JP2010152706A - Biometric authentication device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To improve authentication accuracy of identification processing (1: N authentication) without deteriorating convenience, in personal authentication using biological information. <P>SOLUTION: A biometric authentication device includes: a means for storing contour images of a predetermined body portion of a plurality of registrants and biological images existing in contour shapes of the contour images as registration data; a means for acquiring a contour image of the predetermined body portion of a user and a biological image existing in the contour shape of the contour image as collation data, for identification processing; a means for aligning the positions of the biological image of the collation data with the biological images of the registration data to calculate similarity; a means for selecting the registration data on the basis of the calculated similarity; a means for extracting a difference characteristic from contour images of the registration data and the contour image of the collation data when two or more selected registration data exist; a means for refining the registration data on the basis of the extracted difference characteristic to obtain an authentication result; and a means for outputting the authentication result. <P>COPYRIGHT: (C)2010,JPO&INPIT

Description

  The present invention relates to a technique for improving authentication accuracy in identification processing (1: N authentication) in personal authentication (biometric authentication) using biometric information.

  Conventionally, a magnetic card or a personal identification number (PIN) has been used as means for confirming the identity of a person without entering a person when entering or leaving a facility. However, these means have problems of loss (forgetting) and theft. As a means for identifying a person who is not worried about losing (forgetting) or stealing, personal authentication technology using personal biometric information such as fingerprints, faces, veins and the like has been actively studied.

Patent Document 1 discloses an authentication technique using a vein image (vein pattern) such as a palm as biometric information. Such vein authentication has the following characteristics.
-The burden on the subject is small.
-It is difficult to forge because it is biological information inside the body.
-Resistant to rough surfaces and applicable to most people.
・ Authentication accuracy is relatively high.
・ Since it is shot with near infrared light, it is resistant to fluctuations in external light.

  On the other hand, the personal authentication technology includes a verification process (1: 1 authentication) for determining whether a user is a specific person represented by a card or a number (ID: Identification), and a plurality of registered persons. It is classified into identification processing (1: N authentication) for determining that it is a specific one of them. In the present application, the latter identification processing is targeted.

  The basic operation of the vein authentication apparatus that performs identification processing on a vein image is as follows.

  That is, near-infrared light is irradiated to a part of a human body such as a palm that easily captures a vein, and a vein image is detected by capturing the intensity distribution of reflected or transmitted light. With such a method, a vein image is acquired and registered (stored) for each individual in advance. The registered vein image is called a registered vein image. The registered vein image can be acquired by a registration device provided separately from the device that performs the identification process.

  A vein image is detected by the same method at the time of authentication. The vein image acquired at the time of authentication is called a verification vein image.

Then, a comparison is made to determine whether or not a plurality of registered vein images match the verification vein image, authentication is performed, and an authentication result is output. However, since the biometric information varies to some extent even for the same person, it is necessary to allow a certain variation and compare in determining whether the vein images match.
US Pat. No. 4,699,149 (UK Patent No. 2156127) JP 2003-141542 A Japanese Patent No. 3356144 Japanese Patent Laid-Open No. 9-91434

  The above-described identification process (1: N authentication) has the advantage that it does not require the card presentation or the input operation of the number representing the user at the time of authentication. When biometric information is registered, there is a problem that misidentification is likely to occur.

The following methods are known as conventional techniques for reducing misidentification in identification processing.
(1) Tightening of personal identification threshold value Patent Document 2 discloses a technique for increasing (strictening) a personal identification threshold value when similar face images are registered. Although this method can reduce misidentification, there is a problem that a genuine user is not easily authenticated as the user and convenience is lowered.
(2) Increase in characteristics used for authentication processing As a known technique, a multimodal biometric technique for performing authentication with a plurality of types of characteristics is widely known. For example, if a face and a palm are photographed at the same time, and identification processing is performed by combining face authentication and vein authentication, it is possible to identify two persons who cannot be distinguished by only the face. However, since the biometric information varies to some extent even if it is the same person, there is a problem that if the features used for the authentication process are increased, it becomes difficult for a genuine user to be authenticated as the user and convenience is lowered. Furthermore, depending on the type of feature, there is a problem that the time and effort of photographing increase.
(3) Request for Additional Information (ID, Password, Other Biometric Information, etc.) Patent Document 3 discloses a method of performing alternative authentication using a password or the like when authentication is impossible. Although there is an advantage of reliably reducing misidentification, there is a problem that authentication operation becomes complicated and convenience is lowered.
(4) Refusal of registration of similar biometric information Patent Document 4 describes a method of calculating a similarity with biometric information of another person who has already been registered at the time of registration, and performing registration processing only when there is no biometric information with a high similarity Is disclosed. There is an effect that there is no registered data similar to others and misidentification is reduced, but there is a problem that a user who cannot use biometric authentication appears and convenience is lowered.

  As described above, there are several conventional techniques for reducing misidentification, but there is a problem in that convenience for users is deteriorated and convenience is lowered.

  In view of the above-described conventional problems, an object of the present invention is to improve authentication accuracy for identification processing (1: N authentication) without impairing convenience in personal authentication using biometric information.

