JP5914995B2 - Biological identification device and biological identification method - Google Patents

Description

  The present invention relates to a biometric identification device and a biometric identification method.

  There is known a biometric authentication system that performs biometric authentication such as fingerprints and vein patterns to authenticate individuals. For example, a vein authentication device has been developed as a system for performing authentication based on a finger vein image obtained by imaging a finger.

  When performing personal authentication using a vein authentication device, it is required to increase the accuracy of authentication as much as possible. For this reason, it is important to obtain a finger vein image of optimum quality without being influenced by even when the external environment at the time of imaging is different. For example, there is a method in which a finger vein image is obtained by placing a finger in a guide groove installed on the authentication device, and imaging the finger in a fixed position, and authenticating the individual using the finger vein image. It has been proposed (for example, Patent Document 1).

JP 2006-155575 A

  According to the method of Patent Document 1, since a finger can be imaged at the same position and direction by providing a guide, a finger vein image of stable quality can be acquired and authentication can be performed with high accuracy. It becomes possible.

However, in the method of Patent Document 1, when performing imaging, the user must keep his finger along the guide. Further, only a specific finger (for example, index finger) is limited as an imaging target. On the other hand, considering the situation where the authentication device is actually used, if the finger determined at the time of authentication has to be placed in a predetermined position and direction, the convenience for the user is greatly impaired. Therefore, at the time of authentication, it is desirable to perform highly accurate authentication for an unspecified finger without fixing the position and direction. On the other hand, when authentication is performed without specifying a finger type (for example, index finger) and specifying a position and a direction, a plurality of fingers may be imaged. In such a case, if authentication is performed without considering the positional relationship between the fingers, a mismatch may occur with another finger that should not be matched, and authentication accuracy may be reduced.
Here, the authentication identifies whether or not the authentication (identification) target person is a registrant by collating the registered image and the like with the image obtained at the time of authentication, and based on the identification result, For example, controlling a control target such as an electronic lock. Therefore, since the accuracy of authentication depends on the accuracy of identification, it can be said that the above problem in authentication is a problem in identification.

  In the present invention, in a biometric identification device that identifies an individual using an image of a captured finger, even when a plurality of fingers are imaged, high-precision identification is performed regardless of the position and direction of the finger at the time of identification. The purpose is to do.

A main invention for achieving the above object includes an imaging unit that images a biological pattern of a finger, and a calculation unit, and the calculation unit images the biological pattern of a finger of an identification target person using the imaging unit. The feature amount characterizing the biological pattern of the identification target person from the image obtained as described above, and having the characteristics that are position-invariant and rotation-invariant with respect to the biological pattern even when the imaging range and the imaging direction change Whether or not the feature quantity characterizing the biometric pattern of the person to be registered and registered in association with the finger ID representing the kind of finger in the predetermined registration unit is similar to the feature quantity of the identification target person And calculating a score for each finger of the identification target person according to a finger ID associated with the feature quantity of the registration target person determined to be similar to the feature quantity of the identification target person , Identification Calculating a total value of scores of a plurality of adjacent fingers of the elephant, and identifying that the identification target person is the registration target person when a maximum value among the total values of the scores is greater than a predetermined threshold value The biometric identification device is characterized by the above.

  Other features of the present invention will become apparent from the description of the present specification and the accompanying drawings.

It is a block diagram of vein identification device 1 in this embodiment. 2A and 2B are diagrams illustrating an example in which the vein identification device 1 is used for door unlocking control. It is a figure showing the flow of registration operation | movement. It is a figure showing the flow of a feature-value extraction process. It is a figure showing an example of a feature point. It is a figure showing an example of the obtained brightness | luminance gradient. It is a figure showing an example of the histogram of a brightness | luminance gradient. It is a figure showing an example of the brightness | luminance gradient when aligning a coordinate axis with a reference direction. It is a figure explaining the operation | movement which extracts the feature-value from a vein pattern. It is a figure showing the flow of identification operation | movement. It is a figure showing the flow of collation processing. 12A to 12C are diagrams illustrating an example in which the identification operation is performed using the finger vein pattern of the person X to be identified. 13A to 13C are diagrams illustrating an example in which the identification operation is performed using the finger vein pattern of the person Y to be identified.

  At least the following matters will become clear from the description of the present specification and the accompanying drawings.

  An imaging unit that captures a biometric pattern of a finger, a registration unit that registers and registers a feature amount of a registration target person that characterizes a biometric pattern of a registration target person, and a finger ID that represents a finger type, and a calculation unit. The computing unit extracts a plurality of features of the identification target person from an image obtained by imaging a biological pattern of a finger of the identification target person using the imaging unit, and is registered in the registration unit It is determined whether the feature quantity of the registration target person is similar to the feature quantity of the identification target person, and is associated with the feature quantity of the registration target person determined to be similar to the feature quantity of the identification target person. Whether the identification target person is the registration target person based on a combination of scores of a plurality of adjacent fingers of the identification target person, according to the finger ID. A biometric identification device for identifying whether or not.

According to such a biometric identification device, the calculation unit extracts the registration target person's feature quantity from the registration target person's biometric pattern, and the extracted feature quantity is a finger representing the type of finger from which the feature quantity was extracted. The registered feature value is registered in the registration unit in association with the ID. In addition, the calculation unit extracts a plurality of feature amounts from an image obtained by imaging the biological pattern of the finger of the identification target person using the imaging unit, and compares it with the feature amount of the registration target person, thereby identifying the identification target person. Identify whether or not the user is registered. At that time, the identification target person's score is calculated for each finger based on the finger ID associated with the registered feature quantity similar to the identification target person's feature quantity, and the combination of scores of adjacent fingers is considered. The identification is performed by.
By considering the scores of multiple adjacent fingers, it is less likely that a mismatch will occur in which all of the multiple fingers are supposed to be similar features that are not supposed to match. The influence etc. can be made relatively small. That is, erroneous data due to mismatching is less likely to be reflected in the final identification target score. Therefore, in a biometric identification apparatus that identifies an individual using an image of a captured finger, even when a plurality of fingers are captured, highly accurate identification is performed regardless of the position and direction of the finger at the time of identification. be able to.

