JP3810782B2 - Personal recognition device - Google Patents

Description

  The present invention relates to an individual recognition apparatus, and more particularly to an individual recognition apparatus that detects an individual's physical characteristics, more specifically, a blood vessel image of a hand, in addition to an individual identification number (ID), and performs individual recognition.

  Conventionally, there is a method of using a magnetic card or a personal identification number (PIN) as a method for confirming the identity of a person without entering a person when entering or leaving a facility. However, these methods have problems such as card loss (forgetting) and theft. As a method for identifying a person who is not worried about losing (forgetting) or stealing, a method using a physical characteristic of an individual such as a fingerprint or a face has been studied.

  Verification processing for determining whether a user is an individual represented by a card or a personal identification number depending on a difference in physical characteristics, and identification processing for determining that the user is a specific one of registered persons Done. Here, the verification process and the identification process are collectively referred to as personal recognition.

  Fingerprints, irises, faces, hand blood vessels, etc. are used as physical features in personal recognition. Devices for measuring these characteristics are roughly classified into contact type and non-contact type depending on the measurement method. In entrance / exit management of facilities, it is necessary to recognize an unspecified number of individuals, and it is desirable to be a non-contact type considering hygiene and user psychological resistance.

  The physical feature most commonly used at present is a fingerprint, but it is necessary to press the finger, and the measuring device becomes a contact type. Although there are non-contact types of devices for measuring the iris, there are problems that the device is expensive and that the psychological burden on the user during photographing is large. Face measurement can be performed using a low-cost camera that is non-contact type, but recognition accuracy cannot be increased at present due to changes in facial expressions.

The personal recognition technique using the blood vessel image of the hand was basically disclosed in the following document 1 and document 2. The imaging of the blood vessel image of the hand is characterized in that the burden on the user is small, the recognition accuracy is relatively high, and the imaging is performed using near infrared light, so that it is resistant to fluctuations in external light.
British Patent 2,156,127 U.S. Pat. No. 4,699,149

  The blood vessel image of the hand can be taken without bringing the hand into contact with the photographing apparatus. However, from the viewpoint of improving the recognition accuracy, it is advantageous to photograph with the user's hand fixed to the apparatus. Representative products currently on the market are all contact-type devices that take a blood vessel image by touching a part of the device with a hand. For example, the following document 3 discloses an invention related to a product of a Korean company that takes a blood vessel image by pressing the back of the hand against a camera. This device is a form of picking up a stick by hand and photographing the back of the hand. References 4 and 5 disclose related inventions of a product of a US company that picks up a stick by hand and photographs the palm.

JP-A-10-295664 Japanese National Patent Publication No. 11-512203 US Pat. No. 5,793,881

  As described above, the personal recognition device using the blood vessel image of the hand that is currently on the market is a contact type that requires the hand to be pressed against the device, but the user is burdened from the viewpoint of hygiene and cleanliness. It is large, and there is a problem such as stickiness especially in the summer, and it is desirable to photograph a hand blood vessel without contact. However, when taking a blood vessel image of a hand without contact, there is a problem that it is difficult to obtain a good blood vessel image.

  As a problem when a blood vessel image is photographed in a non-contact manner, first, there is a problem that the manner of opening the hand, the position, the orientation, and the like change every time photographing is performed without fixing the hand. . Secondly, for example, there is a problem that near-infrared light is not sufficient and is easily affected by external light.

  In order to capture a better blood vessel image, it is preferable to illuminate the skin surface with near-infrared light at a substantially right angle, and the optical axis of the camera is also preferably orthogonal to the skin surface. There is also a third problem that the camera and the illumination are not necessarily in an optimal positional relationship, and the quality of the captured image tends to deteriorate.

  In view of the above-described problems, an object of the present invention is to realize a sanitary and clean personal recognition device with less burden on the user by capturing a blood vessel image of a hand without touching the hand. It is. Although it is difficult to acquire a good blood vessel image under non-contact conditions as described above, it is possible to acquire a good blood vessel image even under non-contact conditions by devising the device configuration, and further non-contact imaging. It is also a problem to achieve sufficient recognition accuracy by using a recognition method suitable for the above.

Personal recognition device of the present invention, in order to photograph the blood vessel image of a hand in a non-contact, the position facing shape instructing means for instructing holding the hand of the hand to the user, one or more that irradiates near infrared light on the hand And a photographing device capable of photographing a blood vessel image without contact with a hand. Also, a blood vessel image detection means for detecting a blood vessel image from the captured image, the user storage means for storing a blood vessel image that is registered in each person, and a blood vessel image detected blood vessel image registered Recognizing means for performing personal recognition in comparison.