  In one embodiment of this biometric authentication apparatus, when executing a discrimination process, means for holding a contour image of a predetermined body part of a plurality of registrants and a biometric image existing inside the contour shape of the contour image as registration data. Means for acquiring a contour image of a predetermined body part of a user and a biometric image existing inside the contour shape of the contour image as collation data, and alignment of the biometric image of the collation data and the biometric image of the registration data Means for calculating similarity, means for selecting registration data based on the calculated similarity, and when there are two or more selected registration data, the contour image of the registration data and the collation Means for extracting a difference feature from a contour image of data, means for narrowing down registered data based on the extracted difference feature, obtaining an authentication result, and means for outputting the authentication result.

  In the disclosed biometric authentication device, when there are a plurality of registration data having high similarity to given biometric information, the biometric authentication device extracts the difference features that can identify the registration data and uses them for narrowing down, In personal authentication using information, it is possible to improve the authentication accuracy for identification processing without impairing convenience.

  Hereinafter, preferred embodiments of the present invention will be described.

<Overview>
In the present embodiment, misidentification is reduced in identification processing (1: N authentication) for personal authentication using biometric information. In addition, although the case where a palm outline and a palm vein are used for authentication as biometric information will be described, the present invention can also be applied to outlines and vein images of other parts. Furthermore, biological information other than vein images can be targeted.

  The basic processing procedure is as follows. That is, the similarity between the given verification vein image and each registered vein image is calculated. In the calculation of the similarity, alignment is performed using each contour image (collation contour image, registered contour image) and vein image (collation vein image, registered vein image), and the alignment is completed. The optimal similarity is output. The alignment process will be described later.

  If there is only one registered vein image having a similarity higher than the given threshold, the ID of the registered data is returned as the authentication result. If none exists, an unregistered result is returned as an authentication result.

  Hereinafter, a case where there are a plurality of registered vein images having a similarity higher than a given threshold value will be handled. With the calculation of the similarity, the registration result of each registered vein image and verification vein image is obtained.

  Using the result of registration of the vein image, the contour image is aligned, and the difference feature that can identify the registered data is extracted from the contour shape. Details of the extraction of the difference feature will be described later.

  Then, the registered vein image is narrowed down based on the extracted difference feature, one registered data is selected, and the ID of the registered data is returned as an authentication result.

This method has the following features and has the advantage that it does not impair user convenience.
・ The identity determination threshold is unchanged, and the rejection of identity due to the fluctuation of the threshold does not increase.
・ Since the contour image and vein image can be obtained once, the time and effort of photographing does not increase.
-No additional user operation is required.
・ No rejection at registration.

<Device configuration>
FIG. 1 is a diagram illustrating a configuration example of a biometric authentication apparatus according to an embodiment.

  In FIG. 1, the biometric authentication device 1 includes an irradiation unit 2, an imaging unit 3, a storage unit 4, and an authentication unit 5.

  The irradiation unit 2 has a function of irradiating a near-infrared light to a part P of a human body such as a palm that easily captures a vein image.

  The imaging unit 3 has a function of detecting collation data by imaging the intensity distribution of light reflected or transmitted by the part P. The verification data includes a verification vein image and a verification contour image.

  The storage unit 4 registers and stores registration data for each individual in advance. The registration data includes a registered vein image and a registered contour image.

  The authentication unit 5 performs authentication processing based on the registration data (registered vein image, registered contour image) stored in the storage unit 4 and the matching data (matching vein image, matching contour image) detected by the photographing unit 3, It has a function to output the authentication result. Details of the authentication process will be described later.

  Typically, the storage unit 4 and the authentication unit 5 are realized by a computer device, and the authentication unit 5 is realized by a computer program. However, the storage unit 4 and the authentication unit 5 do not necessarily have to be placed in the same computer device. . For example, a card (IC card or the like) can be used as the storage unit 4. In addition, the storage unit 4 and the authentication unit 5 can be placed on a host computer connected to the network that is remote from the irradiation unit 2 and the imaging unit 3.

  FIG. 2 is a diagram showing an example of the data structure of the storage unit 4, which has items “ID”, “registered vein image”, and “registered contour image”, and associates the registered vein image and registered contour image with the ID. I remember. The ID is a symbol that identifies an individual user. The registered vein image is obtained by converting the value of each pixel of the vein image, the position of each feature point, and the like into a string of numbers according to a certain rule. The registered contour image is obtained by converting the value of each pixel of the contour image, the position of each feature point, and the like into a string of numbers according to a certain rule.

  FIG. 3 is a diagram illustrating a configuration example of the authentication unit 5. In FIG. 3, the authentication unit 5 includes a collation unit 51, a candidate selection unit 52, a feature extraction unit 53, a feature comparison unit 54, and an output unit 55.

  The collation unit 51 reads the collation data and the registration data, performs alignment using the contour image (collation contour image, registration contour image) and vein image (collation vein image, registration vein image), and has completed the alignment. The function of calculating the similarity between the verification vein image and each registered vein image is provided.

