reflected light, and finger vein image luminance changes due to changes in blood flow due to outside air temperature are also large. Omission occurs and authentication accuracy deteriorates.
As described above, since the luminance change of the finger vein image is a factor of deterioration of the authentication accuracy, a method for suppressing the luminance change of the finger vein image has been demanded, but no effective method is found. .

  The present invention has been made in view of such a situation, and an object thereof is to improve authentication accuracy by suppressing a luminance change of a finger vein image.

In order to achieve the above object, an authentication apparatus according to an aspect of the present invention includes:
A light source that irradiates the finger with infrared light;
An imaging means for imaging a finger vein image based on the transmitted light of the finger at a predetermined exposure time;
Luminance calculating means for calculating the luminance of the finger vein image;
Exposure for executing control for changing the exposure time of the imaging means in a range from a minimum time to a maximum time based on an error in the brightness calculated by the brightness calculation means with respect to a target brightness of the finger vein image. A time control means;
Based on the error, a light source light amount control means for executing control for changing the light amount of the light source in a range from 0 light amount to a maximum light amount;
It is characterized by providing.

  Here, the imaging unit may include a near infrared band pass filter.

The exposure time control means controls the exposure time while the light source light quantity control means is executing control to maintain the light source at the zero light quantity,
Even if the exposure time changes up to the maximum time by the control of the exposure time control means, the light source light quantity control means executes control to change the light quantity of the light source,
be able to.

Furthermore, the imaging means amplifies and outputs a signal of the finger vein image by a predetermined gain,
Gain control means for executing control for changing the gain of the imaging means based on the error,
be able to.

  An authentication method and program according to one aspect of the present invention are a method and program corresponding to the above-described authentication apparatus according to one aspect of the present invention.

  According to the present invention, it is possible to improve the authentication accuracy by suppressing the luminance change of the finger vein image.

It is a block diagram which shows the hardware constitutions of the finger vein authentication apparatus which concerns on one Embodiment of the authentication apparatus of this invention. It is a functional block diagram which functions when performing a finger vein authentication / registration process among the functional structures of the finger vein authentication apparatus of FIG. FIG. 3 is a detailed functional block diagram of an image luminance control unit that functions when performing a constant luminance control process among the functional blocks of FIG. 2. FIG. 3 is a detailed functional block diagram of an authentication unit and a registration unit that function when executing finger vein authentication / registration processing among the functional blocks of FIG. 2. 5 is a flowchart for explaining a flow of finger vein authentication / registration processing executed by a finger vein authentication device having the functional configuration of FIGS. 2 to 4; 6 is a flowchart for explaining a detailed flow of a constant luminance control process in the finger vein authentication / registration process of FIG. 5. 6 is a flowchart for explaining a detailed flow of finger vein authentication processing in the finger vein authentication / registration processing of FIG. 5. 6 is a flowchart for explaining the details of the registration process in the finger vein authentication / registration process of FIG. 5.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings.

  FIG. 1 is a block diagram showing a hardware configuration of a finger vein authentication device according to an embodiment of the authentication device of the present invention.

  The finger vein authentication device includes an infrared LED (Light Emitting Diode) 1, a 10-key LCD (Liquid Crystal Display) 2, an LED buzzer 3, a touch sensor 4, an infrared LED drive circuit 5, a CPU ( A central processing unit) 6, a memory 7, a camera 8, a video decoder 9, and an FPGA (Field-Programmable Gate Array) 10.

The four systems of infrared LEDs 1 are light sources that irradiate near-infrared light onto the finger of the user to be authenticated.
The 10-key LCD 2 includes a “10-key” for the user to input a numerical value such as a password, and an LCD (liquid crystal display) for presenting various information to the user by image display.
The LED / buzzer 3 includes an LED that presents various types of information in a lighting / extinction pattern and a buzzer that presents a sound output.
The touch sensor 4 detects that the user's finger is placed on the finger vein authentication device.
The infrared LED drive circuit 5 changes the light quantity of the four infrared LEDs 1 by executing current control on the four infrared LEDs 1. Hereinafter, the light quantity of the four infrared LEDs 1 is referred to as “LED light quantity”, and the parameter is represented using “PI”.
The CPU 6 controls the operation of the entire authentication device.
The memory 7 is configured by an SDRAM (Synchronous Dynamic Random Access Memory), a flash memory, or the like, and appropriately stores data (registered template described later) necessary for the CPU 6 to execute various processes.
The camera 8 captures a near-infrared light image transmitted through the finger of the user to be authenticated, that is, a finger vein image, and outputs an analog signal of the finger vein image. In the present embodiment, the finger vein image is captured by the camera 8 as a moving image.
The video decoder 9 performs A / D conversion processing on the analog signal of the finger vein image (moving image), and the digital signal of the finger vein image (moving image) obtained as a result (hereinafter referred to as “finger vein image data”). Output).
The FPGA 10 converts the finger vein image data into luminance data and transfers it to the CPU 6.

FIG. 2 is a functional block diagram of the functional configuration of such a finger vein authentication apparatus that functions when finger vein authentication / registration processing is executed.
The finger vein authentication / registration process is a series of processes until the user is authenticated based on the finger vein image or the user's finger vein image is registered.

  When the finger vein authentication / registration process is executed, in the CPU 11, the image acquisition unit 21, the authentication unit 22, the registration unit 23, the authentication result output control unit 24, and the image luminance control unit 25 function.

  The image acquisition unit 21 acquires the luminance data of the finger vein image transferred from the FPGA 10 in units of frames and supplies it to the authentication unit 22 and the image luminance control unit 25.

The authentication unit 22 generates data for authentication (hereinafter referred to as “template”) from the luminance data of the finger vein image. Then, the authentication unit 22 uses the generated template as an evaluation target (hereinafter referred to as an “evaluation target template”), and a template already registered in the memory 7 (hereinafter referred to as “registered template”). And the authentication process based on the correlation coefficient is executed.
Although details will be described later with reference to FIG. 4, in this embodiment, the registration templates of N users (N is an arbitrary integer value of 1 or more) are classified into any of a plurality of groups.
Here, the purpose of calculating the correlation coefficient is roughly divided into the purpose of registered user authentication and the purpose of new user registration (template registration of the finger of the new user).
For the purpose of user authentication, the result of the authentication process by the authentication unit 22 is supplied to the authentication result output control unit 24. On the other hand, in the case of the purpose of new user registration, the result of the authentication process by the authentication unit 22 is supplied to the registration unit 23.

Based on the result of the authentication process by the authentication unit 22, the registration unit 23 records, in the memory 7, a template to be evaluated by the authentication unit 22 as a registration template for a new user's finger, so that the new user is recorded. sign up.
That is, the registration unit 23 determines which group to be classified as an evaluation target template in the authentication unit 22 based on the result of the authentication process by the authentication unit 22, and sets it as a registration template belonging to the determined group. Record a new one. In addition, although mentioned later for details, an evaluation object template may be classified into a some group.

  The authentication result output control unit 24 outputs from the 10-key / LCD 2 and the LED / buzzer 3 as the authentication result based on the result of the authentication process by the authentication unit 22 as the authentication result.

The image brightness control unit 25 executes control to make the brightness of the finger vein image that is the basis of the template substantially constant. Here, the control for making the luminance substantially constant means control for making the luminance of the entire finger vein image substantially constant. In the present embodiment, the average value of luminance of each pixel (hereinafter referred to as “average luminance”). ) Is kept constant.
Such control processing by the image luminance control unit 25 is hereinafter referred to as “luminance constant control processing”. As will be described later, the luminance constant control process is executed as a part of the finger vein authentication / registration process in the present embodiment.

  FIG. 3 is a detailed functional block diagram of the image luminance control unit 25 that functions when the luminance constant control process is executed among the functional blocks of FIG.

First, the purpose of the constant brightness control process will be described.
Conventionally, automatic gain adjustment and fixed gain have been used in cameras that capture finger vein images. For this reason, the variation in luminance of the finger vein image (template) increases, and as a result, the authentication accuracy decreases.
In addition, when there is strong external light, it is practically impossible to authenticate beyond the range of automatic camera adjustment.
In addition, there are fingers with low near-infrared light transmittance due to human characteristics, but with such fingers, the gain of the camera increases, resulting in more finger vein image noise, resulting in lower authentication accuracy. To do.
Therefore, in this embodiment, by executing the constant luminance control process, the average luminance of the finger vein image luminance data is made substantially constant, so that it is possible to generate a low-noise template using the dynamic range of the camera 8 sufficiently. The authentication accuracy has been improved compared to the past.

  Here, the camera 8 includes an optical lens 41, a near-infrared bandpass filter 42, an image sensor 43, and an AFE (Analog Front End) 44, as shown in detail in FIG.