Furthermore, a detection unit that detects the position and / or orientation of the hand each time a blood vessel image is detected, and the detected blood vessel image are corrected. If the correction result is appropriate, the corrected image is used as a recognition unit. And a correction unit that gives the position and shape when the correction result is inappropriate.

  FIG. 1 is a block diagram showing the principle configuration of a personal recognition apparatus according to the present invention. FIG. 1 shows the principle of an apparatus for performing personal recognition by detecting a blood vessel image of a hand. The personal recognition apparatus 1 includes an imaging device 2, a blood vessel image detection means 3, a blood vessel image storage means 4, and a recognition means 5.

  The photographing device 2 photographs a position / orientation shape instructing means 6 for instructing the user how to hold the hand, one or more irradiation means 7 for irradiating the hand with near infrared light, and an image by the near infrared light. It is an imaging device that includes one or more imaging means 8 and that can image a blood vessel image without contact with the hand.

  The blood vessel image detecting means 3 detects a blood vessel image from images taken by one or more photographing means 8, and the blood vessel image storage means 4 stores a blood vessel image of a hand registered for each user. The recognizing means 5 performs personal recognition by comparing the registered blood vessel image with the detected blood vessel image.

  In the embodiment of the invention, as the position / orientation shape instructing means 6, the surface on which the user should hold the hand can be configured in a planar shape, and the appropriate position and orientation of the hand can be represented graphically. A hand mold can be used as an example of the diagram. It is also possible to provide a transparent plate or a half mirror that the user should hold in front of the photographing device, or store the photographing device in the hole of the recognition device or along the groove provided in the recognition device.

  Various techniques are used to improve the practicality and recognition accuracy of the personal recognition device. For example, a plate that transmits near-infrared light to the surface on which the user holds the hand and reinforces the strength of the photographing device can be attached, or a mounting base that can adjust the surface on which the hand is held to an arbitrary angle is used. Various methods are used.

  Further, in the embodiment, a plurality of irradiation means are provided, and near-infrared light is irradiated by the plurality of irradiation means 7 in accordance with how the outside light other than near-infrared light enters one or more photographing means 8. It can also be controlled. Next, in order to improve the recognition accuracy, the personal recognition device 1 includes a detection unit that detects the position and / or orientation of the hand from the image of the hand that is captured every time the blood vessel image is detected, and the detected position of the hand. And / or a correction unit that corrects the detected blood vessel image corresponding to the orientation and gives the corrected image to the recognition unit 5 when the correction is appropriate. If the correction is inappropriate, the position / direction shape instructing means 6 can notify the user of the fact.

  In this case, the detection unit can detect the orientation of the hand based on the non-uniformity of the reflected light intensity on the image, and the position and / or orientation of the hand corresponding to the hand shape on the image. Furthermore, the personal recognition device 1 includes hand shape storage means for storing a parameter registered corresponding to the hand shape for each user, and the detection means stores the registered parameter and the parameter on the image. In comparison, the hand orientation can also be determined.

Further, the detection means can approximate the hand in a plane to detect the orientation of the hand, and the correction means can correct the blood vessel image using a parameter representing the plane.
In the embodiment, the personal recognition apparatus 1 determines whether or not a hand is held over the photographing device 2 and causes the recognition means 5 to perform personal recognition when it is determined that the hand is held. A recognition control means can be further provided.

  Furthermore, the personal recognition device 1 of the present invention includes a plurality of photographing means 8, a calculating means for calculating the three-dimensional shape of the hand from the images of the hands photographed by the plurality of photographing means 8, and corresponding to the calculation result, The image processing apparatus may further include correction means for correcting the detected blood vessel image.

  Next, the personal recognition apparatus of the present invention includes one or more irradiation means 7 and one or more photographing means 8, and has a small and portable pen shape and photographs a blood vessel image without contact with a hand. A possible imaging device 2, a blood vessel image detection means 3, a blood vessel image storage means 4, and a recognition means 5 are provided.

  In this case, the personal recognition apparatus 1 may further include a joint mechanism that allows the photographing device 2 to rotate, a tripod that fixes the photographing device, or an attachment portion that attaches the photographing device to the external connection mechanism of the computer.

  As described above, according to the present invention, a blood vessel image of a hand is taken in a non-contact manner, the position and orientation of the hand are detected as necessary, and the position and orientation are corrected if they are inappropriate. Recognition accuracy can be improved.