  In this biometric identification device, the imaging unit captures at least a part of a biometric pattern of one finger of the identification target person and at least a part of a biometric pattern of a finger adjacent to the one finger. It is desirable to do.

  According to such a biometric identification device, it is possible to register the feature amounts of two adjacent fingers as data to be identified, and the accuracy of identification can be increased. When the identification is performed using only the biometric pattern of one finger, the score of the finger may be accidentally calculated due to the influence of noise or the like, which may cause erroneous identification. However, the probability that the score is calculated by chance for adjacent fingers is very small. Therefore, by registering the biometric patterns of two adjacent fingers and using the data for identification, the influence of an accidental mismatch or the like can be relatively reduced. This makes it possible to identify with high accuracy.

  In such a biometric identification device, the calculation unit calculates a total value of scores of a plurality of adjacent fingers of the identification target person, and the maximum value among the total values of the scores is larger than a predetermined threshold value In addition, it is desirable to identify that the identification target person is the registration target person.

  According to such a biometric identification device, the identification accuracy can be further increased. Even if a score is calculated high for a certain finger due to an accidental mismatch or the like, the probability that mismatches occur simultaneously for other fingers is small. Therefore, by calculating the total value of the scores of a plurality of adjacent fingers, it becomes easy to obtain data containing a large amount of accurate data (finger score). And by adopting the maximum value of the total value as the score of the person to be identified, it is possible to calculate a score with a small influence of mismatch and high reliability. By comparing this score with an appropriately set threshold value, the accuracy of authentication can be further increased.

  In this biometric identification device, the calculation unit extracts a feature amount of the registration target person from an image obtained by imaging a biometric pattern of a finger of the registration target person using the imaging unit, and the registration target It is desirable that the feature amount of the person and the finger ID of the person to be registered be associated with each other and registered in the registration unit as the registered feature amount.

  According to such a biometric identification device, the operation of registering the feature value and the operation of identifying the person to be identified can be performed by one biometric identification device, so that convenience for the user can be enhanced. Then, when registering the feature amount, the feature amount extracted from the finger of the person to be registered is registered in association with the finger ID indicating the extracted finger, thereby identifying each finger during the identification operation. And the identification accuracy can be increased.

  In this biometric identification device, it is preferable that the biometric pattern of the finger of the registration target person has a larger area captured by the imaging unit than the biometric pattern of the finger of the identification target person.

  According to such a biometric identification device, in the operation of registering the biometric pattern of the person to be registered, the number of user feature amounts extracted from the image of the biometric pattern captured in a larger range is the ecology of the identification target person. There is a high possibility that it will be larger than the number of features of the identification target person extracted in the operation of identifying the pattern. That is, since the user's feature quantity that can be used for identification can be registered more, the base of the collation data becomes large. Therefore, the feature quantity extracted from the biometric pattern of the person to be identified can be easily compared with the registered feature quantity, so that more accurate identification can be performed.

  In such a biometric identification device, it is preferable that the feature amount is extracted using a SIFT (Scale Invariant Feature Transform).

  According to such a biometric identification device, it is possible to extract a feature quantity of a biometric pattern that is less dependent on position change, rotation, and the like at the time of imaging by extracting a feature quantity using SIFT. Therefore, when imaging a biometric pattern, the biometric pattern is characterized even if the positional relationship between the sensor that performs imaging and the imaging target (biological pattern of the person to be registered or the person to be identified) is not fixed. It is possible to accurately extract feature amounts, and high-precision identification can be performed regardless of imaging conditions.

  Further, a plurality of features of the identification target person characterizing the biological pattern of the finger of the identification target person are extracted from an image obtained by imaging the biological pattern of the finger of the identification target person, and a finger ID representing a finger type Determining whether the feature quantity of the registrant registered in association with the feature quantity of the identification target person is similar, and determining that the feature quantity of the identification target person is similar Calculating the score for each finger of the identification target person according to the finger ID associated with the feature amount of the identification person, and the identification target based on a combination of scores of a plurality of adjacent fingers of the identification target person A biometric identification method comprising: identifying whether or not a person is the registration subject.

=== Embodiment ===
As a form of a biometric identification device for carrying out the invention, a vein identification device 1 will be described as an example.
FIG. 1 is a block diagram of a vein identification device 1 according to this embodiment. The vein identification device 1 includes a calculation unit 10, a sensor unit 20, a light source unit 30, a trigger sensor 40, and a control target 50. The sensor unit 20 is connected to the calculation unit 10 via an interface 28, and the light source unit 30 is connected to the calculation unit 10 via an interface 38. The vein identification device 1 is connected to the trigger sensor 40 and the control target 50 via the interface 48 and the interface 58.

  The calculation unit 10 includes a CPU (Central Processing Unit) 12 that performs calculation, a RAM (Random Access Memory) 14 as a storage device (registration unit), and an EEPROM (Electronically Erasable and Programmable Read Only Memory) 16. The CPU 12 identifies the vein by executing a program stored in the EEPROM 16. In the RAM 14, a feature amount necessary for identifying a vein is registered as a calculation result. The calculation unit 10 can cause the vein identification device 1 to perform two operations, a registration operation and an identification operation described later.

  The sensor unit 20 is a sensor used for imaging a finger vein. The sensor unit 20 captures an image of the approaching finger, and the exposure time can be adjusted at that time. Although details will be described later, the sensor unit 20 only needs to be able to image a vein pattern of a part of the finger of the person to be identified during the identification operation. On the other hand, it is desirable to be able to image a vein pattern in a wide range of fingers as much as possible during the registration operation. Therefore, the vein identification device 1 may include two types of sensors for registration operation and identification operation.

  The light source unit 30 is a device for illuminating a finger to be imaged with light having a predetermined wavelength. Here, a near-infrared LED (Light Emitting Diode) light source is included, and light including a large wavelength band of 700 nm to 900 nm is irradiated. This wavelength band is also referred to as a “biological window”, and is a wavelength band in which the absorption of both blood hemoglobin and water is low, and the biological transmittance is high. In the present embodiment, imaging is performed while irradiating a finger with near-infrared rays in a wavelength band of 700 nm to 900 nm, so that a portion where blood is abundant, that is, a blood vessel portion is captured as a shadow. Therefore, it is possible to obtain an image that accurately represents the shape of the vein in the finger.