  The optical lens 41 is composed of, for example, a focus lens that forms a subject image on the light receiving surface of the image sensor 43.

  The near-infrared bandpass filter 42 passes the near-infrared wavelength band of the light emitted from the optical lens 41 and cuts other wavelength bands. Thereby, even if strong external light is present, an authenticable finger vein image is captured by the image sensor 43.

  The image sensor 43 is configured by, for example, a CMOS (Complementary Metal Oxide Semiconductor) type photoelectric conversion element or the like. The image sensor 43 captures a finger vein image (frame) at regular intervals according to a predetermined clock pulse, and sequentially outputs the finger vein image (frame) as an analog signal.

In this way, a moving image of the finger vein image is configured by a plurality of frames that are continuously output in time from the image sensor 43 as analog signals. In the imaging of one frame, the time during which near-infrared light transmitted through the finger is received by the image sensor 43 is the exposure time, and can be changed by the image sensor 43. That is, changing the exposure time also changes the average brightness of the finger vein image.
Therefore, in the present embodiment, the exposure time is adopted as one of the control objects in the brightness constant control process. Hereinafter, this exposure time parameter will be expressed using “Te”.

The AFE 44 includes, for example, a CDS (Correlated Double Sampling) circuit, an amplifier circuit including an analog amplifier, and the like.
The amplification circuit of the AFE 44 performs gain adjustment on the analog signal of the finger vein image (frame) supplied from the image sensor 43. That is, the level of the analog signal is amplified by a predetermined gain.

Here, the level of the analog signal of each pixel constituting the finger vein image (frame) corresponds to the luminance of each pixel. When the gain changes, the level of the analog signal of each pixel changes, and as a result, the average luminance of the finger vein image also changes.
Therefore, in this embodiment, the gain is adopted as one of the objects to be controlled in the luminance constant control process. Hereinafter, this gain parameter will be expressed using “G”.

As described above, in the constant brightness control process of the present embodiment, the exposure time Te and the gain G in the camera 8 are adopted as control targets.
Further, the average luminance of the finger vein image changes as the light quantity of the infrared LED 4 as the light source changes. Therefore, in the present embodiment, the light amount of the infrared LED 4 is adopted as one of the control targets in the brightness constant control process. Hereinafter, the light amount of the infrared LED 4 is referred to as “LED light amount”, and this parameter is expressed using “PI”.
Here, in the brightness constant control process of the present embodiment, the lower limit of the LED light amount PI is “0”, that is, control that can be authenticated even with only the external light with the infrared LED 4 turned off is realized.

  As described above, when the constant brightness control process in which the exposure time Te, the gain G, and the LED light amount PI are controlled is executed, the image brightness control unit 25 of the CPU 6 includes the average brightness calculation unit 31 and the target. The brightness setting unit 32, the error calculation unit 33, the control target switching unit 34, the exposure time control unit 35, the LED light amount control unit 36, and the gain control unit 37 function.

  The average luminance calculation unit 31 calculates the average luminance of the luminance data of the finger vein image acquired from the image acquisition unit 21. Hereinafter, the parameter of the average luminance is expressed using “La1”.

The target brightness setting unit 32 sets the target value of average brightness as the target brightness. Hereinafter, the target luminance parameter is expressed using “TLa”.
The target luminance TLa can be set to an arbitrary value by a designer or the like, but it is necessary to set the target luminance TLa within a range in which whiteout and blackout do not occur in the finger vein image. Therefore, in this embodiment, it is assumed that the bit depth in the camera 8 is 8 bits, and 128 is set as the target luminance TLa.

The error calculating unit 33 calculates an error (TLa−La1) of the current average luminance La1 as a feedback value with respect to the target luminance TLa.
In the present embodiment, any one of the exposure time Te, the gain G, and the LED light amount PI is changed so that the error (TLa−La1) is almost eliminated (for example, 3% or less). Control is executed. That is, in this embodiment, the control target is determined from the exposure time Te, the gain G, and the LED light amount PI, and is switched as appropriate.
Specifically, the control target determining unit 38 determines the control target based on a predetermined algorithm (described later with reference to FIG. 6 and the like).
The control target switching unit 34 receives an error (TLa-La1), and outputs a functional block that controls the control target determined by the control target determination unit 38 (exposure time control unit 35, LED light amount control described later). Switching to the unit 36 or the gain control unit 37).

The exposure time control unit 35 controls the exposure time Te to change based on the error (TLa−La1).
The LED light amount control unit 36 performs control so as to change the LED light amount PI based on the error (TLa−La1).
The gain control unit 37 performs control so as to change the gain G based on the error (TLa−La1).

  The detailed functional configuration of the image luminance control unit 25 that functions when executing the constant luminance control processing in the functional blocks of FIG. 2 has been described above with reference to FIG. A specific example of the luminance constant control process will be described later with reference to FIG.

Next, detailed functional configurations of the authentication unit 22 and the registration unit 23 that perform finger vein authentication / registration processing using a finger vein image template in which the average luminance is substantially constant by such luminance constant control processing. Will be described.
FIG. 4 is a detailed functional block diagram of the authentication unit 22 and the registration unit 23 that function when executing finger vein authentication / registration processing among the functional blocks of FIG.

  When the finger vein authentication / registration process is executed, the template generation unit 51, the score calculation unit 52, and the registration template selection unit 53 function in the authentication unit 22 of the CPU 11.

The template generation unit 51 generates a template based on the luminance data of the finger vein image acquired by the image acquisition unit 21 (average luminance is made substantially constant by the luminance constant control process).
The form of the template is not particularly limited as long as it can be authenticated. For example, a binarized map may be used. However, in the present embodiment, a four-valued map is adopted.

That is, in this embodiment, each pixel value (luminance value) constituting the luminance data (one frame) of the finger vein image takes a value within the range of 0 to 255.
First, the template generation unit 51 generates image data (hereinafter referred to as “total luminance data of finger vein images”) obtained by summing up pixel values (luminance values) for eight frames of luminance data of finger vein images. To do. That is, each pixel value (total value of luminance values for 8 frames) constituting the total luminance data of the finger vein image takes a value in the range of 0 to 2047.
The template generation unit 51 converts each pixel value constituting the total luminance data of the finger vein image into four values so that the ratio of “0”, “1”, “2”, “3” becomes a predetermined value. Thus, a template that is a quaternary map is generated.
For example, among pixel values (total values of luminance values for 8 frames) constituting total luminance data of a finger vein image, pixel values to be noted as processing targets (hereinafter referred to as “target pixel values”) are as follows: "0" if less than the first threshold, "1" if greater than the first threshold and less than the second threshold, "2" if greater than or equal to the second threshold and less than the third threshold, and "0" if greater than or equal to the third threshold. 3 ". With respect to the total luminance data of the finger vein image, each pixel is sequentially set as a pixel of interest, and the above process is repeated to generate a template that is a quaternary map.
Here, the first threshold value to the third threshold value are set so that, for example, the ratio of “0” is 80%, the ratio of “1” is 5%, and the ratio of “2” is 5%.

  The template may adopt a quaternary map generated from the luminance data of the finger vein image acquired by the image acquisition unit 21. In this embodiment, the patent of the present applicant's patent 4996394 is used. A quaternary map generated by the method disclosed in the publication is adopted.

  That is, the template generation unit 51 generates a template that is a quaternary map by executing the following series of processes. For further details, reference may be made to the above-mentioned patent publication of Japanese Patent No. 4996394.

  First, the template generation unit 51 generates total luminance data of finger vein images after performing camera noise removal and centering. Centering means matching the centers of the frames.

  Next, the template generation unit 51 separates the total luminance data of the finger vein image into each data of the background image and the vein image.

Specifically, the template generation unit 51 performs low-pass filtering on the total luminance data of the finger vein image, thereby separating the background image data (low frequency luminance component signal) and the total luminance of the finger vein image. The data of the vein image (medium frequency luminance component signal) is separated by applying a band pass filter to the data.
In the present embodiment, the background image data (low-frequency luminance component signal) is separated (generated) using a low-pass filter to which a so-called moving average method is applied.
In addition, vein image data (medium frequency luminance component signal) is obtained by moving average method after removing low frequency luminance component using a matrix in which the weighting of each pixel decreases by 2/3 geometric series as it goes away from the center. A mid-pass filter (band-pass filter) that removes the band is used and separated (generated).

  The template generation unit 51 identifies and removes the edge region of the vein image data (medium frequency luminance component signal) and performs correction using the background image data, specifically, each pixel value of the vein image as the background. Correction is performed by dividing by each corresponding pixel value of the image.