  FIG. 2 is a block diagram showing the configuration of the personal recognition apparatus according to the embodiment of the present invention, and FIG. 3 is a flowchart of personal recognition processing. This personal recognition device irradiates each part of the human body, such as the palm, the back of the hand, and the wrist, which is easy to photograph blood vessels, and irradiates near-infrared light to capture the intensity distribution of reflected or transmitted light, and the blood vessel image is obtained from the image Detected. Generally, personal recognition is performed by determining whether or not the blood vessel image stored in the apparatus matches the photographed blood vessel image, and the result is output.

  In FIG. 2, the irradiation part 10 is LED (light emitting diode) which light-emits near infrared light, for example, and many LED can be used as needed. The photographing unit 11 is a CCD camera or a CMOS camera, and a plurality of cameras are used as necessary.

  The position detection unit 12 and the orientation detection unit 13 detect the position and orientation of the hand from the captured image, and the position / orientation correction unit 14 corrects the position and orientation of the blood vessel image based on the detection results. This improves the recognition accuracy. However, if the correction amount is larger than the predetermined value, it is determined that the correction is inappropriate and the correction is not performed. In that case, the position / direction shape instructing unit 15 notifies the user that it is inappropriate.

  In FIG. 2, the registered blood vessel image of each individual is stored, for example, in the form of a card 16, and the card 16 stores the personal ID and blood vessel image data, for example, data corresponding to each pixel of the blood vessel image. Is stored, and the data is read by the card reading unit 17 and given to the recognition unit 18. Here, a card is used as the storage unit 16, but an external storage device such as a hard disk may be used.

  The position detection unit 12, the direction detection unit 13, the position / orientation correction unit 14, the card reading unit 17, and the recognition unit 18 are generally realized as a computer and a program thereof, but it is not always necessary to place all of them in the same computer. For example, the recognition unit 18 may be realized on the host computer connected via a network.

  In FIG. 2, it is assumed that the personal recognition device is applied to entrance management in combination with an automatic door. If the recognition unit 18 recognizes that the person matches the personal ID of the card, the control unit 19 unlocks the door and allows the individual to enter the room.

  When the processing is started in the recognition processing of FIG. 3, first, in step S1, for example, an ID of a card inserted into the card reader and a registered blood vessel image corresponding thereto are read. In step S2, how to hold the hand is instructed. This instruction will be described later. In step S3, the hand is irradiated with near infrared light, and a blood vessel image is detected from the image taken in step S4. This detection will be described later.

  Since the captured image generally includes an image of the entire hand, the position of the hand is detected in step S5 using the image, and the orientation is detected in step S6, and these detection results are used in step S7. The position and orientation of the detected blood vessel image are corrected so that the recognition accuracy is improved. These detection and correction will be described later. If the correction is inappropriate, the hand is again instructed in step S2.

  Subsequently, in step S8, the detected and corrected blood vessel image is compared with the blood vessel image read in step S1, and if the comparison result between the registered blood vessel image corresponding to the ID is OK, step S9. If the door is unlocked and the comparison result is NG, the door is not unlocked and the process is terminated.

  As described above, the personal recognition process is classified into a verification process for identifying a specific individual corresponding to an ID and an identification process for determining that the user is a specific one of a plurality of registered users. These processes will be described in more detail. The processing procedure of the verification process is as follows.

1. ID input Accepts a personal ID specified by a card or PIN . However, when the verification process is repeatedly performed on the same personal ID, the input of the ID can be omitted.
2. Acquisition of registered blood vessel image A registered blood vessel image specified by the inputted personal ID is acquired from the storage unit 16. When the verification process is repeatedly performed, the registered blood vessel image in the previous verification process can be stored and used as it is.
3. Acquisition of imaging blood vessel image An imaging blood vessel image is acquired from the imaging unit 11.
4). Calculation of similarity The similarity that is an amount representing the degree of coincidence between the two is calculated.
5). Determination If the degree of similarity is greater than a predetermined threshold, verification is successful, and the user is determined to be an individual (person) represented by a personal ID. Otherwise, it is a verification failure and it is determined that the user is another person.

On the other hand, the processing procedure of the recognition unit 18 when performing the identification processing is as follows.
1. Acquisition of registered blood vessel images The blood vessel images of a plurality of persons registered in the storage unit 16 are acquired. When the identification process is repeatedly performed, the registered blood vessel image in the previous identification process can be stored and used as it is.
2. Acquisition of imaging blood vessel image An imaging blood vessel image is acquired from the imaging unit 11.
3. Calculation of similarity The similarity that is an amount representing the degree of coincidence between the registered blood vessel image and the captured blood vessel image is calculated for each person.
4). Judgment The individual with the highest similarity is selected. When the maximum value of the similarity is smaller than a predetermined threshold, it is determined that there is no applicable person (not registered). If not, the user determines that the selected individual is the largest individual.