  The trigger sensor 40 is a device that detects the approach of a finger to be imaged and sends a trigger for starting an imaging process to the arithmetic unit 10. As the trigger sensor 40, for example, a capacitance sensor is used. With this trigger sensor 40, when a vein identification device is provided on a door or the like, which will be described later, the sensor unit 20 can automatically start imaging processing of the finger simply by bringing the finger close to the door.

  The control object 50 is an object to be controlled according to the identification result by the vein identification device 1. For example, when the control object 50 is a computer, the access right of the computer is given to the person to be identified according to the identification result by the vein identification device 1. Further, when the control target 50 is an electronic door lock as described later, the door electronic lock unlocking control is performed according to the identification result by the vein identification device 1. Hereinafter, the case where the control target 50 is an electronic lock of a door will be described.

  2A and 2B show an example in which the vein identification device 1 is used for door unlocking control. As shown in the figure, the vein identification device 1 is provided in a door knob portion of a door. A panel-like sensor unit 20 is provided on the front surface of the vein identification device 1 (shaded portion in FIG. 2B), and light source units 30 are provided on both side surfaces of the sensor unit 20 (horizontal line portion in FIG. 2B). The sensor unit 20 is arranged at a position where a finger to be imaged is placed when the identification target person grasps the door knob.

  When the person to be identified grasps the door knob portion to open and close the door, the trigger sensor 40 detects the approach of the finger and starts imaging the vein pattern of the finger. As shown in FIG. 2B, at the time of imaging, the finger pattern of the person to be identified positioned on the sensor unit 20 is irradiated with near infrared rays from the light source unit 30, so that a vein pattern image is obtained for the imaging target portion of the finger. To be acquired. Then, based on the acquired vein pattern image, it is determined whether the identification target person is a registration target person (user). When the identification is permitted, the electronic lock that is the control target 50 is unlocked, and when the identification is rejected, the electronic lock is not unlocked. Details of this identification operation will be described later.

  Note that each time the identification target person opens and closes the door, the gripping method of the door knob will change each time, so the same finger is not always placed at the same position and the same angle of the sensor unit 20. That is, it is considered that the position and imaging direction of the vein pattern imaged as the identification target change each time. However, as will be described later, in the present embodiment, identification can be performed with high accuracy irrespective of the position and direction at the time of imaging or the finger to be imaged.

<Basic operation of vein identification apparatus 1>
In the vein identification device 1, a “registration operation” in which a vein pattern is registered in advance for each finger for each user (registration target person) to be registered, and the individual vein pattern data for each registered finger is used. The “identification operation” is performed to identify the above.

  In order to identify the person to be identified based on the biometric pattern (identification based on the vein pattern in this embodiment), a determination criterion is required. Therefore, first, in the “registration operation”, a finger vein pattern is registered in the registration unit (RAM 14) of the vein identification device 1 for each user (registration target person) (also referred to as a registration mode). The registered vein pattern for each finger is compared with the vein pattern of the person to be identified in the “identification operation”, and if it is determined that both patterns match, it is identified whether the person to be registered is the person to be identified. (Also called the identification mode).

  However, data related to the vein pattern for each finger of the registration target person to be a reference may be stored in a storage device or the like, and the data may be used in the identification mode. In this case, the registration operation is not necessarily performed in the vein identification device 1. That is, the registration operation and the identification operation may be performed by different devices. The vein identification device 1 may refer to data necessary for identification by connecting to an external storage device or the like.

  In the vein identification device 1 of the present embodiment, the setting is normally made to perform the identification mode, and switching between the identification mode and the registration mode is performed by the user or the person to be identified selecting the mode. Each operation will be described below.

=== Registration operation ===
In the registration operation, “feature amount” representing the feature of the vein pattern of the finger of the user who is the registration target is extracted. Then, the extracted feature amount is registered in association with a finger ID representing the type of finger of each user. As the feature amount in the present embodiment, a position invariant and a rotation invariant are employed. This is because the same position is specified for the pattern of the same person regardless of whether the imaging range changes (position change) or the imaging direction rotates (rotation change) (position invariant). The characteristics of the local area around the position can be obtained with the same value as the feature value obtained by quantifying (rotation invariant).

<Flow of registration operation>
FIG. 3 shows a flow of the registration operation. The registration operation is performed by executing each process of S101 to S105.

  When the trigger sensor 40 detects the approach of the finger in the registration mode, the registration operation is started, and first, the imaging parameters are adjusted (S101). Based on the image of the finger (vein) input from the sensor unit 20, the imaging parameter is adjusted by the arithmetic unit 10, and a parameter that can be imaged with the best image quality is set.

  After adjusting the parameters, a vein image of the user to be registered is captured (S102). In order to capture a finger vein pattern image, light including a large wavelength band of 700 nm to 900 nm is emitted from the light source unit 30 to image the finger vein. The vein pattern can be imaged using the sensor unit 20 of the vein identification device 1 or using another imageable device.

  When imaging in the registration mode, the area where the finger is imaged is enlarged so that the above-described feature amount is included as much as possible. In order to increase the accuracy of identification, it is preferable to register all the feature amounts included in an area that may be used for identification. For this reason, in the present embodiment, the region to be imaged is larger in the vein pattern of the person to be registered imaged in the registration operation than the vein pattern of the person to be imaged in the identification operation. For example, for a plurality of fingers of a registration target person, a wide area is imaged, for example, the entire finger is imaged. In order to capture an area larger than the area captured by the identification operation, a sensor that performs imaging in the registration operation and a sensor that performs imaging in the identification operation may be provided separately. Further, when imaging such a wide area, it is not necessary to perform imaging in a lump, and it may be divided into several areas and imaged.

  In the present embodiment, when imaging with respect to a plurality of fingers, a vein pattern image is not acquired with respect to all fingers in one imaging, but imaging with respect to each finger is performed separately. A vein pattern image is acquired for each finger. For example, when imaging with respect to the index finger to the little finger of the person to be registered, the index finger is first imaged, then the middle finger, the ring finger, and the little finger are imaged in order, and four types of vein pattern images are obtained. To get. By performing imaging separately for each finger type, it is easy to extract the feature value in association with the finger type when extracting the feature value from the finger vein pattern image in the next step (S103). In addition, in order to distinguish the type of the imaged finger and give an ID, a registration target person or the like inputs through the user interface at the time of imaging, or the vein identification device 1 sends a finger to the registration target person. Information indicating the type may be stored.