  The template generation unit 51 identifies and removes a separated vein portion with low reproducibility for the corrected vein image data. The separated vein portion means a vein portion that is not shown as a single continuous vein but is shown as being separated into small pieces. The reproducibility of recognition is improved by removing the separated vein portion.

The method of removing the separation vein is not particularly limited. For example, in the present embodiment, a method of executing a series of processes according to the following (1) to (5) is employed.
(1) The template generation unit 51 sets the pixel value f (x, y) at the coordinates (x, y) in the corrected vein image data to a value corresponding to the value “1” after quaternarization ( In the present embodiment, the line segment length is set to “1” if the value is equal to or greater than the first threshold and hereinafter referred to as “luminance 1”.
(2) Furthermore, the template generation unit 51 searches for coordinates having “luminance 1” or more in the vicinity.
(3) The template generation unit 51 adds “1” to the line segment length and continues the search when coordinates of “luminance 1” or more exist in the vicinity.
(4) The template generation unit 51 ends the search when there are no coordinates of “luminance 1” or higher in the vicinity.
(5) If the line segment length at the end of the search is less than “10”, the template generation unit 51 regards it as a separated vein and replaces all pixel values of the line segment with “0”.
By such processes (1) to (5), it is possible to remove the separation vein with less reproducibility.
Here, “periphery” is a range that can be arbitrarily set by a designer or the like, but in this embodiment, (x, y + 1), (x, y−1) with respect to coordinates (x, y). ), (X + 1, y), (x + 1, y + 1), and (x + 1, y-1).

The template generation unit 51 creates a histogram for the vein image data from which the separated vein portion has been removed in this way, and generates a template that is a quaternary map based on the histogram.
The template generated by the template generation unit 51 is supplied to the score calculation unit 52 as an evaluation target template.

The score calculation unit 52 calculates a correlation score (correlation coefficient) between the evaluation target template and a registration template selected by a registration template selection unit 53 described later.
Hereinafter, in order to clarify the distinction between the evaluation target template and the registered template when calculating the correlation score, the evaluation target template is described using “TempR” and the registered template is described using “Temp”. Shall.

ここで、式（１）における相関係数（Ｆ／Ｇ）が、相関スコアに該当する。また、評価画像が、評価対象テンプレートＴｅｍｐＲに該当するため、評価画像の輝度Ｉとは、評価対象テンプレートＴｅｍｐＲの各画素値（４値化後の値）に該当する。一方、登録画像が、登録テンプレートＴｅｍｐに該当するため、登録画像の輝度Ｔとは、登録テンプレートＴｅｍｐの各画素値（４値化後の値）に該当する。
なお、式（１）に従った相関スコア（相関係数）の算出手法のさらなる詳細については、上述の特許４９９６３９４号の特許掲載公報を参照するとよい。 The calculation method of the correlation score is not particularly limited, but in this embodiment, the method disclosed in the patent publication of Japanese Patent No. 4996394 of the present applicant is adopted.
That is, the score calculation unit 52 calculates the correlation score by calculating the following equation (1).

Here, the correlation coefficient (F / G) in Equation (1) corresponds to the correlation score. Further, since the evaluation image corresponds to the evaluation target template TempR, the luminance I of the evaluation image corresponds to each pixel value (value after quaternarization) of the evaluation target template TempR. On the other hand, since the registered image corresponds to the registered template Temp, the luminance T of the registered image corresponds to each pixel value (value after quaternarization) of the registered template Temp.
For further details of the method of calculating the correlation score (correlation coefficient) according to the equation (1), it is preferable to refer to the above-mentioned patent publication of Japanese Patent No. 4996394.

  The registered template selection unit 53 selects a target (target for obtaining a correlation score) that matches the evaluation target template TempR from among a plurality of registered templates registered in the memory 7, and scores the selected template as a registered template Temp. It supplies to the calculation part 52.

  The registered template selection unit 53 includes a group representative selection unit 61 and an in-group selection unit 62.

In the present embodiment, each of N (N is an integer value of 1 or more) registered templates is classified into M (M is an integer value of N or less) groups and registered in the memory 7.
Among one or more registration templates belonging to a predetermined group (hereinafter, the registration templates in the group are described using “TMPG”), a registration template that can be determined to well represent the characteristics of the predetermined group is It is registered in the memory 7 as a template representing a predetermined group (hereinafter referred to as “representative template” and expressed using “TMP”).
One registered template can belong to a plurality of groups, but does not become a plurality of representative templates TMP.

First, at the time of authentication in the present embodiment, a correlation score with each of the M representative templates TMP is obtained for the evaluation target template TempR, and matches with the evaluation target template TempR based on each correlation score (similarity degree). The representative template TMP is selected. In other words, the primary selection is performed.
That is, it is predicted that the registered template TMPG that matches the evaluation target template TempR belongs to the group to which the representative template TMP selected primarily belongs. Therefore, only the group to which the primary selected representative template TMP belongs becomes an evaluation target for the next secondary selection. Therefore, hereinafter, the group to which the representative template TMP selected primarily is referred to as an “evaluation target group”.
When this evaluation target group is selected (when primary selection is performed), the group representative selection unit 61 functions in the registered template selection unit 53. That is, the group representative selection unit 61 selects a representative template TMP that is a matching target with the evaluation target template TempR, and supplies it to the score calculation unit 52.
It should be noted that the method for selecting the representative template TMP is sufficient as long as M representative templates TMP are finally selected, and the selection order of the representative templates TMP is not particularly limited. The selection method employed in this embodiment will be described later with reference to FIG.

Next, at the time of authentication according to the present embodiment, a correlation score with each of P registered templates TMPG (P is an integer value equal to or less than N) belonging to the evaluation target group is obtained for the evaluation target template TempR. Based on each correlation score, a registration template (hereinafter referred to as “match registration template”) that matches the evaluation target template TempR (highly similar) is selected. In other words, a secondary selection is performed.
When this match registration template is selected (secondary selection is performed), the in-group selection unit 62 functions in the registration template selection unit 53. That is, the intra-group selection unit 62 selects a registered template TMPG that is a matching target with the evaluation target template from the evaluation target group and supplies the selected template TMPG to the score calculation unit 52.
Note that the registration template TMPG selection method is not particularly limited as long as P registration templates TMPG that finally belong to the evaluation target group are selected. The selection method employed in this embodiment will be described later with reference to FIG.

As described above, in the authentication according to the present embodiment, since the two-step selection is performed before the match template is selected, it is possible to shorten the entire authentication time as compared with the conventional case.
Specifically, conventionally, correlation scores with all N registered templates are calculated for one evaluation target template TempR. That is, the number of calculation of the conventional correlation score necessary for selecting the match template is N times the same as the registration number N. Therefore, the total authentication time is prolonged in proportion to the increase in the registration number N.
On the other hand, in the present embodiment, for one evaluation target template TempR, correlation scores with M representative templates TMP are calculated in the primary selection. As a result, when the evaluation target group is narrowed down to one evaluation target group, in the secondary selection, correlation scores with P registered templates TMPG in the evaluation target group are calculated for one evaluation target template TempR. Is done. In other words, the correlation score is calculated (M + P) times until a match template is selected in this case, which is smaller than the registered number N times. As a result, the entire authentication time can be shortened compared to the conventional case. The effect of shortening the overall authentication time becomes more remarkable as the number of registrations N increases. In other words, even if the number of registrations N increases, the overall authentication time can be overwhelmingly suppressed compared to the conventional case.

  As described above, the authentication result of the authentication unit 22 as described above is supplied to the authentication result output control unit 24 (FIG. 2) at the time of authentication, and is supplied to the registration unit 23 at the time of registration.

  When the finger vein authentication / registration process at the time of registration is executed, the score comparison unit 71, the group classification unit 72, and the group representative determination unit 73 function in the registration unit 23 of the CPU 11.

The score comparison unit 71 compares the comparison source score with the comparison partner score, and supplies the comparison result to the group classification unit 72 and the group representative determination unit 73.
As the comparison source score, among the correlation scores for the evaluation target template TempR calculated by the score calculation unit 52, the correlation score performed until the secondary selection is adopted. At the time of registration, the evaluation target template TempR becomes a candidate for a new registration template, and therefore, the comparison source score is hereinafter referred to as a “registration target score”. Specifically, the score SCR [k] stored as the authentication result in the process of step S51 in FIG.
As one type of comparison partner score, a “match score” is employed in the present embodiment. The “match score” refers to a correlation score threshold for determining that the evaluation target template TempR matches the existing registered template Temp.
As one type of comparison partner score, a “group affiliation score” is employed in the present embodiment. The “group affiliation score” refers to a correlation score threshold necessary for belonging to a group (being classified).