  FIG. 4 is a configuration example of the photographing unit 11 of FIG. In the figure, the imaging unit 11 includes a visible light cut filter 20 that blocks visible light from input light, a CMOS camera 21, and an image processing unit 22 that extracts a blood vessel image from a captured image of the CMOS camera 21.

  As described above, it is preferable to use an LED that emits near-infrared light as the irradiation unit 10 in FIG. As a camera, a CCD camera can be used instead of the CMOS camera 21. A CMOS camera is small and inexpensive, but a CCD camera can take a more beautiful image and can select either one according to the application.

  The blood vessel is extracted by performing binarization processing on the photographed image by the image processing unit 22 in order to be represented by a dark pattern on the image photographed by the CMOS camera 21, for example, with the image of the entire hand as a background. The The binarization process is realized by setting the pixel value to 1 when the (density) value of each pixel in the image is larger than a predetermined value, and to 0 otherwise. The binarization processing can be realized by a dedicated IC (microcomputer), or can be realized as a program in the computer as preprocessing by the recognition unit 18 in FIG. FIG. 5 is an example of a detected blood vessel image.

The similarity, which is an amount representing the degree of coincidence between the photographed blood vessel image and the registered blood vessel image, is calculated as follows, for example.
1. Initialization Set the target pixel at the top 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 at the target pixel are acquired and compared. If they match, the counter value is incremented by one.
3. Move pixel of interest Move the pixel of interest 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). Repeat the pixel value comparison again. (Repeat from 2)
5). Similarity output The value of the counter is the similarity.

  The similarity calculation method described above has a problem in that the similarity value is affected by the position and orientation of the hand. These effects can be reduced by improving the similarity calculation method. In the above description, a binary image is used as a blood vessel image as it is. However, in order to reduce the storage capacity and speed up the recognition process, feature amounts are calculated from the images and stored. A method of calculating the similarity by comparing the above is also conceivable. As the feature amount, for example, it is conceivable to extract feature points such as branch points and end points from an image and use their positions. In the present embodiment, the details of the expression form of the blood vessel image and the comparison method are not concerned.

  The storage unit 16 stores the personal ID and the blood vessel image in association with each other, and is typically realized in the computer as a data storage format database shown in FIG. The blood vessel image is stored by converting the value of each pixel of the image representing it, the position of each feature point, and the like into a string of numbers according to a certain rule.

The blood vessel image registration process can also be performed using the same apparatus. The processing procedure is as follows.
1. ID input Accepts a personal ID specified by a card or PIN . However, if a personal ID has already been determined, the input of the ID can be omitted.
2. Acquisition of imaging blood vessel image An imaging blood vessel image is acquired from the imaging unit 11.
3. Registration of blood vessel image The personal ID and the photographed blood vessel image are registered in the storage unit 16.

  When taking a blood vessel image of a hand without touching and performing personal recognition as described above, the image is taken based on how the hand opens, changes in position and orientation, the influence of external light, and the relationship between the hand, camera, and lighting. There is a tendency that the quality of the image is lowered and the similarity with the registered blood vessel image of the person is lowered. If the similarity threshold is lowered in order to avoid being erroneously determined to be another person, the probability that the other person is erroneously determined to be the person will increase, and the recognition accuracy will decrease.

  In the present embodiment, in order to avoid such a decrease in recognition accuracy, a position / orientation shape instructing unit 15 that instructs the user how to hold the hand is provided. In addition, the recognition method is also devised. First, the position / orientation shape instruction unit 15 will be described in detail.

  Since the hand is an organ that can be moved most freely in the human body, if the user is given appropriate instructions, fluctuations in the manner of holding can be minimized. Specifically, the irradiation unit 10 (LED or the like) and the camera are stored in a thin planar photographing device so that the user naturally spreads his / her hand without being conscious of it. This makes it possible to make the illumination direction and the optical axis of the camera substantially orthogonal, thereby improving the quality of the captured image. Furthermore, there is an effect of stabilizing the opening of the hand.

  In order to prevent the position and orientation of the hand from changing every time the image is taken, the appropriate position and orientation of the hand are represented graphically on the surface on which the user should hold his hand. As an example of the diagram, a hand mold can be used. Alternatively, the tip position of the finger can be indicated by a point. The hand mold is composed of a planar seal. Alternatively, a three-dimensional shape using a material such as plastic is used. In an environment used by infants and the like, in order to weaken the sense of resistance of infants, it may be possible to perform personal recognition by taking a handshake in the shape of a life-size hand such as a popular doll.

