JP2016096987A - Biometric authentication device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a technology which can perform highly accurate authentication by obtaining information on a plurality of biological tissues overlapping one another in a living body.SOLUTION: A biometric authentication device includes: a placement area on which a finger is placed; at least one light source 3 for irradiating the finger with light; imaging device 9 for imaging the light from the light source 3 to obtain at least one image; an authentication processing part 10 for processing the image; and a storage device 14 for storing registration information on at least one of a plurality of biological characteristics overlapping one another in the finger. The authentication processing part 10 obtains information on the distribution of pigment concentration related to a plurality of biological characteristics overlapping one another in a finger 1 from an image of the finger, obtains inputted information on at least one biological characteristic from the information on the distribution of pigment concentration, and uses the inputted information and the registration information and calculates a degree of the similarity to perform authentication based on the degree of the similarity.SELECTED DRAWING: Figure 1A

Description

  The present invention relates to a biometric authentication apparatus that authenticates an individual using a biometric.

  Among various biometric authentication techniques, finger vein authentication is known as a technique that can realize highly accurate authentication. Finger vein authentication achieves excellent authentication accuracy using a blood vessel pattern inside the finger. Since finger vein authentication is more difficult to counterfeit and tamper than fingerprint authentication, a high level of security can be realized.

  In recent years, there have been an increasing number of cases in which biometric authentication devices are installed in devices such as mobile phones, notebook PCs (Personal Computers), mobile terminals such as smartphones and tablet terminals, lockers, safes, and printers to ensure the security of each device. ing.

  In addition, in addition to entrance / exit management, attendance management, log-in to computers, etc., biometric authentication has been used in recent years as a field to which biometric authentication is applied. In particular, it is important for a biometric authentication device used in the public to realize reliable personal authentication. Furthermore, in view of the recent spread of tablet-type mobile terminals and the trend of wearable computing, it is also an important requirement to realize downsizing of the apparatus while ensuring convenience as described above.

  Patent Document 1 uses a tendency that light is absorbed in blood vessels and reflected or scattered in adipose tissue, and obtains three-dimensional digital information of a subcutaneous tissue structure photographed by focusing on a sensor array through a lens. A technique for identifying an individual from a photographing apparatus and a grayscale image of a picture photographed using the apparatus is disclosed.

  Non-Patent Document 1 discloses, as a conventional technique, a technique for irradiating skin with white visible light, photographing the reflected light with a color camera, and independently acquiring the distribution of hemoglobin and melanin in the skin. ing.

JP 2007-044532 A

`` Image-based skin color and texture analysis / synthesis by extracting hemoglobin and melanin information in the skin '', in Proc. Of SIGGRAPH, ACM Press, 770-779, 2003.

  In order to realize a small-sized, easy-to-use and highly accurate biometric authentication device, it is important to extract biometric features effective for personal authentication from a narrow area.

  Patent Document 1 discloses that shading in an image represents various structural features related to the subcutaneous tissue structure, with dark pixels corresponding to blood vessels and light pixels corresponding to adipose tissue. In Patent Document 1, the image processor classifies each pixel into a pixel having blood and a pixel not having blood, and uses adipose tissue around the vein for identification.

  However, in Patent Document 1, it is described that a portion other than the vein is a portion corresponding to the adipose tissue with respect to the imaging of the adipose tissue, and thus it is not considered that the vein and the adipose tissue are superimposed. . Therefore, with the technique of Patent Document 1, it is not possible to perform personal authentication in consideration of a plurality of living tissues that are superimposed.

  Further, in the technique described in Non-Patent Document 1, while it is possible to separate hemoglobin and melanin by irradiating the skin with white reflected light, disclosure and suggestion regarding personal authentication based on skin tissue Not done.

  Therefore, the present invention provides a technique that can perform authentication with high accuracy by acquiring information related to a plurality of living tissues that are superimposed.

  In order to solve the above problems, for example, the configuration described in the claims is adopted. The present application includes a plurality of means for solving the above-described problems. To give an example, a placement area on which a finger is placed, at least one light source that irradiates light on the finger, and a light source from the light source. An imaging unit that captures light to acquire at least one image, an authentication unit that processes the image, and a storage unit that stores registration information regarding at least one of a plurality of biometric features superimposed on the inside of the finger; Are provided. In the biometric authentication apparatus, the authentication unit acquires information on a pigment concentration distribution related to a plurality of biometric features superimposed on the inside of the finger from the image, and at least one biometric feature from the information on the pigment concentration distribution. Input information is acquired, a similarity is calculated using the input information and the registration information, and authentication is performed based on the similarity.

  ADVANTAGE OF THE INVENTION According to this invention, the biometric authentication apparatus which can authenticate with small size and high precision can be provided by acquiring the information regarding the some biological tissue which overlaps and exists. Further features related to the present invention will become apparent from the description of the present specification and the accompanying drawings. Further, problems, configurations and effects other than those described above will be clarified by the description of the following examples.