  The group classification unit 72 performs group classification of the evaluation target template TempR (registration template candidate) based on the comparison result by the score comparison unit 71, and records (registers) it in the memory 7 as the registration template TMPG belonging to the classified group. .

  Specifically, for example, when the maximum value of the registration target score is equal to or greater than the matching score, it means that the registration template TMPG that matches the evaluation target template TempR is already registered in the memory 7. Therefore, in this case, the group classification unit 72 executes a predetermined error process (hereinafter referred to as “multiple registration error process”) for indicating multiple registration.

Further, for example, when there is a registration target score that has a maximum registration target score less than the matching score and is equal to or higher than the group affiliation score, the group classification unit 72 selects the evaluation target template TempR (registration template candidate) as a group. The registration target score that is equal to or higher than the affiliation score is classified into the same group as the registered template TMPG.
Here, the registration template is allowed to belong to a plurality of groups. Therefore, when there are a plurality of registration target scores that are equal to or higher than the group affiliation score and they are calculated from a plurality of different groups, the group classification unit 72 selects the evaluation target template TempR (registration template candidate) as the plurality of groups. It classifies each.

Furthermore, when the evaluation target template TempR (registered template candidate) is classified into a predetermined group, there is a possibility that it becomes the representative template TMP of the predetermined group.
Therefore, in such a case, the score comparison unit 71 compares the registration target score with the highest score in the predetermined group.
“Highest score” is one of the comparison partner scores, and refers to the following correlation score. That is, in this embodiment, when the evaluation target template TempR (registration template candidate) is registered in a predetermined group, the registration target score is also registered in association with it. That is, each of the P registration templates TMPG belonging to the predetermined group is associated with the registration target score used at the time of registration. The highest value of these P registration target scores (generally, the registration target score associated with the representative template TMP) is the highest score.

  When the registration target score of the evaluation target template TempR (registered template candidate) exceeds the highest score in the predetermined group, the group representative determining unit 73 selects the evaluation target template TempR (registered template candidate) as the representative template. Record (register) in the memory 7 as TMP.

  When the registration target score of the evaluation target template (registration template candidate) is less than the group affiliation score, the group classification unit 72 generates a new group and sets the evaluation target template (registration template candidate) as the new group. Shall be classified as

Next, a finger vein authentication / registration process executed by the finger vein authentication apparatus having the functional configuration shown in FIGS. 2 to 4 will be described with reference to FIGS.
FIG. 5 is a flowchart for explaining the flow of finger vein authentication / registration processing executed by the finger vein authentication apparatus having the functional configuration of FIGS.

In step S1, the touch sensor 4 (FIG. 1) determines whether or not a finger has been detected.
When the finger is not detected by the touch sensor 4, it is determined as NO in Step S1, and the process returns to Step S1. That is, until the finger is detected by the touch sensor 4, the determination process in step S1 is repeated, and the finger vein authentication / registration process is in a standby state.
When the finger is detected by the touch sensor 4, it is determined as YES in Step S1, and the process proceeds to Step S2.

In step S <b> 2, the image luminance control unit 25 and the like (FIG. 2) of the CPU 11 executes a constant luminance control process so that the luminance of the finger vein image for the finger detected by the touch sensor 4 is constant.
Details of the constant brightness control processing will be described later with reference to FIG.

In step S3, the authentication unit 22 or the like (FIG. 2) generates a finger vein image template for the finger detected by the touch sensor 4 as an evaluation target template. The authentication unit 22 or the like extracts an evaluation target group (primary selection) by calculating a correlation score between the evaluation target template and the representative template. And authentication part 22 grade | etc., Calculates a correlation score between the said evaluation object template and the registration template which belongs to an evaluation object group (secondary selection).
Hereinafter, the process of step S3 executed by the authentication unit 22 or the like is referred to as “finger vein authentication process”. Details of the finger vein authentication process will be described later with reference to FIG.

  In step S4, the authentication unit 22 or the like determines whether it is registration or authentication.

In the case of authentication, it is determined as NO in step S4, the result of the finger film authentication process in step S3 is supplied from the authentication unit 22 to the authentication result output control unit 24 (FIG. 2), and the process proceeds to step S5. move on.
In step S <b> 5, the authentication result output control unit 24 causes the authentication result to be output from the 10-key LCD 2 or the LED buzzer 3.
As a result, the finger vein authentication / registration process ends.

On the other hand, in the case of registration, it is determined as YES in step S4, and the result of the finger film authentication process in step S3 is supplied from the authentication unit 22 to the registration unit 23 (FIG. 2). Proceed to S6.
In step S <b> 6, the registration unit 23 classifies the finger vein image template for the finger detected by the touch sensor 4, that is, the template treated as the evaluation target template in step S <b> 3, and registers it in the memory 7. Execute the process.
Hereinafter, the process of step S6 executed by the registration unit 23 is referred to as “registration process”. Details of the registration process will be described later with reference to FIG.

  Further, in the finger vein authentication / registration process of FIG. 5, the details of the constant brightness control process in step S2, the finger vein authentication process in step S3, and the registration process in step S6 will be described individually in that order. .

  First, the prerequisites and the like for the constant luminance control process will be described.

As described above, the purpose of executing the constant brightness control process is to perform the authentication process in step S3 and the like with a low noise template using the dynamic range of the camera 8 by aligning the average brightness of the finger vein images. As a result, the authentication rate is improved.
Furthermore, acquiring a finger vein image that can be authenticated even in a state of only external light without a light source such as the infrared LED 1 is one of the purposes of executing the constant luminance control processing.

  As described above, the target luminance TLa is set to 128 by the target luminance setting unit 32, but any value can be set within a range in which a finger vein image without overexposure or underexposure can be acquired. is there.

  The gain G of the camera 8 is constant at a set value (initial setting value of the camera 8 in step S11 in FIG. 6 described later) prepared in advance in the brightness constant control process of FIG. 6 described below for convenience of explanation. It is assumed that the gain control unit 37 controls the above. The set value is not particularly limited, but if the gain G is increased, noise in the finger vein image increases, so it is preferable to use a low value. As will be described later, it goes without saying that variable control of the gain G of the camera 8 can be easily realized.

Here, when the gain G of the camera 8 is constant, the average luminance La of the finger vein image can be approximated by the following equation (2).
La = Ke × Vs × (Pe + PI) × Te × Kg (2)
In Equation (2), Ke indicates a coefficient of the exposure time Te and the average luminance La. Vs indicates the near infrared light average transmittance of the finger. Pe indicates the near-infrared light amount of external light. Kg represents a coefficient of the gain G and the average luminance La of the camera 8.
In the luminance constant control process of FIG. 6 described below, the average luminance La is controlled to be constant using the exposure time Te and the LED light amount PI as variable parameters (control target) in Expression (2). Specifically, as described above, the luminance constant control in FIG. 6 is performed so that the error (TLa−La1) of the average luminance La1 as a feedback value with respect to the target luminance TLa is almost eliminated (for example, 3% or less). Processing is executed.
The range that the exposure time Te can take is set to be not less than the minimum exposure time (Min_Te) and not more than the maximum exposure time (Max_Te). Specifically, in the following example, 2 / (60 × 510) [s] is adopted as the minimum exposure time (Min_Te), and (1 / frame rate) [ s] is employed.

  FIG. 6 is a flowchart for explaining the detailed flow of the constant luminance control process of step S2 in the finger vein authentication / registration process of FIG.

In step S11, the image brightness control unit 25 of the CPU 6 in FIG.
As the initial setting, for example, 1 is set as the gain G of the camera 8, 60 [Hz] is set as the frame rate, and (1/60) [s] is set as the exposure time.

In step S <b> 12, the LED light amount control unit 36 sets 0 (PI = 0) as the LED light amount PI setting (hereinafter referred to as “IR-LED setting”).
That is, in the brightness constant control process, control is initially executed with the goal of keeping the average brightness of the finger vein image constant even in the case of only the external light without the light source of the infrared LED 4.

In step S13, the image acquisition unit 21 acquires luminance data of the finger vein image, and the average luminance calculation unit 31 calculates average luminance La1 (feedback value) from the luminance data.
Here, since the peripheral portion of the finger vein image is easily affected by leakage of external light, the average luminance calculation unit 31 calculates the average luminance La1 based on each pixel in the central portion of the finger vein image. .
Specifically, for example, among the angle of view of the finger vein image conforming to the VGA (Video Graphics Array) standard, the angle of view of the finger is (x1, y1) = (0, 140) to (x2, y2) = (640 , 340). In this case, the average luminance calculation unit 31 calculates the average luminance of each pixel included in the range of (x1, y1) = (80, 160) to (x2, y2) = (560, 300) as the average luminance La1. To do.