  When it is determined that the correction processing described later is inappropriate, it is possible to instruct the user of an appropriate position and orientation. For example, it is possible to attach an LED to the tip position of a finger and blink a portion that is inappropriate for correction, or arrange LEDs side by side and indicate an appropriate direction by the flow of light. In addition, the user can be notified by voice, bell, lamp lighting, or the like. The notification content simply tells you that it is inappropriate or the reason for the inappropriateness, for example, that it is tilted.

  It is preferable to give the photographing device strength in consideration of the possibility of being hit with a palm. One method of providing strength is to put a material such as a transparent plastic (on the near infrared light) on the surface to be photographed. At this time, if a material that is opaque with respect to visible light is used, the internal structure is concealed to provide a security measure. The photographic equipment should be sized so that it can be easily transported and installed so that it can be placed in a position where you can easily hold your hand depending on the shooting environment. A mounting base can be installed so that it can be installed at an angle. FIG. 7 shows the appearance of the photographing device.

  If the hand is very close to the camera, the registered part may not be captured because the range of hands that can be taken is limited. As a method for solving this problem, there are a method using a wide-angle lens such as a fisheye lens and a method of arranging a plurality of cameras (FIG. 8). Images taken by a plurality of cameras are integrated into a single taken image. Another method is to install a transparent plate or half mirror on the front of the camera to increase the distance to the camera while allowing the user to widen the palm (FIG. 9).

  Although shooting is performed by irradiating near-infrared light, it is difficult to be influenced by external light, but it is desirable to remove external light as much as possible in an environment where a large amount of external light enters, such as outdoors. As an apparatus shape for removing external light, there is the following form in which the photographing apparatus of FIG. 7 is stored in a hole or groove (FIG. 10). In the left of FIG. 10, photographing devices are installed on the front and rear surfaces of the hole and on the upper and lower surfaces in the center, so that the palm and the back can be photographed simultaneously. There is also a form in which a goodwill-type shielding plate is placed behind the hand to be presented (left in FIG. 11). A reflector can be attached to the inside of the goodwill to eliminate the need to put the hand into the goodwill (right in FIG. 11). The angle of the reflector is preferably variable.

  As another means for reducing the influence of external light, there is a method in which a plurality of LEDs are arranged to increase the illumination intensity of the illumination and apply it evenly (FIG. 12). The illumination may be adjusted or switched so that the influence of external light is reduced. For example, the direction of sunlight is calculated from the mounting position and the current time, and the LED in the direction close to it is turned on.

  Another possible method is to reduce the size of the photographic equipment so that it can be easily transported and to remove the influence of external light by installing it in a place where external light does not enter during use. As a device shape for realizing this, there is a pen type shown in FIG. When photographing with a camera and LED stored at the tip of the device, the joint is bent and used on a fixed base. A fixed base with a hole or a tripod is used as an instrument to fix the device. When used with a notebook-type portable computer, it can also be attached to the computer and fixed (such as clipped or inserted into an expansion slot).

The position and orientation of the hand is detected for each shooting, and the following processing is performed based on the detected position and orientation to solve the problem that the position and orientation of the hand changes for each shooting.
A proper correction process is performed in the position / orientation correction unit 14.
If the correction is inappropriate, the position / orientation shape instructing unit 15 notifies the user that it is inappropriate and asks the user to correct it.

  As a method for detecting the orientation of the hand, for example, there is a method for detecting non-uniformity of the reflection intensity. This is based on the principle that when the palm is tilted and presented to the camera, the reflection intensity from a distant place is reduced (FIG. 14).

  The non-uniformity is detected as follows, for example. The captured image is divided into four parts vertically and horizontally, and the average of the pixel values (partial images) in each region is calculated. The difference between the calculated four average values is sufficiently small (for example, whether or not it is below a threshold value) ), It is determined that the reflection intensity is uniform and the direction of the hand is flat. Otherwise, it is determined that the region with the highest average value is inclined in the direction of approaching.

  As another method for detecting the orientation of the hand, there is a method for obtaining a hand shape on an image. In this case, the field of view of the camera is set wide so that the entire hand can be photographed. Multiple cameras may be combined to expand the field of view. Since the hand is illuminated, the value of the pixel corresponding to the hand is large, and since the background is not illuminated, the value of the pixel corresponding to the background is small (FIG. 15). Therefore, a hand shape on the image can be obtained by extracting a pixel having a large value by binarization processing. The orientation of the hand is obtained by comparing the extracted hand shape with the registered (or average) hand shape. For example, the length in the vertical and horizontal directions is obtained from each hand shape, and the ratio (vertical length / horizontal length) is obtained. When the difference between the ratios of the two hand shapes is smaller than a predetermined threshold value, it is determined that the direction of the hand is flat. For example, in FIG. 16, the extracted hand shape is shortened in the vertical direction, and is determined to be inclined. As another comparison standard, the ratio of the circumference of the hand and the width of one finger, for example, can be considered. This is because the finger width is almost constant even when the hand is tilted.