  Next, feature values are extracted for each finger from the captured finger vein image (S103). As described above, by capturing a vein pattern image for each finger type, it is possible to extract a feature amount for each finger. Details of the feature amount extraction processing (S103) will be described later.

  After the feature amount is extracted from the finger vein image, it is determined whether or not the extracted feature amount is appropriate (S104). In this embodiment, since the vein pattern is identified based on the extracted feature quantity, a sufficient amount (number) of feature quantities must be extracted for accurate identification in the discrimination operation. There is. In other words, unless a sufficient number of feature quantities are extracted, they cannot be used as identification criteria. Therefore, the calculation unit 10 determines whether or not the number of extracted feature quantities is equal to or greater than a predetermined number. If the number is less than the predetermined number (No in S104), the process returns to S101 because the feature quantity is inappropriate. The image is picked up again from the parameter adjustment.

  If it is determined that the number of extracted feature values is equal to or greater than the predetermined number (YES in S104), the extracted feature value corresponds to a finger ID representing the type of finger, assuming that the feature value is appropriate. Attached and registered in the RAM 14 (S105).

  With the above processing, the registration operation is completed.

<Details of Feature Extraction (S103)>
Details of the processing content of the feature amount extraction processing (S103) will be described. FIG. 4 shows a flow of the feature amount extraction process (S103). The feature amount is extracted by sequentially executing the processes of S131 to S133.

  First, image correction of the captured vein image is performed (S131). The image correction is performed mainly for the following three reasons. (1) The finger transmittance varies from person to person, and the overall brightness of the acquired image may vary. (2) A light-dark distribution may occur due to individual differences in finger transmittance. For example, a bright image is acquired at the joint portion of the finger, and a dark image is acquired between the joints. (3) The living tissue between the vein and the epidermis may diffuse light and blur the captured vein pattern.

  In order to solve these problems, filter processing is performed. In order to solve the above problem (1), normalization is required, and therefore, it is necessary to remove the average value (DC component). Moreover, in order to solve the above problem (2), it is necessary to make uniform, and for this reason, it is necessary to remove moderate fluctuations. Therefore, in order to solve the problems (1) and (2), a high-pass filter is applied to the vein image.

  Further, in order to solve the above problem (3), sharpness processing is necessary. Therefore, an unsharp mask is applied to the vein image to emphasize high frequency components. That is, a filter integrating these high-pass filter and unsharp mask is created and applied. Specifically, the frequency response (MTF: Modulation Transfer Function) of the two filters is integrated in the frequency space, and a filter obtained by performing inverse Fourier transform on this is applied.

  Note that when there is almost no variation in luminance in the captured image, the correction processing does not necessarily have to be performed.

  Next, feature point extraction (S132) is performed. Here, the “feature point” refers to a point that appears at a fixed position of the vein pattern even when the vein image rotates or moves between a plurality of captured images. That is, even if the position / angle is shifted, the relative position with respect to the vein pattern does not change.

  In the present embodiment, since the positional relationship between the sensor and the finger is not fixed as described above, in order to calculate the feature amount of the vein pattern, there is a request for obtaining the center position (reference position). is there. The point at the center position is a feature point. As an example of a method that satisfies these requirements, SIFT (Scale Invariant Feature Transform) is employed in feature point extraction and feature amount extraction in the present embodiment. Hereinafter, a method for performing feature point extraction and feature amount extraction using SIFT will be described.

  In the feature point extraction (S132) process, first, in order to remove noise and obtain a stable feature point, an averaging process is performed by applying a Gaussian filter to the vein image. And the process which cuts the component more than a certain frequency is performed. Further, the second derivative of the image to which the Gaussian filter is applied is calculated, and the extreme value is set as a feature point candidate. Furthermore, in order to remove feature points derived from noise, only points whose absolute value of the extreme value is equal to or greater than a predetermined threshold are adopted as feature points. In the above description, the second derivative is calculated in order to obtain the feature point candidate because the edge portion having a change is extracted from the image instead of the uniform region. In addition, when light source irradiation is performed obliquely during photographing, a region that changes with a certain inclination may appear in the image, but such a region is not used as a feature point candidate. Specifically, the calculation of the second derivative is performed by convolution integration of a vein image and a Gaussian derivative.

  FIG. 5 is a diagram illustrating an example of feature points. FIG. 5 shows a partially enlarged view of a captured vein image (shaded portion in the figure) and feature points specified at the branch point of the vein. As the feature point, the location of the extreme value of the second derivative of the luminance gradient is selected, so that a portion having a large luminance change amount is selected. Further, since the place of the extreme value of the second derivative is a place where the amount of change of the first derivative is maximized, a point having a larger curvature (ie, how to bend) than the surroundings is selected. Therefore, the branch point of the vein and the inside of the blood vessel are also selected as the feature points. That is, a place that separates a vein from a place that is not so is automatically selected as a feature point.

  Next, feature amount extraction is performed (S133). The feature amount is extracted by performing the following processing on each feature point obtained in the above processing. First, the brightness gradient around the feature point is calculated.

  FIG. 6 is a diagram illustrating an example of the obtained luminance gradient. FIG. 6 shows a plurality of grids (in this embodiment, 8 × 8 grids) centering on the feature points. And the brightness | luminance gradient in each square is shown as a vector amount.

  Next, a histogram is created for the luminance gradient obtained as shown in FIG. The direction with the highest frequency is set as the reference direction of the feature amount.

  FIG. 7 is a diagram illustrating an example of a histogram of luminance gradient. The horizontal axis in FIG. 7 represents each direction when all directions (360 degrees) are divided into a predetermined number of directions (in the case of FIG. 7, the case is divided into 36 directions), and the vertical axis represents the luminance in each direction. Represents the size h. That is, in FIG. 7, a histogram in 36 directions is shown. And the value of the direction described as "peak" is the highest. Therefore, this direction becomes the reference direction of the feature amount.

  Next, an 8 × 8 cell is created again around the feature point in accordance with the reference direction selected in the above process. Then, the 8 × 8 cells are made to correspond to 4 × 4 cells, and the luminance gradient for each 4 × 4 cell is decomposed into eight-direction vectors for each cell.