Here, at the end of the processing in step S13, the control target determining unit 38 determines that the exposure time Te is a control target, and the output destination of the control target switching unit 34 is switched to the exposure time control unit 35 side. Yes.
For this reason, the error (TLa−La1) is calculated by the error calculation unit 33 using the average luminance La1 acquired in the process of step S13, and the error (TLa−La1) is exposed via the control target switching unit 34. When supplied to the time control unit 35, the process proceeds to step S14.

In step S14, the exposure time control unit 35 calculates the exposure time Te according to the following equation (3).
Te = Te + (TLa-La1) / Ke (3)
According to Expression (3), when the average luminance La1 is lower than the target luminance TLa, that is, when the error (TLa−La1) is a positive value, the exposure time control unit 35 performs control so as to increase the exposure time Te. . Here, the exposure time Te cannot exceed the frame rate. Therefore, the frame rate is adjusted as appropriate according to the exposure time Te.
On the other hand, when the average luminance La1 is higher than the target luminance TLa, that is, when the error (TLa−La1) is a negative value, the exposure time control unit 35 performs control so as to shorten the exposure time Te.

In step S15, the exposure time control unit 35 determines whether or not the exposure time Te calculated in the process of step S14 is equal to or greater than the minimum exposure time (Min_Te) (Te ≧ Min_Te).
If the exposure time Te calculated in step S14 is equal to or longer than the minimum exposure time (Min_Te), it is determined as YES in step S15, and the process proceeds to step S16.
In step S16, the exposure time control unit 35 determines whether or not the exposure time Te calculated in step S14 is equal to or less than the maximum exposure time (Max_Te) (Te ≦ Max_Te).
If the exposure time Te calculated in the process of step S14 is less than or equal to the maximum exposure time (Max_Te), it is determined as YES in step S16, and the process proceeds to step S17.

In step S17, the image acquisition unit 21 acquires luminance data of the finger vein image, and the average luminance calculation unit 31 calculates average luminance La1 (feedback value) from the luminance data.
That is, in the process of step S17, the average luminance La1 that is a feedback value for the exposure time Te calculated in the process of step S14, that is, the exposure time Te after control is calculated.

  In step S18, the exposure time control unit 35 determines whether the average brightness La1 calculated in the process of step S17, that is, the average brightness La1 after the control of the exposure time Te substantially matches the target brightness TLa (La1≈TLa). Determine whether or not.

  For example, in this embodiment, when the average luminance La1 after the control of the exposure time Te is within ± 3% relative to the target luminance TLa, that is, when the error (TLa−La1) is within 3%, YES in step S18. It is determined that there is, and the luminance constant control process ends. That is, the finger vein authentication process (FIG. 5 etc.) of the subsequent step S3 is executed using the exposure time Te calculated in the process of the previous step S14. As a result, a template having a substantially constant average luminance is used and the finger vein authentication process (FIG. 5 and the like) in step S3 is executed, so that the authentication accuracy is improved as compared with the conventional case.

On the other hand, when the average luminance La1 after the control of the exposure time Te exceeds ± 3% with respect to the target luminance TLa, that is, when the error (TLa−La1) exceeds 3%, further exposure is performed. It is necessary to execute control for changing the time Te to suppress the error (TLa−La1) within 3%. Therefore, in such a case, it is determined as NO in Step S18, the process returns to Step S14, and the subsequent processes are repeated.
That is, until the error (TLa−La1) is suppressed to within 3%, the loop processing of steps S14 to S18 is repeatedly executed, and the control is performed in which the exposure time Te is sequentially changed in step S14 of each loop processing. Is done.

However, as described above, the range that the exposure time Te can take is set to the minimum exposure time (Min_Te) or more and the maximum exposure time (Max_Te) or less, and the loop processing of steps S14 to S18 is within this range. Repeated on condition that there is time Te.
That is, when the exposure time Te is out of this range during the loop processing of steps S14 to S18, the loop processing ends and the following series of processing is executed.

That is, for example, when the exposure time Te becomes less than the minimum exposure time (Min_Te), it is determined as NO in Step S15, and the process proceeds to Step S26.
In step S26, the image luminance control unit 25 executes predetermined error processing. As a result, the luminance constant control process ends.

  For example, when the exposure time Te exceeds the maximum exposure time (Max_Te), in order to keep the average luminance La1 constant (substantially coincide with the target luminance TLa), only the light from the infrared LED 4 is necessary in order to make the average luminance La1 constant. Therefore, it is necessary to control the LED light amount PI. Therefore, in such a case, it is determined as NO in Step S16, the process proceeds to Step S19, and the following series of processes is executed.

In step S <b> 19, the image luminance control unit 25 performs initial setting of the camera 8.
In step S20, the LED light amount control unit 36 sets the LED light amount PI = 50 as the IR-LED setting.
In step S21, the image acquisition unit 21 acquires luminance data of the finger vein image, and the average luminance calculation unit 31 calculates average luminance La1 (feedback value) from the luminance data.

Here, at the end of the process in step S21, the control target determining unit 38 determines that the LED light amount PI is a control target, and the output destination of the control target switching unit 34 is switched to the LED light amount control unit 36 side. Yes.
For this reason, the error (TLa-La1) is calculated by the error calculation unit 33 using the average luminance La1 acquired in the process of step S21, and the error (TLa-La1) is transmitted to the LED via the control target switching unit 34. When supplied to the light quantity control unit 36, the process proceeds to step S22.

In step S22, the LED light amount control unit 36 calculates the LED light amount PI according to the following equation (4).
PI = PI + (TLa-La1) / Ke (4)
According to Equation (4), when the average luminance La1 is lower than the target luminance TLa, that is, when the error (TLa−La1) is a positive value, the LED light amount control unit 36 performs control so as to increase the LED light amount PI. On the other hand, when the average luminance La1 is higher than the target luminance TLa, that is, when the error (TLa−La1) is a negative value, the LED light amount control unit 36 performs control so as to decrease the LED light amount PI.

In step S23, the LED light quantity control unit 36 determines whether or not the LED light quantity PI calculated in step S22 is equal to or less than the maximum light quantity (Max_PI) (Te ≦ Max_PI).
If the LED light amount PI calculated in step S22 is less than or equal to the maximum light amount (Max_PI), it is determined as YES in step S23, and the process proceeds to step S24.

In step S24, the image acquisition unit 21 acquires luminance data of the finger vein image, and the average luminance calculation unit 31 calculates average luminance La1 (feedback value) from the luminance data.
That is, in the process of step S24, the average luminance La1 as a feedback value for the LED light amount PI calculated in the process of step S22, that is, the LED light amount PI after control is calculated.

  In step S25, the LED light amount control unit 36 determines whether the average luminance La1 calculated in the process of step S24, that is, the average luminance La1 after the control of the LED light amount PI substantially matches the target luminance TLa (La1≈TLa). Determine whether or not.

  For example, in the present embodiment, when the average luminance La1 after the control of the LED light amount PI is within ± 3% of the target luminance TLa, that is, when the error (TLa−La1) is within 3%, YES is determined in step S25. It is determined that there is, and the luminance constant control process ends. That is, the LED light amount PI calculated in the immediately preceding step S22 is used, and the finger vein authentication process (FIG. 5, etc.) in the subsequent step S3 is executed. As a result, a template having a substantially constant average luminance is used and the finger vein authentication process (FIG. 5 and the like) in step S3 is executed, so that the authentication accuracy is improved as compared with the conventional case.

On the other hand, when the average luminance La1 after the control of the LED light amount PI exceeds ± 3% with respect to the target luminance TLa, that is, when the error (TLa−La1) exceeds 3%, the LED further It is necessary to suppress the error (TLa−La1) within 3% by executing control for changing the light quantity PI. Therefore, in such a case, it is determined as NO in Step S25, the process is returned to Step S22, and the subsequent processes are repeated.
That is, until the error (TLa−La1) is suppressed to 3% or less, the loop processing of Steps S22 to S25 is repeatedly executed, and the control in which the LED light amount PI is sequentially changed in Step S22 of each loop processing is executed. Is done.

However, the range that the LED light quantity PI can take is set to be equal to or less than the maximum light quantity (Max_PI) determined according to the number and properties of the infrared LEDs 4, and the loop processing of steps S22 to S25 is within this range. It is repeated on condition that there is LED light quantity PI.
That is, when the LED light amount PI falls outside this range during the loop processing of steps S22 to S25, the loop processing ends, and the following series of processing is executed.

That is, when the LED light amount PI exceeds the maximum light amount (Max_PI), it is determined as NO in Step S23, and the process proceeds to Step S26.
In step S26, the image luminance control unit 25 executes predetermined error processing. As a result, the luminance constant control process ends.