  If there is only one registered vein image whose similarity is higher than the given threshold, the candidate selection unit 52 passes the ID of the registered data to the output unit 55, and the similarity is higher than the given threshold. In the case where there are a plurality of registered vein images, there is a function of passing the IDs of a plurality of registered vein images having a degree of similarity up to a predetermined order to the feature extraction unit 53 in the next stage. If there is no registered vein image having a similarity higher than the given threshold value, the candidate selecting unit 52 passes information indicating unregistered (no corresponding person) to the output unit 55.

  The feature extraction unit 53 uses the alignment result in the matching unit 51 to perform alignment of the contour image, extracts a difference feature that can identify the registered data from the contour shape, and compares the difference feature with the ID in association with the feature. The function of passing to the unit 54 is provided.

  The feature comparison unit 54 has a function of narrowing down registration data based on the difference features extracted by the feature extraction unit 53, selecting one registration data, and passing the ID of the registration data to the output unit 55.

  The output unit 55 has a function of outputting registration data ID or unregistered information as an authentication result.

<Operation>
FIG. 4 is a flowchart showing an overall processing example of the above embodiment.

  In FIG. 4, the biometric authentication device 1 starts processing (step S1), and irradiates a near-infrared light from the irradiation unit 2 to a part P of a human body such as a palm that easily captures a vein image (step S2).

  The imaging unit 3 detects the collation data by imaging the intensity distribution of the light reflected or transmitted by the part P (step S3). The verification data includes a verification vein image and a verification contour image.

  An example of a technique for detecting a verification vein image and a verification contour image from an image captured by the imaging unit 3 will be described below.

On the photographed image, the background is bright, the palms etc. are slightly dark, and the veins are darkest. For example, by collating the photographed image as follows, the collation vein image and the collation contour image are obtained. Can be detected.
(1) When the value of each pixel of the photographed image is larger than a predetermined first threshold, the value representing the background is set to “0”.
(2) If the value of each pixel does not correspond to the above (1) and is larger than a predetermined second threshold value (a value smaller than the first threshold value), the value “2” representing the palm or the like is set.
(3) In other cases, the value “1” representing the vein is used.
(4) After the ternarization according to the above (1) to (3), all the pixels are scanned one by one, the target pixel is the value “2” representing the palm and the like, and the adjacent pixel represents the background In the case of the value “0”, the value of the target pixel is changed to a value “4” representing the contour.
(5) The pixel of the value “1” representing the vein is left as it is, and the value of the other pixels is set to “0”, thereby obtaining a verification vein image.
(6) From the processing result of (4) above, the pixel value other than the value “4” representing the contour is set to “0”, and the value “4” representing the contour is changed to the value “1”. A contour image is obtained.

  Returning to FIG. 4, the authentication unit 5 performs an authentication process based on a plurality of registered data (registered vein image, registered contour image) and verification data (verification vein image, verification contour image) stored in the storage unit 4. (Step S4), the authentication result is output (Step S5), and the process is terminated (Step S6).

  FIG. 5 is a flowchart showing a processing example of the authentication unit 5.

  In FIG. 5, when the authentication unit 5 starts processing (step S11), the verification unit 51 acquires the verification data (verification vein image and verification contour image) detected by the imaging unit 3 (step S12), and the storage unit 4 The registration data (ID, registered vein image, registered contour image) of a plurality of persons is acquired from (step S13). When the identification process is repeatedly performed, the registration data in the previous identification process can be held and used as it is. In that case, registration data acquisition (step S13) can be omitted.

  FIG. 6 is a diagram illustrating an example of input data of the collation unit 51, where (a) illustrates collation data, and (b) illustrates three registration data with IDs “0001”, “0002”, and “0003”. In both (a) and (b), the left side shows a vein image, and the right side shows a contour image. The vein image and the contour image are detected together, but separated into separate binary images and passed. Here, both the vein image and the contour image are binary images. The binary image is expressed by arranging the leftmost pixel one stage below the rightmost pixel in the row on the memory.

  Next, referring back to FIG. 5, the collation unit 51 performs alignment using the respective contour images and vein images for the collation data and the plurality of registration data. A similarity indicating the degree of matching of images is calculated (step S14).

  The alignment is performed to prevent the similarity value from being affected by the position and orientation of the hand. When performing registration using only the vein image, the similarity is repeatedly calculated while changing the positional relationship between the registered vein image and the verification vein image (parallel movement, enlargement / reduction, rotation, etc. of the vein image), and their similarity A method using the maximum value of degree can be considered. When aligning, one vein image is often used as a reference, and the other vein image is moved. However, the method of reference is not essential, and both vein images are moved for alignment. It is also possible to do this. However, this method has a problem that the processing is complicated and the processing cannot be performed at high speed.

  On the other hand, the contour shape of the contour image is simple compared to the vein image and is common to other people, and therefore, there is an advantage that alignment between the contours can be easily performed. For example, Japanese Patent Application Laid-Open No. 2002-92616 discloses a technique for aligning a finger vein pattern based on position information of a finger outline. Since the contour is obtained as a secondary process in the process of detecting the vein image, no new imaging is required.

  However, a contour image such as a palm is easily affected by the position and orientation, and is often not detected as stably as a vein image. Therefore, it is desirable to finely adjust the alignment between the vein images after roughly aligning with the contour image. The matching process is speeded up by keeping the alignment with a large number of registered vein images within the range of fine adjustment.