  FIG. 8 is a diagram illustrating an example of the luminance gradient when the coordinate axis is aligned with the reference direction. In the left diagram of FIG. 8, the direction of the thick arrow is the above-described reference direction, and the 8 × 8 squares aligned with the reference direction are recreated and the luminance gradient is obtained again. The right diagram of FIG. 8 corresponds to the 8 × 8 cells corresponding to the 4 × 4 cells, and the luminance gradient is decomposed into vectors in eight directions for each cell.

  Here, since it is decomposed into vectors in eight directions, vector scalar quantities are obtained in the respective directions of 0 °, 45 °, 90 °, 135 °, 180 °, 225 °, 270 °, and 315 °. It is done. In addition, since these scalar amounts are obtained for each of the 4 × 4 squares, 4 × 4 × 8 = 128-dimensional scalar amounts can be obtained. In the present embodiment, the feature quantity at the feature point is a multi-dimensional scalar quantity.

  FIG. 9 is a diagram illustrating an operation for extracting a feature amount from a vein pattern in the present embodiment. The upper diagram in FIG. 9 represents finger vein pattern images captured separately for each of the index finger to the little finger, and the lower diagram represents feature value data extracted for each finger. Note that four fingers do not necessarily have to be imaged at the time of registration. For example, three fingers may be imaged, or fingers of both hands may be used.

  By capturing the finger vein image, a vein pattern image for each finger as shown in the upper diagram of FIG. 9 is obtained. Then, feature points are extracted from each finger vein pattern image. A plurality of white circles displayed in the vein image in the figure are extracted feature points. A feature amount is extracted for each of these feature points. For example, n feature amounts U1 to Un are extracted from the index finger image, and m feature amounts V1 to Vm are extracted from the middle finger image. Each feature amount is represented by a 128-dimensional scalar amount as described above. All feature amounts extracted from the image of each finger are defined as a feature amount group for each finger, and are converted into data as shown in the lower side of FIG.

  The converted feature quantity group is stored in the RAM 14 as a set together with the ID and finger ID of the person to be registered. For example, when n feature values (U1 to Un) are extracted for the index finger of a certain registration target person A as shown in FIG. 9, the feature value group [ID-A index finger (U1 to Un)] For example, each person to be registered is registered in association with the finger ID. In this specification, all feature amount data registered in association with the finger ID for each person to be registered is referred to as “registered feature amount”.

=== About Identification Operation ===
The identification operation is performed by comparing the feature amount (registered feature amount) registered in association with the finger ID for each registration target person (user) with the feature quantity extracted from the vein pattern image of the identification target person. This is a process for determining whether any registered person to be registered and the person to be identified are the same person.

  The vein identification device 1 stands by in the identification mode in a normal use scene (for example, when the door is locked in FIG. 2). In this state, when the trigger sensor 40 detects that the finger of the person to be identified approaches the vein identification device 1, the following identification operation is started.

<Flow of identification operation>
FIG. 10 shows a flow of the identification operation. The identification operation is performed by the calculation unit 10 executing each process of S501 to S509.

  First, the imaging parameters are adjusted in the same manner as S101 in the registration operation (S501). That is, the imaging parameter is adjusted by the calculation unit 10 based on the finger (vein) image input from the sensor unit, and the parameter that can be imaged with the best image quality is set.

  After adjusting the imaging parameters, the finger vein image of the person to be identified is captured (S502). The vein image is captured at the timing when the person to be identified grasps the door knob to open the door (see FIG. 2). At this time, one imaging is performed using the sensor unit 20 while irradiating near infrared rays including many wavelength bands of 700 nm to 900 nm from the light source unit 30. In the present embodiment, identification is performed using a feature amount having a position invariant / rotation invariant characteristic. Therefore, the finger position / direction at the time of identification may be different from the finger position / direction at the time of registration. In other words, the user does not need to make fine adjustments such as finger positioning at the time of imaging, which is highly convenient for the user.

  Moreover, the range of the finger to be imaged can be narrower than that during the registration operation. This is because, in a normal case, it is possible to extract a necessary number of feature quantities for individual identification only by imaging a part of a finger vein pattern. In the registration operation, vein patterns of a plurality of fingers are imaged for each registration target person, and feature amount groups are extracted for each finger (for example, four fingers from the index finger to the little finger) (the registration described above) Feature value). For this reason, in the identification operation, the feature quantity of the person to be identified extracted from the range corresponding to a part of it (for example, a part of the index finger and the middle finger) can be collated with the registered feature quantity. At this time, it is desirable that a part of a vein pattern of a certain finger of the identification target person and a part of a vein pattern of a finger adjacent to the finger are simultaneously imaged. The identification operation using the feature amount extracted from a plurality of adjacent fingers (for example, the index finger and the middle finger) is more effective than the case where the identification operation is performed using the feature amount extracted from only one finger. Since the influence of noise or the like is reduced, the identification accuracy can be increased. Details will be described later.

  Next, a feature amount is extracted from the captured vein image of the person to be identified (S503). The feature amount extraction can be performed in the same manner as in the registration operation (S103), and a plurality of feature amounts (feature amount group) are extracted from one vein image. For example, W1 to Wp are extracted as feature amounts representing the vein pattern of the person to be identified and temporarily stored in the RAM 14. Note that in the finger vein pattern imaging (S502) in the identification operation, an unspecified finger is imaged, and therefore it is impossible to determine from which finger vein pattern each extracted feature amount is extracted. . That is, at the stage of S503, the above-described feature amounts W1 to Wp are data that does not include finger information.

  If the extracted feature quantity group is appropriate, that is, if a sufficient number of feature quantities can be extracted for vein identification (S504), collation for the extracted feature quantity is performed. Processing (S505) is performed. Note that S504 can be performed in the same manner as S104 in the registration operation.

  In the matching process (S505), the score of the identification target person is calculated based on the extracted feature quantity of the finger vein pattern of the identification target person.