In the finger vein authentication / registration process of FIG. 5, when the brightness constant control process of step S2 of FIG. 6 is executed, the finger vein authentication process of step S3 shown in FIG. 7 is executed.
FIG. 7 is a flowchart for explaining the detailed flow of the finger vein authentication process in step S3 in the finger vein authentication / registration process of FIG.

  In step S41, the image acquisition unit 21 of the CPU 6 in FIG. 4 captures the luminance data of the finger vein image (the average luminance is fixed by the luminance constant control process in FIG. 6).

  In step S42, the template generation unit 51 creates an evaluation target template TempR based on the luminance data of the finger vein image captured in step S41.

In step S43, the group representative selection unit 61 selects one of M group representative templates TMP as a registration template Temp for which a correlation score is to be calculated.
Here, each of the M groups is assigned a number i (i is an integer value of 0 to M−1 or less), and the representative template TMP of the group of the number i is referred to as “TMP [i]”. It shall be written. In the process of step S43, i = 0 is initially set, and the representative template TMP [0] is selected as the registration template Temp to be calculated.
Note that the number j = 0 is also initialized, but the number j will be described later in step S51.

  In step S44, the score calculation unit 52 calculates the correlation score for the evaluation target template TempR created in step S42 and the representative template TMP [i] selected as the calculation target registration template Temp according to the above-described equation (1). score1 is calculated.

In step S45, the score calculation unit 52 determines whether the correlation score score1 is less than the threshold value thresh1 (score1 <thresh1).
The threshold value thresh1 is used for selecting an evaluation target group, and can be arbitrarily set by a designer or the like. However, the number of registration templates N and the registration templates TMPG belonging to each of the M groups are registered. Is preferably set based on the number P (particularly the maximum number).

  That is, the case where the correlation score score1 is equal to or greater than the threshold value thresh1 means that the group to which the representative template TMP [i] for which the correlation score score1 is calculated belongs is primarily selected as the evaluation target group. In this case, it is determined as NO in Step S45, and a series of processes of Steps S48 to S53 corresponding to the secondary selection is executed. A series of these processes will be described later.

  On the other hand, when the correlation score score1 is less than the threshold value thresh1, the group to which the representative template TMP [i] for which the correlation score score1 is calculated is not an evaluation target group (it is not necessary to perform secondary selection). It means that it was judged. In this case, it is determined as YES in Step S45, and the process proceeds to Step S46.

In step S <b> 46, the group representative selecting unit 61 newly selects another one of the M group representative templates TMP as the registration template Temp for which the correlation score is to be calculated.
It should be noted that the selection method of the representative template TMP in step S46 suffices if an unselected one is selected, but in this embodiment, selection is performed in the order of the number i assigned to the group, that is, Temp = TMP [i ++. ].

In step S47, the group representative selection unit 61 determines whether or not the calculation of the correlation score score1 has been completed for all of the M group representative templates TMP.
If there is a M template representative template TMP for which the correlation score score1 has not yet been calculated, the process returns to step S44, and the subsequent processes are repeated.
That is, by repeating the loop processing of steps S44 to S47, the correlation score score1 with the evaluation target template TempR created in step S42 is calculated for each of the M groups of representative templates TMP. Of the M groups, a group having a representative template TMP in which the correlation score score1 is equal to or greater than the threshold value thresh1 is determined as an evaluation target group. And about this evaluation object group, a series of processes of step S48 thru | or S53 equivalent to secondary selection are performed.
Thus, when the calculation of the correlation score score1 is completed for all of the M group representative templates TMP, it is determined as YES in step S47, and the finger vein authentication processing is ended.

  Next, processing in steps S48 to S53 corresponding to the secondary selection will be described.

In step S48, the in-group selection unit 62 selects one of the registration templates TMPG in the evaluation target group (the group determined to be NO in step S45) as the registration template Temp for the correlation score calculation target. To do.
Here, each of the P registration templates TMPG in the group is assigned a number k (k is an integer value of 0 to P−1 or less), and the registration template TMPG of the number k is designated as “TMPG [k ] ". In the processing of step S48, k = 0 is initially set, and the registration template TMPG [0] in the evaluation target group is selected as the calculation target registration template Temp.

  In step S49, the score calculation unit 52 calculates the correlation score score2 for the evaluation target template TempR created in step S42 and the calculation target registration template Temp (registered template TMPG [k]) according to the above-described equation (1). calculate.

In step S50, the score calculation unit 52 determines whether or not the correlation score score2 is greater than or equal to a threshold value thresh2 (score1 ≧ thresh1).
The threshold value thresh2 is used for selecting a match template (or a candidate thereof) and can be arbitrarily set by a designer or the like. However, the number of registered templates is N and M groups are included in each group. It is preferable to set based on the number P (particularly the maximum number) of registered templates TMPG to which it belongs.

That is, the case where the correlation score score2 is greater than or equal to the threshold value thresh2 means that the registered template TMPG [k] in the evaluation target group is secondarily selected as a match template (or candidate thereof). In this case, it is determined as YES in Step S50, and the process proceeds to Step S51.
In step S51, the score calculation unit 52 stores the score SCR [k] = score2 and the evaluation target group [j] = i as authentication results. The storage location is not particularly limited, and may be the memory 7 or a buffer (not shown). When the process of step S51 ends, the process proceeds to step S52.

  On the other hand, the case where the correlation score score2 is less than the threshold value thresh2 means that it is determined that the registered template TMPG [k] in the evaluation target group is not a match template (or candidate thereof). In this case, it is determined as NO in Step S50, and the process proceeds to Step S52.

In step S52, the group representative selection unit 61 newly selects another one from the registration template TMPG for the evaluation target group as the registration template Temp for the correlation score calculation target.
The registration template TMPG selection method in step S52 only needs to be selected in the evaluation target group, but in this embodiment, the registration templates TMPG are selected in the order of the number k assigned to each registration template TMPG. Suppose that Temp = TMPG [k ++].

In step S53, the group representative selecting unit 61 determines whether or not the calculation of the correlation score score2 has been completed for all the registered templates TMPG in the evaluation target group.
If there is a registered template TMPG for which the correlation score score2 has not yet been calculated in the evaluation target group, the process returns to step S49, and the subsequent processes are repeated.
That is, by repeating the loop processing of steps S49 to S53, the correlation score score2 with the evaluation target template TempR created in step S42 is calculated for each registered template TMPG in the evaluation target group. Among the evaluation target groups, a registered template TMPG [k] whose correlation score score2 is equal to or greater than the threshold value thresh2 is determined as a match template (or candidate thereof), and the score SCR [k] = score2 is evaluated as an authentication result. The target group GRP [j] = i is stored.
Thus, when the calculation of the correlation score score2 is completed for all the registered templates TMPG in the evaluation target group, it is determined as YES in Step S53, and the process proceeds to Step S46, and the next representative template TMP is determined. The calculation target is selected, and the above-described series of processing is repeated.

When the finger vein authentication process of FIG. 7 is completed, that is, when the process of step S3 of FIG. 5 is completed, in the case of authentication, it is determined NO in step S4, and in step S5, the process of FIG. While the processing result of step S51 is output as the authentication result, in the case of registration, it is determined as YES in step S4, and the registration processing of step S6 shown in FIG. 8 is executed.
FIG. 8 is a flowchart for explaining the details of the registration process of step S6 in the finger vein authentication / registration process of FIG.

  In step S61, the score comparison unit 71 of the CPU 6 in FIG. 4 acquires the score SCR [k] as the registration target score from the authentication result (the processing result in step S51 in FIG. 7).

  In step S62, the score comparison unit 71 determines whether the registration target score is equal to or higher than the matching score.

When the maximum value among the one or more registration target scores is equal to or higher than the matching score, as described above, the evaluation target template TempR (created by the process of step S42 in FIG. 7) in which the registration target score is calculated; This means that the registration template TMPG of the same finger has already been registered in the memory 7. Therefore, in such a case, it is determined as YES in Step S62, and the process proceeds to Step S63.
In step S63, the group classification unit 72 executes a multiple registration error.
As a result, the registration process ends. That is, the process of step S6 in FIG. 5 ends, and the finger vein authentication / registration process ends.

On the other hand, when the registration target score is less than the coincidence score, it is determined as NO in Step S62, and the process proceeds to Step S64.
In step S64, the score comparison unit 71 determines whether or not the registration target score is equal to or greater than the group affiliation score.