FIG. 7 is a diagram illustrating an example of a positioning method using a contour image in the matching unit 51. 7, if the contour SH ₀ contour SH ₁ and collation data registered data is given, by converting to match the standard profile that defines both advance and contour SH _{1 '} The contour shape SH ₀ ′ is obtained. The vein image inside the contour shape is also aligned by performing the same conversion.

An example of a technique for aligning the two contour images by converting the vein images so that the top, bottom, left, and right ends of the contour shapes coincide with each other is as follows.
(1) Prepare areas for recording the top, bottom, left and right edges of the contour shape.
(2) The two contour images representing the contour shape are scanned one by one for each pixel, and the contents of the area for recording the top, bottom, left and right edges of the contour shape are updated. For example, when the contour shape exists above the recorded upper end of the image, the contents of the area where the upper end is recorded are replaced with the vertical coordinate value of the pixel.
(3) For the two contour shapes, a coordinate conversion rule for mapping one up / down / left / right end position to the other up / down / left / right end position is determined. For example, coordinate conversion is performed separately for the horizontal direction and the vertical direction, and a linear expression that uses the left end as the left end and the right end as the right end in the horizontal direction is used. The vertical direction is determined in the same manner.
(4) All the pixels are scanned one by one with respect to the vein image representing the vein shape, and the coordinate value obtained by applying the coordinate conversion rule to the coordinate value of the pixel representing the vein shape is obtained. A vein image whose value is a vein shape is generated.

The calculation of the similarity in the matching unit 51 is performed by the following procedure, for example. The vein image is expressed as, for example, a binary image (“0” for pixels not corresponding to veins and “1” for pixels corresponding to veins).
(1) The initialization target pixel is set at the upper left of the image. A variable (counter) that holds the number of matched pixels is initialized to zero.
(2) Comparison of pixel values The pixel values of two images in the target pixel are acquired and compared. If they match, the counter value is incremented by one.
(3) Moving the target pixel The target pixel is moved to the right by one. In the case of the right end, it moves to the lower left one. If it is in the lower right, exit.
(4) The pixel values are compared again and again.
(5) The value of the similarity output counter is set as the similarity.

  FIG. 8 is a diagram illustrating an example of output data of the collation unit 51. Each of the registration data IDs “0001”, “0002”, and “0003” includes the similarity to the collation data and the alignment amount. . As the similarity, a value obtained by dividing the number of pixels having the same pixel value by the total number of pixels is used. The alignment amount is an alignment amount that maximizes the degree of similarity. For example, (1, 0) indicates that the registered vein image is moved by one pixel to the right in the horizontal direction.

  In the above description, the binary image is used as it is as the contour image or vein image. However, in order to reduce the storage capacity and further speed up the authentication process, the feature amount is calculated from the contour image or vein image. A method of storing them and calculating the degree of similarity by comparing feature amounts at the time of authentication is also conceivable. As the feature amount, for example, coordinate values of feature points such as vertices, branch points, and end points in a polygonal line expression can be considered. The present application is not concerned with the details of the expression form of the contour image and the vein image.

  Returning to FIG. 5, the candidate selection unit 52 determines whether or not there are a plurality of registered vein images having a degree of similarity higher than the given threshold (step S15), and when there is only one (NO in step S15). The ID of the registration data is passed to the output unit 55. The output unit 55 outputs the ID as an authentication result (step S19) and ends the process (step S20).

  If there is no registered vein image with a similarity higher than the given threshold, the candidate selection unit 52 passes information indicating unregistration to the output unit 55, and the output unit 55 outputs unregistered information as an authentication result. (Step S19), the process is terminated (Step S20).

  On the other hand, when there are a plurality of registered vein images having a degree of similarity higher than the given threshold value (YES in step S15), the candidate selection unit 52 specifies the IDs of the plurality of registered vein images whose similarity is higher than the predetermined order. To the next feature extraction unit 53 (step S16).

  FIG. 9 is a diagram illustrating an example of output data of the candidate selection unit 52. In this example, “0.95” is used as a determination threshold for the similarity of each registered data in FIG. 8, and IDs “0001” and “0003” are selected as registered data exceeding the determination threshold “0.95”. It is shown that the total number is 2. In this example, since there are a plurality (two) of similar registration images, the difference feature extraction processing is performed by the feature extraction unit 53 in the next stage.

  Returning to FIG. 5, the feature extraction unit 53 uses the alignment result in the matching unit 51 to perform alignment of the contour image, extracts a difference feature that can identify registered data from the contour shape, and associates it with the ID. The difference feature is transferred to the feature comparison unit 54 (step S17).