It is a figure which shows the structure of the whole biometric authentication system of a 1st Example. It is a figure which shows the structure of the authentication process part of the biometrics system of a 1st Example. It is sectional drawing of the input device of the biometrics authentication system of a 1st Example. It is the figure which looked at the input device of the biometrics system of the 1st example from the upper surface. It is a schematic diagram of the skin cross section for demonstrating the biological tissue of the body surface vicinity. It is an example of the processing flow which measures a fat pattern using the input device of the biometrics authentication system of a 1st Example. It is a schematic diagram for demonstrating the method of calculating the density | concentration of a biological tissue from color space. FIG. 5B is a schematic diagram for explaining a technique for calculating the concentration of a biological tissue from a color space, and is a partial enlarged view of FIG. 5A. It is a schematic diagram for demonstrating the method of calculating the density | concentration of a biological tissue from color space. It is explanatory drawing of the method of estimating the color vector of a biological tissue. It is explanatory drawing of the method of estimating the color vector of a biological tissue. It is explanatory drawing of the method of estimating the color vector of a biological tissue. It is explanatory drawing of the method of estimating the color vector of a biological tissue. It is a figure explaining the change of a fat crest when the way of pressing a finger is weak. It is a figure explaining the change of a fat crest when the way of pressing a finger is strong. It is a flowchart explaining one Example of the registration procedure of the registration data in the biometric authentication system of a 1st Example. It is a flowchart explaining one Example of the collation procedure of the information and registration data which were obtained from the user in the biometrics system of a 1st Example. It is an example of the registration data used with the method of template collation of the fat crest by weighting. It is an example of the input data used with the method of template collation of the fat crest by weighting. It is an example of the weight table used with the method of template collation of the fat crest by weighting. It is one Example of the biometrics authentication system using a fat crest and a finger vein which is a 2nd Example. It is a processing flow of one Example of the authentication using the biometric authentication system of a 2nd Example. It is a schematic diagram of the image | photographed finger vein pattern. It is the schematic diagram of the pattern of the image | photographed fat pattern. It is the schematic diagram of the image | photographed finger vein and the pattern of a fat crest. It is a schematic diagram of a finger vein pattern image used in a feature point matching method that combines fat prints and feature amounts of finger veins. It is a schematic diagram of a fatprint image used in a feature point matching method combining a feature quantity of a fatprint and a finger vein. It is a schematic diagram which shows extraction of the feature point in the image of FIG. 14A. It is a schematic diagram which shows extraction of the feature point in the image of FIG. 14B. 14C is a schematic diagram of a feature point composite image obtained by superimposing the image of FIG. 14C and the image of FIG. 14D. It is a figure explaining one Example of the collation process which collates a fat crest and a finger vein independently, and performs a position correction complementarily. It is a figure explaining the process which codes and authenticates the overlap part of a finger vein and a joint pattern. It is a figure explaining the process which codes and authenticates the overlap part of a finger vein and a joint pattern. It is a figure explaining the process which codes and authenticates the overlap part of a finger vein and a joint pattern. It is one Example of the biometric authentication apparatus which comprises the visible light source for reflected light irradiation and the color camera which are 3rd Examples in the portable terminal. It is one Example of the biometric authentication apparatus which comprises the visible light source for reflected light irradiation and the color camera which are 3rd Examples in the portable terminal. It is an example of operation of the fingertip authentication apparatus for smartphones which authenticates by tapping a finger. It is an example of operation of the fingertip authentication apparatus for smartphones which authenticates by tapping a finger.

  Embodiments of the present invention will be described below with reference to the accompanying drawings. The accompanying drawings show specific embodiments in accordance with the principle of the present invention, but these are for the understanding of the present invention, and are never used to interpret the present invention in a limited manner. is not. In each drawing, the same reference numerals are assigned to common components.

[First embodiment]
FIG. 1A is a diagram illustrating an overall configuration of a biometric authentication system using a finger blood vessel according to a first embodiment. Needless to say, the present invention may be configured not as a system but as an apparatus in which all or a part of the configuration is mounted on a housing. The device may be a biometric authentication device including an authentication process. The authentication process may be performed by an external device, and the device may be a blood vessel image acquisition device or a blood vessel image extraction device specialized for acquiring blood vessel images. Therefore, the functions described below may be configured by a plurality of devices existing on the network. The device may also be implemented as a form terminal as described below.

  The biometric authentication system includes an input device 2, an authentication processing unit 10, a storage device 14, a display unit 15, an input unit 16, a speaker 17, and an image input unit 18.

  The input device 2 is a device that acquires an image including a biometric feature from the authentication subject's biometric. The input device 2 includes a light source 3 installed in a housing and an imaging device 9 installed in the housing. The image input unit 18 acquires an image captured by the imaging device 9 of the input device 2 and inputs the acquired image to the authentication processing unit 10. The authentication processing unit 10 performs image processing on the image input from the image input unit 18 and executes authentication processing.