  First, between the feature quantity of the identification target person extracted in the identification operation and the feature quantity (registered feature quantity) of the registration target person registered in association with the finger ID in the registration operation, Presence or absence is determined. When there is a registered feature amount similar to the feature amount of the identification target person, the score for each finger of the identification target person is added according to the finger ID associated with the registered feature amount. The score of the person to be identified is calculated from the score for each finger. Details of the collation processing (S505) will be described later.

  In this method, the score is increased according to the number of similar feature amounts between the feature amount of the identification target person and the registered feature amount. Therefore, the higher the score is, the more similar the vein pattern of the person to be identified and the vein pattern of any registered user (registered person).

  Then, the score of the person to be identified obtained as a result of the matching process is compared with a predetermined threshold value (S506). As described above, the higher the score, the higher the degree of similarity between the vein pattern of the person to be identified and the vein pattern of the person to be registered, so that the individual can be accurately identified by appropriately setting the threshold value. be able to.

  When the score of the identification target person is greater than the predetermined threshold (S506: Yes), the calculation unit 10 determines that the identification target person and the registration target person match (S507). On the other hand, when the score of the identification target person is smaller than the predetermined threshold (No in S506), it is determined that the identification target person and the registration target person do not match (S508).

  The determination result is transmitted to the controlled object 50 (S509), and control according to the result is performed. For example, as shown in FIG. 2A, when the control target 50 is an electronic lock of a door, if it is determined that the identification target person matches the registration target person, it is unlocked, and the identification target person matches the registration target person. If it is determined not to be unlocked, it is not unlocked.

<Details of collation processing (S505)>
Details of the collation processing (S505) will be described. FIG. 11 shows a flow of collation processing. The collation processing is performed by executing each processing of S551 to S556 by the calculation unit 10. In the following description, it is assumed that p feature amounts (W1 to Wp) are extracted from the finger vein pattern of the person to be identified in the feature amount extraction process (S503).

  First, the j-th (j = 1, 2, 3,...) Feature quantity Wj is selected for the p feature quantities (W1 to Wp) included in the feature quantity group of the identification target person (S551). For example, at the start of the collation process, first, the first feature amount W1 is selected.

  Next, it is determined whether or not there is a feature amount similar to Wj in the above-described registered feature amount (S552). The “similarity” of the feature quantity can be determined by calculating the Euclidean distance between the two feature quantity vectors to be compared and comparing the calculated Euclidean distance with a predetermined threshold value set in advance. When these distance values are smaller than a predetermined threshold value, it indicates that the two vectors are close to each other, so that it can be determined that the two feature quantities have high similarity.

  For example, between the feature amount W1 selected in S551 and a certain feature amount Ul (ID-A index finger (Ul)) associated with the index finger of the registration target person A registered as the registered feature amount, A Euclidean distance is calculated for each. If the Euclidean distance between W1 and Ul is smaller than a predetermined threshold, it is determined that W1 and Ul are similar feature amounts. Further, if W1 and Ul are dissimilar, the similarity (similarity) between the next feature amount U2 (ID-A index finger (U2)) associated with the index finger of the person A to be registered is similar. Whether or not) is determined. Then, when there is no similar feature amount in the feature amount (ID-A forefinger (U1 to Un)) associated with the index finger of the registration subject person A, the feature amount is then associated with the middle finger of the registration subject person A. A feature amount (similarity with ID-A middle finger (V1 to Vm)) is determined.

  In this way, the similarity determination is sequentially performed between each feature amount of the selected identification target person and the registered feature amount.

  Here, the similarity determination is performed based on the Euclidean distance, but the similarity determination may be performed based on the city distance or the Mahalanobis distance.

  If a similar feature value is found in the registered feature value for the selected Wj (Yes in S552), a score is determined for each finger of the identification target person according to the finger ID associated with the registered feature value. Is increased (S553). For example, if it is determined that the feature value W1 selected in the above example is similar to the feature value Ul associated with the index finger of a certain registration target person A in the registered feature value, the index finger of the identification target person One point is added to the score. The points to be added can be adjusted. For example, when the feature amounts are similar, two points may be added for each finger.

  When the feature quantity Wj of the selected identification target person is not similar to the feature quantity of any finger of any registration target person in the registered feature quantity (No in S552), any finger of the identification target person No score is added.

  If the next feature amount (j + 1th feature amount Wj + 1) exists (S554 is No), the above-described processing (S551 to S553) is repeated for the feature amount Wj + 1 (S554).

  When there is no next feature quantity (Yes in S554), the score of the identification target person is calculated from the score of each finger of the identification target person (S555). When calculating the score, the total value of the scores of a plurality of adjacent fingers among the scores calculated for each finger of the person to be identified in the steps up to S554 is calculated, and the combination of the fingers with the maximum total value is calculated. Is calculated as the score of the person to be identified. For example, when calculating the score of the person to be identified from the total score of two adjacent fingers, the total of the index and middle finger scores, the middle and ring finger scores, and the ring finger and little finger scores are calculated. The maximum value among them is used as the score of the person to be identified.

  Then, the calculated score of the identification target person is output (S556), and by comparing this value with the above-described threshold (S506), it is determined whether the identification target person is a user (registrant). The

<Specific example of identification operation>
The flow of the identification operation will be described using a specific example. 12A to 12C are diagrams illustrating an example in which the identification operation is performed using the finger vein pattern of the person X to be identified. In this example, it is assumed that finger vein data (ID-A to ID-C) for three persons to be registered A to C are registered as registered feature amounts. The finger vein data is a feature amount registered in association with four types of finger IDs of the index finger and the ring finger. For example, as the finger vein data for the registration target person A, [ID-A index finger (U1 to Un)], [ID-A middle finger (V1 to Vm)], etc. The associated feature amount data is registered.

  In this example, when the finger vein pattern image of the person X to be identified is captured (S502), it is assumed that the imaging is performed in the range represented by the hatched portion of the sensor unit 20 in FIG. 12A. In the figure, the middle finger of the identification target person X is imaged larger than the other fingers, and a part of the index finger and ring finger are imaged. With respect to the finger vein pattern of the identification target person X included in the imaging range, the feature amounts (W1 to Wp) are extracted by the above-described method (S503).

  Next, in the matching process (S505), the similarity between each feature quantity in the registered feature quantity and each feature quantity (W1 to Wp) of the extracted identification target person X is determined. When there are similar feature amounts in the registered feature amount, the finger score of the identification target person is increased according to the finger ID associated with the feature amount. And the score for every finger is calculated like FIG. 12B by totaling the increased score for every finger.