Hereinafter, one method for extracting the difference features in the feature extraction unit 53 will be described. Note that the contour image and the vein image are expressed as binary images, and the alignment by the collation processing is completed.
(1) One contour shape is used as a reference. The reference contour shape is called a reference contour shape, and an image including this reference contour shape is called a reference contour image. The reference contour shape is selected from, for example, a predetermined standard contour shape, a contour shape corresponding to a registered vein image having the highest similarity, or a contour shape corresponding to a verification vein image.
(2) For each pixel of the reference contour image, when the target pixel represents a contour shape (when the value of the target pixel is “1” in the contour image), the following processing is performed.
(3) A partial contour image is created by cutting out an image around the pixel of interest.
(4) An image centered on the target pixel for all pixels in the vicinity of the corresponding target pixel in the contour image of another feature extraction target (registered contour image or collation contour image that is not a reference contour image). A partial contour image is created by cutting out.
(5) The two created partial contour images are aligned, and the movement distance is calculated.
(6) The maximum value (or mean square) of the movement distance is calculated for each target pixel of the partial contour image.
(7) The coordinate value of the pixel having the maximum moving distance (or the root mean square) is the difference feature.

  FIG. 10 is a diagram illustrating an example of extraction of a difference feature in the feature extraction unit 53. In this example, with respect to the reference contour image indicated by the broken line, the contour image of the feature extraction target indicated by the solid line has the maximum amount of alignment movement in the vicinity of the child finger and the coordinate value of the child finger is the difference feature. Extracted.

  Note that since the detection accuracy of the contour shape is generally not so high, an inappropriate difference feature may be extracted due to a detection error. As a method for solving this problem, for example, a method may be considered in which a plurality of different features are extracted, and a registered contour image in which the median of the difference from the different feature value with respect to the matching contour is the smallest is used as the identification result.

  Further, as a different feature extraction method different from that described above, pixels whose pixel values do not match between the aligned contour images can be obtained.

  FIG. 11 is a diagram illustrating an example of processing data in the feature extraction unit 53. (A) shows an example of a contour image after alignment of ID “0001”, and (b) shows an example of a contour image after alignment of ID “0003”. Here, the contour image of (a) and the contour image of (b) are compared on a pixel basis, and the matching pixel is set to “0” and the mismatching pixel is set to “1”, as shown in (c). An image of the difference feature is obtained. In this image, a non-matching pixel indicated by “1” becomes a difference feature.

  Returning to FIG. 5, the feature comparison unit 54 narrows down the registration data based on the difference features extracted by the feature extraction unit 53, selects one registration data, and passes the ID of the registration data to the output unit 55 (step). S18). That is, the feature comparison unit 54 compares the difference feature of the collation data with the difference feature of each registered data, and narrows down the registered data depending on whether or not they have the same difference feature. When the difference feature is given as a coordinate value, both coordinate values are compared. When the difference feature is given by non-matching pixels as shown in FIG. 11C, the registered contour image is compared with the matching contour image only for the non-matching pixels, and the ratio of the matching pixel values is indicated as the difference feature. And the registered data having the highest similarity of the difference features is specified.

  FIG. 12 is a diagram showing an example of output data of the feature comparison unit 54, and shows a state in which the registered data of ID “0003” having the highest similarity of the different features is finally selected.

  Returning to FIG. 5, when the output unit 55 receives the ID of the registration data from the feature comparison unit 54, the output unit 55 outputs the ID as an authentication result (step S19), and ends the process (step S20).

<Comparison with other methods>
FIG. 13 is a diagram showing a comparative example with multimodal biometrics. Multimodal biometrics is a method that simply combines contour authentication and vein authentication as described above.

  In FIG. 13, if there is similar registration data in addition to the registration data of the user himself / herself in the registration data, the similarity between the contour images and the similarity between the vein images are compared for each of the verification data and each registration data. You can't identify similar people just by doing. On the other hand, in the above-described embodiment, since the difference feature is extracted from the contour image and compared after performing the alignment by the vein image, both can be identified.

  FIG. 14 is a diagram showing a comparative example of a method for combining a contour image and a vein image. Matching the contour image and the vein image means applying the vein authentication by considering the contour image and the vein image as one virtual vein image.

  In this method, as shown in FIG. 13 described above, it is possible to identify two registration data even when the registration data includes similar registration data of another person in addition to the registration data of the user himself / herself. is there. However, in general, the detection accuracy of the contour shape is often lower than the detection accuracy of the vein image. Therefore, in this comparison method, as shown in FIG. Disappear. On the other hand, in the above-described embodiment, the difference feature is extracted only when there are a plurality of registered vein images having a high degree of similarity. Therefore, even when the contour shape as shown in FIG. It is.

  FIG. 15 is a diagram illustrating an example in which erroneous identification is reduced according to the present embodiment when there are a plurality of registered vein images having a high degree of similarity.

  There are two registered vein images having a high degree of similarity to the verification vein image, and both cannot be identified by the vein image alone. If the registration result with the verification vein image is used, the registration between the two registered vein images can also be performed. The registration result of the two registered contour images can also be registered using the registration result of the vein image.

  In this example, if two registered contour images after alignment are compared, and it is found that the height of the thumb of the child finger is different, the height of the finger of the child finger is extracted as a difference feature. . By comparing the height of the thumb of the secondary finger in the matching contour image and the height of the secondary finger of both registered contour images, the one having a close height is selected. Thereby, identification of both registration data is attained. In addition, since the contour image cannot be compared with anything other than the child finger, there is a strong advantage in the variation of the contour image.