A method for correcting the orientation by the recognition unit 18 will be described. Based on the detected hand inclination, an image obtained when the hand is viewed from the front is synthesized, thereby correcting the orientation, for example, by approximating the palm with a plane and obtaining the orientation of the surface. The plane is expressed by three parameters a, b, and c (average distance and vertical and horizontal inclinations) as z = ax + by + c. When estimating these parameters based on the above-described non-uniformity detection, it is assumed that the reflection intensity is inversely proportional to the square of the distance, for example, as the relationship between the reflection intensity and the distance. With this assumption, the average distance in each area of the image can be calculated,

Average distance = a * Coordinate value in the horizontal direction of the center of the region + b * Coordinate value in the vertical direction of the center of the region + c

The following relational expression is obtained for each region. By solving this equation, the parameters a 1 , b, and c are obtained. By using these parameters, it is possible to synthesize an image when the hand is photographed from the front by projective transformation. Details are described in the following document 6, for example.
Image Analysis Handbook, University of Tokyo Press

  In the obtained plane parameters, a and b represent the inclination of the hand. These absolute values represent the degree of correction. When the degree of correction is large, the image quality after correction deteriorates and the recognition accuracy decreases. Therefore, for example, whether correction is appropriate or inappropriate is determined based on whether or not | a | + | b | is larger than a given value. If it is determined that the correction is inappropriate, the user is notified without performing the correction process.

  When correcting the orientation from the above-described hand shape on the image, use the above-mentioned method of obtaining the plane parameters and performing the projective transformation, or as a simpler method, the image in the ratio of the vertical and horizontal directions. The method of enlarging is used. When features extracted from an image are used as a blood vessel image, features obtained from an image photographed from the front may be directly synthesized without synthesizing the image photographed from the front.

  Next, a method for detecting the position of the hand will be described. The hand position is detected by obtaining the hand shape on the image. Set the field of view of the camera wide so that you can shoot the entire hand. Multiple cameras may be combined to expand the field of view. As described with reference to FIG. 15, a hand shape on an image can be obtained by extracting a pixel having a large value by binarization processing. The average position of the pixels corresponding to the hand is determined as the average position of the hand. The average position of the hand is also obtained in advance for the registered data, and the positional deviation between the two is corrected at the time of recognition. Specifically, the displacement is corrected by performing a parallel movement process of the amount represented by the difference between the two on the pixel value of the captured image.

  Finally, the time (timing) at which recognition processing is performed will be described. In entry / exit, etc., the time at which personal recognition is required is not usually determined, so basically the photographing unit 11 must be operated at all times. However, in the case where operations such as card presentation and password input are preceded, it is sufficient to operate the photographing unit 11 for a certain period thereafter. In addition, the position / orientation shape instruction unit 15 may prompt the user to present a hand. The photographing unit 11 repeats the process of photographing and extracting the blood vessel image. From the viewpoint of preventing malfunction of the blood vessel image extraction process, it is preferable to determine whether or not the hand is held up and extract the blood vessel image only when the hand is held up. Whether or not the hand is held up is determined by, for example, whether or not the average value of the pixels of the captured image is equal to or greater than a threshold value. It is also possible to select an optimal image from a plurality of captured images that are continuous in time. As the selection criterion, for example, the one with the maximum average value of the pixels of the photographed image is used. Alternatively, the one having the largest hand shape area on the image is used.

Next, a method for further improving the recognition accuracy by taking advantage of non-contact will be described.
1) Combined hand shape Utilizing the detection of the shape and orientation of the hand and using them for recognition improves the recognition accuracy. The hand shape is also registered at the time of registration, and the match of the hand shape is also added to the judgment material at the time of recognition. The user can freely determine the hand shape at the time of registration. For example, a blood vessel image with a thumb attached is registered while holding the hand. At the time of recognition, whether or not the hand is held is determined from, for example, the ratio of the length of the hand shape in the vertical and horizontal directions. Only when the similarity of the blood vessel image is equal to or greater than the threshold value and the hand is held is determined to be the person himself / herself.

2) The recognition accuracy is improved by recognizing a plurality of locations of a plurality of recognizers. A plurality of cameras are arranged to shoot a wide range. In particular, in a device of a type in which a hand is put in a hole, both sides of the hand can be photographed as described above. For example, the recognition method is determined when the person is determined to be the person at each location. Alternatively, it is determined that the person who has the highest sum of similarities.