  First, for each feature quantity (W1 to Wp) of identification target person X, feature quantities U1 to Un ([ID-A index finger (U1)] to [ID-A index finger ( Un)]) are collated in order. It is assumed that 14 points out of the feature amounts (W1 to Wp) of the identification target person X are similar as a result of collation with the registered feature amount associated with the index finger of the registration target person A. In this case, as the finger score of the identification target person X, the score of the index finger is added by 14 points (a bar graph represented by hatching in FIG. 12B).

  Next, a comparison is made with the feature amount [ID-A middle finger (V1 to Vm)] associated with the middle finger of the registration target person A, and the score of the middle finger is calculated as 5. Similarly, the score of the ring finger is calculated as 10 and the score of the little finger is calculated as 13. This finger score is a collation result between the identification target person X and the registration target person A.

  Similarly, each feature quantity of the identification target person X and the registration feature quantity (ID-B index finger) to (ID-B pinky finger) of the registration target person B are sequentially collated. As a result of collation with the registered feature amount (ID-B) of the person B to be registered, the index finger score is 16, the middle finger score is 34, the ring finger score is 17, and the little finger score is 7 (FIG. 12B). Bar graph represented in black).

  Further, as a result of collation with the registered feature amount (ID-C) of the person C to be registered, the index finger score is 11, the middle finger score is 4, the ring finger score is 10, and the little finger score is 11 (FIG. 12B is a bar graph expressed in white).

  Next, among the scores calculated for each finger of the identification target person X, the total value of the scores of two adjacent fingers is calculated. For example, in the case of ID-B in FIG. 12B, the total score of the index finger and the middle finger is 50 (= 16 + 34). The total score of the middle finger and the ring finger is 51 (= 34 + 17), and the total score of the ring finger and the little finger is 24 (= 17 + 7).

  Among these, the combination with the maximum total score is selected as a candidate score for the identification target person X. In the above-described example, as shown in FIG. 12C, 51 that is the total score of the middle finger and the ring finger is the identification target person X calculated from the comparison with the registration feature amount (ID-B) of the registration target person B. (A bar graph represented by black in FIG. 12C). Similarly, the score candidate of the identification target person X based on the result of collation with the registration feature amount (ID-A) of the registration target person A is 23 (= 10 + 13) (the bar graph represented by the oblique lines in FIG. 12C), and the registration target The score candidate of identification target person X based on the result of collation with the registered feature amount (ID-C) of person C is 21 (= 10 + 11) (bar graph represented by white in FIG. 12C).

  Among these candidates, since the score (51) obtained by collation with ID-B is the maximum, this score is the total score of the person X to be identified.

  Then, the size relationship between the score (51) of the identification target person X and a predetermined threshold is compared (S506), and whether or not the identification is permitted is determined. For example, if the threshold value is set to 40 in this example, since the score (51) of the identification target person X is greater than the threshold value (40), identification is permitted (S507). In this case, it is determined that the identification target person X is the same person as the registration target person B. As for the setting of the threshold value, an optimum value is determined by conducting an experiment in advance.

  Moreover, you may calculate the total value of three adjacent fingers as a score of identification object person X. In the case of a total of three fingers, the maximum score value according to the collation result with ID-A is 29 (= 14 + 5 + 10). Similarly, the result of collation with ID-B is 67 (= 16 + 34 + 17), and the result of collation with ID-C is 25 (= 11 + 4 + 10). Among these, since the score (67) obtained by collation with ID-B is the maximum, this score is eventually the total score of the person X to be identified. By appropriately adjusting the threshold value, it is possible to determine that the identification target person X is the same person as the registration target person B as in the case described above.

  13A to 13C are diagrams illustrating an example in which the identification operation is performed using the finger vein pattern of the person Y to be identified. The basic flow is the same as in the above example.

  In this example, when the finger vein pattern image of the identification target person Y is captured (S502), as shown in FIG. 13A, the index finger and middle finger of the identification target person Y are included in the imaging range. That is, the characteristic amount of the identification target person Y is extracted from a part of the vein pattern of the index finger and the middle finger.

  Next, as in the above-described example, each feature quantity of the identification target person Y is collated with the registration feature quantities of the registration target persons A to C. As a result of collation with the registered feature amount associated with the index finger of the registration target person A, as shown in FIG. 13B, the finger score of the identification target person Y is 24 for the index finger, 23 for the middle finger, and the ring finger Is calculated as 10 and the score of the little finger is calculated as 9 (a bar graph represented by diagonal lines in FIG. 13B). This finger score is a matching result between the identification target person Y and the registration target person A. Further, each feature quantity of the identification target person Y is collated with the registration feature quantities of the registration target persons B and C, and as a result, the bar graph (ID-B) represented by black in FIG. 13B and the white paint are represented. The finger score of the bar graph (ID-C) is calculated.

  Then, among the scores calculated for each finger of the identification target person X, the total value of the scores of two adjacent fingers is calculated, and the score 45 (= 23 + 22) obtained by collation with ID-A is the maximum. Therefore (FIG. 13B), this score becomes the total score of the person X to be identified.

  Since the calculated score and the threshold are compared, and the score of the threshold identification target person Y (45)> the threshold (40), the identification is permitted (S507). In this case, the identification target person X is determined to be the same person as the registration target person A.

  Thus, finger vein identification can be performed with high accuracy by considering a plurality of adjacent finger scores.

  The reason for considering the scores of a plurality of adjacent fingers is to reduce the influence of a mismatch that occurs by chance (a feature quantity that should not originally match is erroneously determined to be similar). For example, in FIG. 13B, the finger score for the index finger of the identification target person Y obtained by matching with the registration feature amount of the registration target person B (ID-B) is 20. On the other hand, since it is known that the identification target person Y and the registration target person B are non-identical persons, this score is considered to be calculated by the influence of noise or the like that occurred by chance.