  Next, the point that the contour position is stabilized by alignment by vein image matching processing will be supplemented. Prior to the vein image matching process, it is possible to perform alignment by taking correspondence between the contour images. However, in general, the detection accuracy of the contour image is often lower than the detection accuracy of the vein image, so that the alignment by correspondence between the contour images is often low in accuracy. For this reason, as shown in FIG. 16, the position of the child finger is not necessarily close to each other in the registered contour image of the person and the collation contour image. By performing the alignment of the contour image using the alignment result obtained by the vein image matching process, the contour position is stabilized and can be correctly identified.

<System application example 1>
FIG. 17 is a diagram showing a configuration example of a system applied to entry / exit management. In FIG. 17, the automatic door 100 and the host computer 300 are connected by a communication path 200.

  The automatic door 100 is provided with an irradiation unit 101, a photographing unit 102, an information processing unit 103, a display unit 104, an opening / closing unit 105, and a communication unit 106. The irradiation unit 101 and the imaging unit 102 correspond to the irradiation unit 2 and the imaging unit 3 in the biometric authentication device 1 of FIG. The information processing unit 103 transmits the collation data (collation contour image and collation vein image) acquired from the imaging unit 102 to the communication unit 106, and the display unit 104 and the opening / closing unit 105 based on the authentication result acquired from the communication unit 106. Take control. The display unit 104 displays an authentication result for a user standing in front of the automatic door 100 using a lamp, a buzzer, or the like. The opening / closing unit 105 performs door opening / closing control. The communication unit 106 communicates with the host computer 300 via the communication path 200.

  The host computer 300 includes a communication unit 301, an information processing unit 302, an authentication unit 303, and a storage unit 304. The authentication unit 303 and the storage unit 304 correspond to the authentication unit 5 and the storage unit 4 in the biometric authentication device 1 of FIG. However, in addition to the registration data, the storage unit 304 is provided with an area for managing the entry / exit time of the registrant. The communication unit 301 communicates with the automatic door 100 via the communication path 200. The information processing unit 302 transmits the verification data acquired from the communication unit 301 to the authentication unit 303 and also transmits the authentication result acquired from the authentication unit 303 to the communication unit 301.

  FIG. 18 is a flowchart showing an example of processing for entry / exit management.

  In FIG. 18, when a user who desires to enter the room starts processing by holding his hand over the irradiation unit 101 and the photographing unit 102 installed in or near the automatic door 100 (step S21), the irradiation unit 101 The hand is irradiated with near infrared light (step S22). The imaging unit 102 captures near-infrared light reflected or transmitted from the hand, and detects collation data including a contour image and a vein image (step S23). The information processing unit 103 transmits the collation data to the host computer 300 via the communication unit 106 and the communication path 200 (step S24).

  In the host computer 300, the authentication unit 303 receives the verification data via the communication unit 301 and the information processing unit 302, and each verification data stored in the storage unit 304 and the verification sent from the imaging unit 102 of the automatic door 100 An authentication process is performed based on the data (step S25). If it is determined that the user is one of the registrants as a result of the authentication process, the authentication unit 303 records the room entry time in association with the registrant ID in the storage unit 304 (step S26). Further, the authentication unit 303 transmits the authentication result to the automatic door 100 via the information processing unit 302 and the communication path 200 (step S27).

  In the automatic door 100, the information processing unit 103 receives the authentication result via the communication path 200 and the communication unit 106, and performs processing according to the authentication result (step S28). That is, when it is determined that the user is one of the registrants, the display unit 104 notifies the user of the authentication result of entry permission, and the opening / closing unit 105 opens the door. If it is determined that the user is not a registrant, the display unit 104 notifies the user of the authentication result indicating that entry is not permitted. Thereafter, the process ends (step S29).

<System application example 2>
FIG. 19 is a diagram illustrating a configuration example of a system applied to attendance / leaving management. In FIG. 18, the shared terminal 400 and the host computer 600 are connected by a communication path 500.

  The shared terminal 400 includes an irradiation unit 401, an imaging unit 402, an information processing unit 403, a display unit 404, a communication unit 405, an authentication unit 406, and a storage unit 407. The irradiation unit 401, the imaging unit 402, the authentication unit 406, and the storage unit 407 correspond to the irradiation unit 2, the imaging unit 3, the authentication unit 5, and the storage unit 4 in the biometric authentication device 1 of FIG. The information processing unit 403 transmits the verification data (matching contour image and verification vein image) acquired from the imaging unit 402 to the authentication unit 406, and controls the display unit 404 based on the authentication result acquired from the authentication unit 406 to perform authentication. The result is transmitted to the communication unit 405. The display unit 404 displays the authentication result to the user using an LCD (Liquid Crystal Display) or the like. The communication unit 405 communicates with the host computer 600 via the communication path 500.

  The host computer 600 is provided with a communication unit 601, an information processing unit 602, and a storage unit 603. The communication unit 601 communicates with the shared terminal 400 via the communication path 500. The information processing unit 602 records attendance / leaving for the storage unit 603 based on the authentication result acquired from the communication unit 601.

  FIG. 20 is a flowchart showing an example of attendance / leaving management processing.