If the registration target score is greater than or equal to the group affiliation score, it is determined as YES in step S64, and the process proceeds to step S65.
In step S65, the group classification unit 72 registers the evaluation target template TempR (registration template candidate) created in step S42 of FIG. 7 in the corresponding group.
As described above, the corresponding group refers to the same group as each of the one or more registration templates TMPG for which the registration target score that is equal to or higher than the group affiliation score is calculated. That is, when there are a plurality of registration target scores, a plurality of groups corresponding to each of the plurality of registration target scores is a corresponding group.

In step S67, the score comparison unit 71 determines whether or not the registration target score is equal to or higher than the highest score of the corresponding group.
When the registration target score is equal to or higher than the highest score of the corresponding group (when there are a plurality of corresponding groups, the score is equal to or higher than the highest score of at least one group), YES is determined in step S67, and the process proceeds to step S68. move on.
In step S68, the group representative determining unit 73 registers the evaluation target template TempR (registration template candidate) as a group representative of the corresponding group, that is, as a representative template TMP. In addition, when there are a plurality of corresponding groups, as described above, only one group representative template TMP can be obtained. Therefore, the evaluation target template TempR (registered template candidate) is registered as the representative template TMP of the group with the highest highest score. Shall be. As a result, the registration process ends. That is, the process of step S6 in FIG. 5 ends, and the finger vein authentication / registration process ends.
If the registration target score is less than the highest score of the corresponding group (if there are a plurality of relevant groups, if it is less than the highest score of all of the plurality of groups), it is determined as NO in step S67, and the registration process ends. It becomes. That is, the process of step S6 in FIG. 5 ends, and the finger vein authentication / registration process ends.

The series of processes when the registration target score is equal to or higher than the group affiliation score, that is, the series of processes when it is determined as YES in Step S64 has been described.
Next, a series of processes when the registration target score is less than the group affiliation score, that is, a series of processes when it is determined NO in Step S64 will be described.

If it is determined NO in step S64, it means that there is no group to which the evaluation target template TempR (registered template candidate) should belong, and thus a new group is created. Therefore, the process proceeds to step S66, and the following process is executed.
In step S66, the group classification unit 72 registers the evaluation target template TempR (registration template candidate) in the new group.
In step S68, the group representative determination unit 73 registers the evaluation target template TempR (registration template candidate) as the group representative of the new group, that is, as the representative template TMP.
As a result, the registration process ends. That is, the process of step S6 in FIG. 5 ends, and the finger vein authentication / registration process ends.

As described above, by executing the registration process of the present embodiment, the evaluation target template TempR (registration template candidate) is correlated with the registration target score (mutual authentication correlation calculated in the finger vein authentication process in step S3 shown in FIG. 7). (Score) is classified into a group having a group score or higher.
That is, a template having a mutual authentication score equal to or higher than the group score, that is, a similar template belongs to the predetermined group as the registered template TMPG. In other words, similar templates are grouped so as to be classified into the same group according to the correlation score of mutual authentication of registered templates.
Further, a template having a high mutual authentication score in the group, that is, a template having a high correlation within the group is registered as a representative template TMP representing the group.
As a result, at the time of authentication, as described above, the number of times of authentication itself is smaller than that of the conventional N times (and the speed is correspondingly increased), but the authentication accuracy equivalent to that of the conventional case can be obtained.

  Although the embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and modifications, improvements, and the like within the scope that can achieve the object of the present invention are included in the present invention. is there.

  For example, in the above-described embodiment, four infrared LEDs 1 are employed as light sources, but the present invention is not particularly limited to this, and any light source capable of irradiating infrared light onto a finger can be employed.

Further, for example, in the brightness constant control process, the average brightness of the finger vein image is controlled to be constant in the above-described embodiment, but it is sufficient if the brightness of the finger vein image is controlled to be constant. It is not limited to this.
Here, the luminance of the finger vein image refers to a value based on the luminance value (pixel value) of each pixel of the finger vein image. That is, the value calculated using the luminance value (pixel value) of at least a part of each pixel of the finger vein image is the luminance of the finger vein image. The calculation method in this case is not particularly limited. That is, in the above-described embodiment, the calculation method of taking the average of the luminance values of the pixels near the center of the finger vein image is adopted, but this is only an example, and other than that, the calculation of each pixel near the center of the finger vein image is performed. Arbitrary methods such as a calculation method for taking the median of the luminance value and a calculation method for taking the root mean square of the luminance values of the respective pixels near the center of the finger vein image can be adopted.

For example, as the luminance constant control process, the process according to the flowchart of FIG. 6 is employed in the above-described embodiment, but the present invention is not particularly limited thereto. For example, instead of the predetermined error processing in step S26, the following processing may be executed.
That is, the predetermined error processing in step S26 is for abnormal termination due to insufficient luminance. Therefore, instead of the predetermined error processing, the gain control unit 37 in FIG. 3 may execute control for increasing the gain of the camera 8. Thereafter, for the control for substantially eliminating the error (TLa−La1) (for example, 3% or less), any one of the three types of the gain G and the exposure time Te of the camera 8 and the LED light amount PI, and any Two types or all three types can be controlled.

  In addition, for example, as a finger vein authentication process, a correlation score (correlation coefficient) based on Expression (1) is used in the above-described embodiment for comparing the evaluation target template TempR and the registered template Temp. However, the present invention is not limited thereto, and a correlation coefficient calculated using a pattern matching method based on an arbitrary normalized correlation method may be employed.

  Further, for example, as finger vein authentication processing, in the above-described embodiment, after selection of an evaluation target group by matching using a representative template TMP (primary selection), matching by using a registered template TMPG in the evaluation target group. Although match template selection (secondary selection) has been performed, that is, two-step authentication has been performed, the present invention is not particularly limited thereto, and the number of authentication steps may be three or more.

  For example, after the group is further classified into small groups, a representative template (hereinafter referred to as “small representative template”) is set for each small group, and the evaluation target group is selected by the primary selection, FIG. The authentication unit 22 performs matching with the small representative template in the same manner as the primary selection, and selects a matching small group (hereinafter referred to as “match small group”) (secondary selection). Thereafter, the authentication unit 22 selects a match template by performing matching with a registered template belonging to the small match group (third-order selection).

  Further, for example, it is possible to perform multi-stage selection by combining a primary map and a secondary map disclosed in the patent publication of Japanese Patent No. 4996394.

Here, the primary map and the secondary map will be briefly described.
A template (quaternary map) referred to in the present embodiment is adopted as a secondary map.
The primary map is a reduced quaternary map generated by convolving a plurality of cells of the secondary map.

The primary map is generated as follows, for example.
An image (map) composed of a vertical pixel number a and horizontal pixel number b (a row and b column matrix with each pixel as a component) is referred to as an “a × b image (map)”. For example, a 2 × 2 primary map is generated by convolving a 2 × 2 pixel group (cell group) into one from a 4 × 4 secondary map.

That is, if the pixel position (coordinates) in the Xth row and the Yth column in the image is represented by “(X, Y)”, (1, 1), (1, 2), (2 , 1), (2, 2), the pixel value (luminance value) of (1, 1) of the primary map is calculated based on the pixel value (luminance value) of the pixel group.
Similarly, based on the pixel values (luminance values) of the pixel groups (1, 3), (1, 4), (2, 3), (2, 4) of the secondary map, 1, 2) pixel values (luminance values) are calculated.
Based on the pixel values (luminance values) of the pixel groups (3, 1), (3, 2), (4, 1), and (4, 2) of the secondary map, ) Pixel value (luminance value) is calculated.
Based on the pixel values (luminance values) of the pixel groups (3, 3), (3,4), (4, 3), (4, 4) of the secondary map, ) Pixel value (luminance value) is calculated.

  Specifically, for example, the pixel value (luminance value) of the primary map can be calculated according to the procedure of the following steps Sa to Sc using each pixel value (luminance value) of the pixel group of the secondary map.

  In step Sa, the template generation unit 51 in FIG. 4 sums the median threshold values of the pixel values of the pixel group of the secondary map.

Here, the “threshold value” refers to the following value.
That is, the secondary map is a quaternary map (template) in the above-described embodiment, and is generated based on luminance data of finger vein images for 8 frames. Since each pixel value (luminance value) of the luminance data of the finger vein image takes a value in the range of 0 to 255, for example, the total value of the pixel values for 8 frames takes a value in the range of 0 to 2040. The total value of the pixel values for the eight frames is converted into one of four values “0”, “1”, “2”, and “3”. The threshold value (first to third threshold values described above) used when performing this conversion is the “threshold value” in step Sa.
The “threshold median” means (threshold upper limit value + threshold lower limit value) / 2.