  When photographing continuously in time, the recognition accuracy is improved by sequentially recognizing a plurality of positions in the hand. The judgment of which part of the hand is presented is determined from the hand shape on the image, for example. Furthermore, the order of presentation can also be used for recognition.

Other recognition accuracy improvement methods will be further described.
1) Calculation of three-dimensional shape A technology for calculating the three-dimensional shape of a hand from a captured image will be described as a technology that enables more accurate recognition. Since a technique for calculating the three-dimensional shape of an object from an image is a known technique, detailed description thereof is omitted. For example, see Document 6. A plurality of cameras are installed, and the three-dimensional shape of the hand is obtained by stereo measurement. Stereo measurement is performed by obtaining corresponding points from two images (FIG. 17).

  If the calculated three-dimensional shape is used, it is possible to perform more accurate correction processing of the captured image. The correction processing method is the same as described above. Further, a three-dimensional shape of the blood vessel is obtained by performing stereo measurement on the extracted blood vessel image. By performing similarity calculation in three dimensions, recognition with higher accuracy becomes possible. A three-dimensional shape can be calculated from motion parallax using a plurality of temporally continuous images instead of a plurality of cameras. Also, the three-dimensional shape can be calculated by switching the illumination or giving the illumination a specific pattern.

2) Combined use of visible light When photographing the palm of the hand, wrinkles (palm) on the surface of the palm are also captured. Since wrinkles can be copied (stolen) relatively easily, it is desirable not to use them for personal recognition. If combined with imaging with visible light, wrinkles and blood vessel images can be separated. Shooting with visible light is performed with another camera, for example. Alternatively, the visible light cut filter 20 of FIG. 4 may be partially removed. Since wrinkles appear only with visible light, wrinkles are removed by removing the detection results with visible light from the detection results with infrared light. The blood vessel image excluding wrinkles is registered and compared with the imaged blood vessel image excluding wrinkles. Alternatively, the amount of wrinkles removed is calculated, and when there are many wrinkles, a warning is displayed to the user to change how the hand opens.

  Next, a method for improving user convenience will be described. For example, in personal recognition in entry / exit management or the like, there is mainly a usage form in which a user first presents an ID to a recognition device with a card (IC card or the like), and then verifies the ID with a hand. When using a non-contact type IC card, there is no particular problem in operation. However, when using a contact type or insertion type card, the user must hold the hand again after operating the card. There is a problem that the operation becomes complicated. In the following, a method for solving this problem will be described.

1) Shoot the wrist or back of the hand while holding the card When using the card recognition mechanism that makes the card surface contact the card reader, the camera of the personal recognition device is installed under the card reader to photograph the wrist ( FIG. 18). Alternatively, a camera is installed next to the card reader and the back of the hand is photographed by rotating the hand after the card reading process. Thereby, individual recognition can be performed by one operation.

2) When using a card recognition mechanism that inserts the palm into the card reader after inserting the card, install the camera of the personal recognition device above the card insertion part or above the insertion part (high position) The palm is photographed after insertion (FIG. 19). The back of the hand may be photographed by rotating the hand. Thereby, individual recognition can be performed by one operation. Note that this method applies not only to the card recognition mechanism in which the card is ejected from the insertion section, but also to the card recognition mechanism in which the user passes the gate and receives the card at another place. Applicable.

  As another form of use of personal recognition devices, when managing attendance of meetings and classes, if personal recognition devices are installed at the entrance of a conference room or classroom, it is expected that people will be rushed and confused before the start of the conference . In the following, a method for solving this problem will be described.

  The personal recognition device storing the blood vessel images of all the prospective attendees is circulated during the meeting, and the personal recognition is sequentially performed. In order to make the attendee confirm that he / she has been registered for attendance, the circulation device preferably includes a recognition unit 18 and a display unit for displaying the result. As the personal recognition processing, verification is performed by obtaining a personal ID by presenting a card, or identification processing without a card is performed. The recognition result is stored in the personal recognition device, and after collecting the device, the data is transferred to the attendance management device and summed up. When an encryption function is incorporated in the personal recognition device, it may be transmitted to the attendance management device using a wireless communication mechanism every time recognition is performed. The encryption is performed, for example, in order not to reveal the attendees of important secret meetings.

  Although the details of the personal recognition apparatus of the present invention have been described above, this personal recognition apparatus can naturally be configured as a computer system including a photographing device. FIG. 20 is a block diagram showing the configuration of such a computer system, that is, a hardware environment.

  In FIG. 20, the computer system includes a central processing unit (CPU) 50, a read only memory (ROM) 51, a random access memory (RAM) 52, a communication interface 53, a storage device 54, an input / output device 55, and a portable storage medium reading device. 56, and a bus 57 to which all of them are connected.