  In FIG. 20, when a user who goes to work or leaves the office starts holding the hand over the irradiation unit 401 and the photographing unit 402 attached to the shared terminal 400 (step S31), the irradiation unit 401 receives near-infrared light in the hand. Is irradiated (step S32). The imaging unit 402 captures near-infrared light reflected or transmitted from the hand, detects verification data including a contour image and a vein image (step S33), and passes the verification data to the authentication unit 406 via the information processing unit 403. .

  The authentication unit 406 performs authentication processing based on each registration data and collation data stored in the storage unit 407 (step S34), and passes the authentication result to the information processing unit 403.

  The information processing unit 403 displays the authentication result on the user through the display unit 404 (step S35) and asks for confirmation (step S36). If the user confirms the display contents and the authentication result is wrong, the authentication process is performed again according to an instruction from the user.

  If it is appropriate, the user ID and authentication time are notified to the host computer 600 via the communication unit 405 and the communication path 500 (step S37).

  In the host computer 600, the information processing unit 602 receives the user ID and authentication time via the communication path 500 and the communication unit 601, and records the attendance and attendance time in the storage unit 603 in association with the user ID. (Step S38). If it is the first day of the day, it will be treated as work and if it is the second time, it will be treated as work. The ID of the registered data and the employee ID in the storage unit 407 may be the same number, but the correspondence table of both may be managed on the host computer side and converted into the employee ID. Thereafter, the process is terminated (step S39).

<Summary>
The present invention has been described above by the preferred embodiments of the present invention. While the invention has been described with reference to specific embodiments, various modifications and changes may be made to the embodiments without departing from the broad spirit and scope of the invention as defined in the claims. Obviously you can. In other words, the present invention should not be construed as being limited by the details of the specific examples and the accompanying drawings.
(Appendix 1)
Means for holding, as registration data, a contour image of a predetermined body part of a plurality of registrants and a biological image existing inside the contour shape of the contour image;
Means for acquiring, as collation data, a contour image of a predetermined body part of a user and a biometric image existing inside the contour shape of the contour image when executing the identification process;
Means for aligning the biometric image of the verification data and the biometric image of the registration data, and calculating a similarity;
Means for selecting registration data based on the calculated similarity;
Means for extracting a difference feature from the contour image of the registered data and the contour image of the matching data when there are two or more selected registered data;
A means for narrowing down registration data based on the extracted difference features and obtaining an authentication result;
A biometric authentication apparatus comprising: means for outputting an authentication result.
(Appendix 2)
In the biometric authentication device according to attachment 1,
A biometric authentication device using a palm as the body part.
(Appendix 3)
In the biometric authentication device according to any one of Appendix 1 or 2,
A biometric authentication device using a vein image as the biometric image.
(Appendix 4)
In the biometric authentication device according to any one of appendices 1 to 3,
As means for calculating the similarity, first, the contour image of the collation data and the contour image of the registered data are aligned, and then the biometric image of the collation data and the biometric image of the registered data are aligned. A biometric authentication device.
(Appendix 5)
In the biometric authentication device according to any one of appendices 1 to 4,
A biometric authentication apparatus characterized in that, as means for selecting the registration data, registration data having a similarity equal to or greater than a predetermined threshold is selected.
(Appendix 6)
In the biometric authentication device according to any one of appendices 1 to 5,
A biometric authentication apparatus, wherein the registration data is selected by selecting registration data having a degree of similarity up to a predetermined order.
(Appendix 7)
In the biometric authentication device according to any one of appendices 1 to 6,
A biometric authentication apparatus characterized in that, as means for extracting a difference feature from the contour image, alignment between the contour images is performed before the extraction of the difference feature.
(Appendix 8)
In the biometric authentication device according to attachment 7,
A biometric authentication device characterized in that, as alignment between contour images before extraction of the difference feature, alignment between contour images is performed based on a result of alignment by means for calculating the similarity.
(Appendix 9)
In the biometric authentication device according to any one of appendices 1 to 8,
As means for extracting a difference feature from the contour image, a reference contour image is determined, a corresponding point of each contour image is searched for each point on the contour shape of the reference contour image, and a coordinate value of the corresponding point is changed. A biometric authentication device characterized in that a predetermined number of contour point coordinates having the largest is defined as a difference feature.
(Appendix 10)
A step of holding, as registered data, a contour image of a predetermined body part of a plurality of registrants and a biometric image existing inside the contour shape of the contour image;
A step of acquiring, as collation data, a contour image of a predetermined body part of a user and a biometric image existing inside the contour shape of the contour image when executing the identification process;
Aligning the biometric image of the verification data with the biometric image of the registration data, and calculating the similarity;
Selecting registration data based on the calculated similarity;
A step of extracting a difference feature from the contour image of the registered data and the contour image of the matching data when there are two or more selected registered data;
The process of narrowing the registration data based on the extracted difference features and obtaining the authentication result,
And a step of outputting an authentication result.
(Appendix 11)
In the biometric authentication method according to attachment 10,
A biometric authentication method using a palm as the body part.
(Appendix 12)
In the biometric authentication method according to any one of Appendix 10 or 11,
A biometric authentication method using a vein image as the biometric image.
(Appendix 13)
In the biometric authentication method according to any one of appendices 10 to 12,
The step of calculating the similarity first aligns the contour image of the collation data and the contour image of the registered data, and then aligns the biometric image of the collation data and the biometric image of the registered data. A biometric authentication method.
(Appendix 14)
In the biometric authentication method according to any one of appendices 10 to 13,
The biometric authentication method characterized in that the step of selecting the registration data selects registration data having a similarity equal to or higher than a predetermined threshold.
(Appendix 15)
In the biometric authentication method according to any one of appendices 10 to 14,
The biometric authentication method characterized in that the step of selecting the registration data selects registration data having a degree of similarity up to a predetermined upper order.
(Appendix 16)
In the biometric authentication method according to any one of appendices 10 to 15,
The step of extracting a difference feature from the contour image includes performing alignment between the contour images before extracting the difference feature.
(Appendix 17)
In the biometric authentication method according to attachment 16,
A biometric authentication method characterized by performing alignment between contour images based on a result of alignment in the step of calculating the similarity as alignment between contour images before extraction of the difference feature.
(Appendix 18)
In the biometric authentication method according to any one of appendices 10 to 17,
The step of extracting the difference feature from the contour image determines a reference contour image, searches for the corresponding point of each contour image for each point on the contour shape of the reference contour image, and changes the coordinate value of the corresponding point. A biometric authentication method characterized in that a predetermined number of contour point coordinates having the largest is defined as a difference feature.