  If the identification is performed using only the score for one finger, as shown in FIG. 13B, the score of the identification target person Y based on the verification with the registration target person A is 23 (ID-A index finger). Moreover, the score by collation with the registration subject person B is 20 (ID-B index finger), and the score by collation with the registration subject person C is 13 (ID-C ring finger) (see FIG. 13B). In this case, depending on the setting of the threshold (for example, when threshold = 19 is set), identification of the identification target person Y is permitted not only in the registration target person A but also in the relationship with the registration target person B. Misidentification occurs.

  On the other hand, when the scores of a plurality of adjacent fingers are taken into consideration as in the present embodiment, the possibility of mismatches occurring in all of the plurality of fingers is extremely low. As a result, the influence of the accidental mismatch can be made relatively small, so that the identification can be performed with high accuracy.

  It is also important to use adjacent finger data. For example, if the sum of the finger scores for the two fingers that are the maximum of the four finger scores is handled as the score of the person to be identified, the accuracy of the identification is deteriorated. This is because the data generated by the mismatch as described above is positively adopted.

  The number of fingers to be considered (adjacent) is determined according to the imaging range. For example, when imaging is performed as shown in FIGS. 12 and 13, if the sum of the finger scores for the index finger to the little finger is handled as the score of the person to be identified, the accuracy of the identification is deteriorated. Since the finger score calculated for the little finger outside the imaging range cannot be said to be reliable, taking into consideration the score of the little finger results in a decrease in identification accuracy. Therefore, in the case of the imaging range as shown in FIGS. 12 and 13, only two or three adjacent fingers are considered. Note that the number of fingers to be considered can be determined in consideration of the area of the imaging unit (area that can be imaged) and the like.

<Summary>
In the present embodiment, in the identification operation, a plurality of feature quantities characterizing the vein pattern are extracted from an image obtained by capturing the vein pattern of the person to be identified regardless of the position and direction. Then, the registered feature quantity registered in association with the finger ID representing the finger type is collated with the extracted feature quantity of the identification target person. As a result of the collation, when the feature quantity of the identification target person is similar to the registered feature quantity, a score is calculated for each finger of the identification target person according to the finger ID associated with the registration feature quantity. From the calculated score for each finger, the maximum value when the scores of a plurality of adjacent fingers are combined is set as the score of the person to be identified, and vein identification is performed by comparing the score with a predetermined threshold.

  According to this method, mismatches that occur with fingers that are outside the imaging range or fingers that are not adjacent to each other are hardly reflected in the final score of the person to be identified. Furthermore, since the influence of the feature amount that is considered to be caused by noise or the like becomes relatively small, highly accurate identification can be realized. In addition, since it is not necessary to perform positioning and guide installation at the time of imaging, it is highly convenient for users.

  Therefore, in a biometric identification device that identifies an individual using a captured finger image, it is possible to perform highly accurate identification regardless of the position and direction of the finger at the time of identification.

=== Other Embodiments ===
Although the vein identification device as one embodiment has been described, the above embodiment is for facilitating the understanding of the present invention, and is not intended to limit the present invention. The present invention can be changed and improved without departing from the gist thereof, and it is needless to say that the present invention includes equivalents thereof. In particular, the embodiments described below are also included in the present invention.

<About biometric identification device>
In the above-described embodiment, the vein identification device 1 that performs identification using a vein pattern has been described as an example of the biometric identification device. It may be a device that performs the above. For example, it is possible to identify an individual using an image such as a fingerprint of a finger.

<SIFT feature value>
In the above-described embodiment, an example in which SIFT feature values are used as a technique for extracting feature points and feature values has been described, but this is not restrictive. For example, techniques such as SURF (Speeded Up Robust Features) and GLOH (Gradient Location and Orientation Histogram) may be used.

1 vein identification device,
10 arithmetic units, 12 CPU, 14 RAM, 16 EEPROM,
20 Sensor part,
30 light source,
40 trigger sensor,
50 Control target

Claims (6)

An imaging unit for imaging a biological pattern of a finger;
An arithmetic unit,
The computing unit is
A characteristic amount characterizing the biological pattern of the identification target person from an image obtained by imaging the biological pattern of the finger of the identification target person using the imaging unit, even when the imaging range and the imaging direction change. Extract multiple features that have position-invariant and rotation-invariant characteristics with respect to the biological pattern,
Determining whether or not the feature quantity characterizing the biometric pattern of the person to be registered registered in association with the finger ID representing the kind of finger in the predetermined registration unit is similar to the feature quantity of the person to be identified;
In accordance with the finger ID associated with the feature quantity of the registration subject determined to be similar to the feature quantity of the identification subject person, the score for each finger of the identification subject person is calculated, and Calculate the total score of multiple adjacent fingers,
When the maximum value among the total values of the scores is larger than a predetermined threshold, the identification target person is identified as the registration target person.
A biometric identification device. The biometric identification device according to claim 1,
The imaging unit images at least a part of a biological pattern of one finger of the identification target person and at least a part of a biological pattern of a finger adjacent to the one finger. Identification device. The biometric identification device according to claim 1 or 2 ,
The computing unit is
From the image obtained by imaging the biometric pattern of the finger of the registration subject using the imaging unit, the feature amount of the registration subject is extracted,
The biometric identification device, wherein the registration target person's feature quantity and the registration target person's finger ID are associated with each other and registered as a registration feature quantity in the registration unit. The biometric identification device according to claim 3 ,
Than the biometric pattern of the identification subject's finger,
The biometric identification device, wherein the biometric pattern of the finger of the person to be registered has a larger area imaged by the imaging unit. The biometric identification device according to any one of claims 1 to 4 ,
The biometric identification device, wherein the feature amount is extracted using a SIFT (Scale Invariant Feature Transform). A characteristic amount characterizing the biological pattern of the finger of the identification target person from an image obtained by imaging the biological pattern of the finger of the identification target person, and even if the imaging range and imaging direction change, Extracting a plurality of features having position-invariant and rotation-invariant characteristics;
Determining whether or not the feature quantity of a registration target person registered in association with a finger ID representing a finger type is similar to the feature quantity of the identification target person;
In accordance with the finger ID associated with the feature quantity of the registration subject determined to be similar to the feature quantity of the identification subject person, the score for each finger of the identification subject person is calculated , and Calculating a total value of scores of a plurality of adjacent fingers;
Identifying the identification target person as the registration target person when the maximum value of the total score is greater than a predetermined threshold;
A biometric identification method comprising:

Images