Specifically, for example, a lower limit “0” and an upper limit “400” (the upper limit is the above-described first threshold) can be employed as the threshold value of the pixel value “0” of the secondary map. In this case, for a pixel having a pixel value of “0” in the pixel group of the secondary map, the median “200” of the lower limit “0” and the upper limit “400” is the median of the threshold values of the pixel values. .
Similarly, as the threshold value of the pixel value “1” of the secondary map, for example, a lower limit “400” (lower limit is the above-described first threshold value) and an upper limit “800” (upper limit is the above-described second threshold value) may be employed. it can. In this case, for the pixel having the pixel value “1” in the pixel group of the secondary map, the median “600” of the lower limit “400” and the upper limit “800” is the median of the threshold values of the pixel values. .
As the threshold value of the pixel value “2” of the secondary map, for example, a lower limit “800” (lower limit is the above-described second threshold value) and an upper limit “1200” (upper limit is the above-described third threshold value) can be employed. In this case, for the pixel having the pixel value “2” in the pixel group of the secondary map, the median “1000” of the lower limit “800” and the upper limit “1200” is the median of the threshold values of the pixel values. .
As the threshold value of the pixel value “3” of the secondary map, for example, a lower limit “1200” (the lower limit is the above-described third threshold value) and an upper limit “2047” can be employed. In this case, for the pixel having the pixel value of “3” in the pixel group of the secondary map, the approximate median value “1600” of the lower limit “1200” and the upper limit “2047” is the median of the threshold values of the pixel values. Become. The reason why the approximate median value is used instead of the median value itself is to simplify the calculation.

Thus, for example, in the secondary map, the pixel value of (1, 1) is “0”, the pixel value of (1, 2) is “0”, and the pixel value of (1, 3) is “0”. It is assumed that the pixel value of (1, 4) is “1”.
In this case, the median of the pixel values of (1, 1) is “200”, the median of the pixel values of (1, 2) is “200”, and the median of the pixel values of (2, 1) is “200”. 200 ”, and the median of the pixel values of (2, 2) is“ 600 ”.
Therefore, in step Sa, the total value of these medians, that is, (200 + 200 + 200 + 600) = 1200 is calculated.

In step Sb, the template generation unit 51 divides the sum of the median threshold values of the pixel values of the pixel groups of the secondary map (calculation result of step Sa) by the number of pixels to be convolved.
For example, in the secondary map, the sum of the median threshold values of the pixel values of the pixel groups (1, 1), (1, 2), (1, 3), (1, 4) (in step Sa) When “1200” in the above example is calculated as the calculation result), since the number of pixels to be convolved is “4”, “300 (= 1200/4)” is the calculation result of step Sb. As obtained.

  In step Sc, the template generation unit 51 converts the value obtained by division in step Sb into a quaternary value to obtain a pixel value of the primary map. As in the case of generating the secondary map, the ratio of the values after quaternarization is a predetermined ratio.

  Although the primary map generation method has been briefly described above, for further details, reference may be made to the patent publication of Japanese Patent No. 4996394 of the present applicant.

It is possible to perform authentication at higher speed by executing authentication processing by multi-stage selection using such a primary map (a reduced image of the secondary map) as appropriate.
For example, a primary map can be adopted as a representative template. Thereby, the time which selects an evaluation object group can be shortened more.
In addition, for example, in matching with the registration template TMPG in the evaluation target group after the primary selection, a method of performing secondary selection using the primary map and selecting only the secondary map corresponding to the selected primary map is tertiary. May be adopted. Thereby, compared with the above-mentioned embodiment which performs matching in the evaluation object group after primary selection only by a secondary map, time to select a match template can be shortened more. For details of this method, refer to the patent publication of Japanese Patent No. 4996394 of the present applicant.
For example, when the inside of a group is further classified into small groups, a primary map can be adopted as the small representative template described above. Thereby, the time for selecting the small match group can be further shortened.

Here, the series of processes described above can be executed by hardware or can be executed by software.
In other words, the functional configurations of FIGS. 2 to 4 are merely examples, and are not particularly limited. That is, it is sufficient that the authentication apparatus has a function capable of executing the above-described series of processes as a whole, and what kind of functional blocks are used to realize this function is limited to the examples of FIGS. Not.
In addition, one functional block may be constituted by hardware alone, software alone, or a combination thereof.

When a series of processing is executed by software, a program constituting the software is installed on a computer or the like from a network or a recording medium.
The computer may be a computer incorporated in dedicated hardware. The computer may be a computer capable of executing various functions by installing various programs, for example, a general-purpose personal computer.

  The recording medium including such a program is not only constituted by a removable medium (not shown) distributed separately from the apparatus main body in order to provide the user with the program, but is also provided to the user in a state of being pre-installed in the apparatus main body. It is composed of a provided recording medium or the like. The removable medium is composed of, for example, a magnetic disk (including a floppy disk), an optical disk, a magneto-optical disk, or the like. The optical disk is composed of, for example, a CD-ROM (Compact Disk-Read Only Memory), a DVD (Digital Versatile Disk), or the like. The magneto-optical disk is configured by an MD (Mini-Disk) or the like. Further, the recording medium provided to the user in a state of being incorporated in advance in the apparatus main body is configured by, for example, the memory 7 in FIG. 1 in which a program is recorded.

  In the present specification, the step of describing the program recorded on the recording medium is not limited to the processing performed in time series along the order, but is not necessarily performed in time series, either in parallel or individually. The process to be executed is also included.

  DESCRIPTION OF SYMBOLS 1 ... Infrared LED, 2 ... 10 key LCD, 3 ... LED buzzer, 4 ... Touch sensor, 5 ... Infrared LED drive circuit, 6 ... CPU, 7 ..Memory, 8 ... Camera, 9 ... Video decoder, 10 ... FPGA, 21 ... Image acquisition unit, 22 ... Authentication unit, 23 ... Registration unit, 24 ... Authentication Result output control unit 25 ... Image luminance control unit 31 ... Average luminance calculation unit 32 ... Target luminance setting unit 33 ... Error calculation unit 34 ... Control target switching unit 35 ... Exposure time control unit, 36 ... LED light quantity control unit, 37 ... Gain control unit, 41 ... Optical lens, 42 ... Near infrared bandpass filter, 43 ... Image sensor, 44... AFE, 51... Template generation unit, 52... Score calculation unit, 53. Template selecting section, 61 ... representative group selection unit, the selection unit 62 ... group, 71 ... score comparing section, 72 ... group classification unit, 73 ... representative group determination unit

Claims (6)

A light source that irradiates the finger with infrared light;
An imaging means for imaging a finger vein image based on the transmitted light of the finger at a predetermined exposure time;
Luminance calculating means for calculating the luminance of the finger vein image;
Exposure for executing control for changing the exposure time of the imaging means in a range from a minimum time to a maximum time based on an error in the brightness calculated by the brightness calculation means with respect to a target brightness of the finger vein image. A time control means;
Based on the error, a light source light amount control means for executing control for changing the light amount of the light source in a range from 0 light amount to a maximum light amount;
An authentication device comprising: The authentication apparatus according to claim 1, wherein the imaging unit includes a near-infrared bandpass filter. While the light source light quantity control means is executing control to maintain the light source at the zero light quantity, the exposure time control means controls the exposure time,
The light source light quantity control means executes control to change the light quantity of the light source when the error does not become a certain value even if the exposure time changes up to the maximum time by the control of the exposure time control means. 2. The authentication device according to 2. The imaging means amplifies and outputs a signal of the finger vein image by a predetermined gain,
Gain control means for executing control for changing the gain of the imaging means based on the error,
The authentication device according to claim 2 or 3. A light source that irradiates the finger with infrared light;
An imaging means for imaging a finger vein image based on the transmitted light of the finger at a predetermined exposure time;
An authentication method executed by an authentication device comprising:
A luminance calculating step for calculating the luminance of the finger vein image;
Control is performed to change the exposure time of the imaging means in a range from a minimum time to a maximum time based on an error in the luminance calculated by the processing of the luminance calculation step with respect to a target luminance of the finger vein image. An exposure time control step,
A light source light quantity control step for executing control to change the light quantity of the light source in a range from 0 light quantity to the maximum light quantity based on the error;
An authentication method that includes: A light source that irradiates the finger with infrared light;
An imaging means for imaging a finger vein image based on the transmitted light of the finger at a predetermined exposure time;
A computer for controlling an authentication device comprising:
A luminance calculating step for calculating the luminance of the finger vein image;
Control is performed to change the exposure time of the imaging means in a range from a minimum time to a maximum time based on an error in the luminance calculated by the processing of the luminance calculation step with respect to a target luminance of the finger vein image. An exposure time control step,
A light source light quantity control step for executing control to change the light quantity of the light source in a range from 0 light quantity to the maximum light quantity based on the error;
A program that executes control processing including

Images