It is a figure which shows the structural example of the biometrics apparatus concerning one Embodiment. It is a figure which shows the example of a data structure of a memory | storage part. It is a figure which shows the structural example of an authentication part. It is a flowchart which shows the example of the whole process of embodiment. It is a flowchart which shows the process example of an authentication part. It is a figure which shows the example of the input data of a collation part. It is a figure which shows the example of the method of position alignment by the outline image in a collation part. It is a figure which shows the example of the output data of a collation part. It is a figure which shows the example of the output data of a candidate selection part. It is a figure which shows the example of the extraction of the difference feature in a feature extraction part. It is a figure which shows the example of the processing data in a feature extraction part. It is a figure which shows the example of the output data of a feature comparison part. It is a figure which shows the comparative example with multimodal biometrics. It is a figure which shows the comparative example with the system which matches an outline image and a vein image. It is a figure which shows the example in which misidentification reduces by this embodiment. It is a figure which shows the example of stabilization by the collation process of a vein image. It is a figure which shows the structural example of the system applied to the entrance / exit management. It is a flowchart which shows the process example. It is a figure which shows the structural example of the system applied to the attendance management. It is a flowchart which shows the process example.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Biometric apparatus 2 Irradiation part 3 Imaging | photography part 4 Memory | storage part 5 Authentication part 51 Verification part 52 Candidate selection part 53 Feature extraction part 54 Feature comparison part 55 Output part P site | part

Claims (6)

Means for holding, as registration data, a contour image of a predetermined body part of a plurality of registrants and a biological image existing inside the contour shape of the contour image;
Means for acquiring, as collation data, a contour image of a predetermined body part of a user and a biometric image existing inside the contour shape of the contour image when executing the identification process;
Means for aligning the biometric image of the verification data and the biometric image of the registration data, and calculating a similarity;
Means for selecting registration data based on the calculated similarity;
Means for extracting a difference feature from the contour image of the registered data and the contour image of the matching data when there are two or more selected registered data;
A means for narrowing down registration data based on the extracted difference features and obtaining an authentication result;
A biometric authentication apparatus comprising: means for outputting an authentication result. The biometric authentication device according to claim 1,
As means for calculating the similarity, first, the contour image of the collation data and the contour image of the registered data are aligned, and then the biometric image of the collation data and the biometric image of the registered data are aligned. A biometric authentication device. In the biometric authentication device according to any one of claims 1 and 2,
A biometric authentication apparatus characterized in that, as means for extracting a difference feature from the contour image, alignment between the contour images is performed before the extraction of the difference feature. The biometric authentication device according to claim 3.
A biometric authentication device characterized in that, as alignment between contour images before extraction of the difference feature, alignment between contour images is performed based on a result of alignment by means for calculating the similarity. The biometric authentication device according to any one of claims 1 to 4,
As means for extracting a difference feature from the contour image, a reference contour image is determined, a corresponding point of each contour image is searched for each point on the contour shape of the reference contour image, and a coordinate value of the corresponding point is changed. A biometric authentication device characterized in that a predetermined number of contour point coordinates having the largest is defined as a difference feature. A step of holding, as registered data, a contour image of a predetermined body part of a plurality of registrants and a biometric image existing inside the contour shape of the contour image;
A step of acquiring, as collation data, a contour image of a predetermined body part of a user and a biometric image existing inside the contour shape of the contour image when executing the identification process;
Aligning the biometric image of the verification data with the biometric image of the registration data, and calculating the similarity;
Selecting registration data based on the calculated similarity;
A step of extracting a difference feature from the contour image of the registered data and the contour image of the matching data when there are two or more selected registered data;
The process of narrowing the registration data based on the extracted difference features and obtaining the authentication result,
And a step of outputting an authentication result.

Images