  As the storage device 54, various types of storage devices such as a hard disk and a magnetic disk can be used, and the program shown in the flowchart of FIG. Programs such as detection and correction of blood vessel images are stored, and when such a program is executed by the CPU 50, personal recognition using a hand blood vessel image photographed in a non-contact manner in this embodiment becomes possible.

  Such a program is stored in, for example, the storage device 54 via the network 59 and the communication interface 53 from the program provider 58 side, or stored in a portable storage medium 60 that is commercially available and distributed, It can also be set in the reading device 56 and executed by the CPU 50. As the portable storage medium 60, various types of storage media such as a CD-ROM, a flexible disk, an optical disk, and a magneto-optical disk can be used, and a program stored in such a storage medium is read by the reading device 56. As a result, the personal recognition by the blood vessel image of the hand in this embodiment becomes possible.

  As described above in detail, according to the present invention, it is possible to provide a sanitary and clean personal recognition device with a small burden on the user by capturing a blood vessel image of a hand without contact. . In order to avoid degradation of recognition accuracy due to non-contact imaging, devise the device form and recognition method, for example, by detecting the position and orientation of the hand and correcting the blood vessel image according to the detection result This greatly contributes to the improvement of the practicality of the personal recognition device using the blood vessel image of the hand.

  As described above, the personal recognition device of the present invention requires verification and identification of individuals, such as entrance / exit management of facilities and attendance management of meetings, by using blood vessel images of hands taken without contact. It can be used in all industries.

1 is a block diagram illustrating the principle configuration of the present invention. It is a block diagram which shows the structure of the personal recognition apparatus in this embodiment. It is a flowchart of a personal recognition process. It is a figure which shows the structural example of an imaging | photography part. It is a figure which shows the example of a blood vessel image. It is an example of the memory content of a memory | storage part. It is a figure explaining the external appearance of an imaging device. It is explanatory drawing of the imaging | photography method using a some camera. It is a figure which shows the example which installs a transparent plate or a half mirror in the front surface of a camera. It is explanatory drawing of how to put in the hand for removing the influence of external light. It is explanatory drawing of utilization of a goodwill and a reflecting plate for removing the influence of external light. It is a figure explaining utilization of the some illumination for reducing the influence of external light. It is an example of the apparatus shape of the imaging device which is easy to carry. It is explanatory drawing of the detection method of the direction of a hand. It is explanatory drawing of the hand shape extraction process on an image. It is a figure explaining the comparison with the hand shape on an image, and the registered hand shape. It is a figure explaining the stereo measurement of a hand shape. It is a figure which shows the combination of a personal recognition apparatus and a contact-type card reader. It is a figure which shows the combination of a personal recognition apparatus and an insertion-type card reader. It is a figure explaining loading to the computer of the program in this embodiment.

Claims (5)

In the device that detects the blood vessel image of the hand and performs personal recognition,
Position orientation shape instructing means for instructing the user how to hold the hand, one or more irradiating means for irradiating the hand with near infrared light, and one or more photographing means for photographing an image by the near infrared light Equipped with a photographing device capable of photographing a blood vessel image without touching the hand,
A blood vessel image detecting means for detecting a blood vessel image from an image photographed by the one or more photographing means;
A blood vessel image storage means for storing a blood vessel image of a hand registered for each user;
Recognizing means for performing personal recognition by comparing the registered blood vessel image and the detected blood vessel image;
Each time the blood vessel image is detected, the detection means detects the position and / or orientation of the hand from the captured hand image, and the detection is performed in correspondence with the detected position and / or orientation of the hand. blood vessel image by correcting the, further comprising a correction unit when the correction is appropriate to provide an image after the correction to said recognition means, said position facing shape indicating means when the correction is inappropriate it characterized in that it notifies the user of personal recognition device. The detection means, individual recognition apparatus according to claim 1, wherein the detecting the orientation of the hand by the non-uniformity on the image of the incident light intensity to the imaging means. The detection means, corresponding to the shape of the hand on the image, the position of the hand, and / or personal recognition system according to claim 1, characterized in that it detects the orientation. The parameters to be registered corresponding to the shape of the front Kite, further comprising a hand shape storage means for storing for each individual's user,
4. The personal recognition apparatus according to claim 3 , wherein the detecting means compares the registered parameter with a hand shape parameter on the image to determine a hand orientation. The detection means detects the orientation of the hand by approximating the hand in a plane;
It said correction means, individual recognition apparatus according to claim 1, wherein the correcting the blood vessel image using parameters representing the plane.

Images