  The image processing function portion of the authentication processing unit 10 or the image processing function including the image input unit 18 may be referred to as an image processing unit. In any case, the authentication processing unit 10 has an image processing function.

  The light source 3 is a light emitting element such as an LED (Light Emitting Diode), for example, and irradiates light on the finger 1 presented on the input device 2. The imaging device 9 captures an image of the finger 1 presented on the input device 2.

  As shown in FIG. 1A, the authentication processing unit 10 includes a central processing unit (CPU) 11, a memory 12, and various interfaces (IF) 13. The CPU 11 performs various processes by executing a program stored in the memory 12. The memory 12 stores a program executed by the CPU. The memory 12 temporarily stores the image input from the image input unit 18.

  FIG. 1B shows a functional block diagram of the authentication processing unit 10. The authentication processing unit 10 includes an authentication unit 10a and a registration unit 10b. The authentication unit 10a compares the input data input from the image input unit 18 with the registration data registered in the storage device 14, and authenticates the user. The registration unit 10 b creates registration data from the image acquired by the input device 2 and stores it in the storage device 14.

  Each processing unit of the above-described authentication processing unit 10 can be realized by various programs. For example, various programs stored in the storage device 14 are expanded in the memory 12. The CPU 11 executes a program loaded on the memory 12. The CPU 11 executes processes and operations mainly performed by the authentication processing unit 10 described below.

  The interface 13 connects the authentication processing unit 10 and an external device. Specifically, the interface 13 is connected to the input device 2, the storage device 14, the display unit 15, the input unit 16, the speaker 17, the image input unit 18, and the like.

  The storage device 14 stores user registration data in advance. The registration data is information for collating users, and is, for example, an image of a finger vein pattern. Usually, a finger vein pattern image is an image obtained by capturing blood vessels (finger veins) distributed mainly under the skin on the palm side of a finger as a dark shadow pattern.

  The display unit 15 is, for example, a liquid crystal display, and is an output device that displays information received from the authentication processing unit 10. The input unit 16 is, for example, a keyboard, and transmits information input from the user to the authentication processing unit 10. The speaker 17 is an output device that transmits information received from the authentication processing unit 10 as an acoustic signal (for example, voice).

  FIG. 2A is a cross-sectional view illustrating the structure of the input device 2 of the biometric authentication system according to the first embodiment. FIG. 2B is a view of the input device 2 as viewed from above. The input device 2 captures a biometric feature included in the multilayer structure of the fingertip skin. The input device 2 includes a casing 21 in which various devices are installed. As shown in FIG. 2B, the housing 21 has a cylindrical shape, and the housing 21 has an opening 22.

  A transparent acrylic plate 23 is fitted into the opening 22, and the fingertip of the finger 1 is placed thereon. The transparent acrylic plate 23 becomes a placement area for the finger 1 described below.

  The light source 3 is provided inside the housing 21 and can project visible light toward the finger 1 through the opening 22. It is assumed that the light source 3 in the present embodiment is an LED, and visible light of three types of wavelengths of blue, green, and red can be irradiated with arbitrary intensities from one LED element. As an example of a specific wavelength, a wavelength of 450 nm for blue, 550 nm for green, and 620 nm for red is selected. However, the LED elements of the respective wavelengths may not be integrated. Moreover, although it is good also as a white light source with a continuous spectrum including these wavelength bands, it is desirable that the light quantity can be individually adjusted for each wavelength.

  In the present embodiment, four light sources 3 are arranged around the imaging device 9 and can uniformly illuminate the subject (fingertip) through the opening 22. In addition, when the light emitted from the light source 3 is reflected on the acrylic plate 23, the subject cannot be seen. Therefore, the four light sources 3 shown here are lighted in time series to continuously photograph the subject, and each image is captured by HDR (High Dynamic Range). ) The image may be synthesized by a technique to obtain a clear subject having no reflection component.

  The imaging device 9 is assumed to be a color camera and has a plurality of light receiving elements having sensitivity in visible and infrared wavelength bands. The color camera 9 has, for example, three types of CMOS or CCD elements having sensitivity to blue (B), green (G), and red (R), and these are arranged in a grid for each pixel of the image. In addition, a light receiving element having sensitivity to red has sensitivity to near infrared. The color camera 9 has three light receiving sensors having different peak wavelengths of light receiving sensitivity. The light receiving sensitivity of each light receiving sensor is composed of sensors with light receiving sensitivity peaks, for example, around 450 nm for blue, 550 nm for green, and 620 nm for red, and the space of light emitted from the living body based on the sensitivity of each wavelength. Color distribution can be obtained. Further, it is assumed that the color image to be shot is in an RGB color image format in which each RGB color plane can be acquired independently. Note that the imaging device 9 may be a multispectral camera having more than three wavelengths. In addition, the imaging device 9 is provided with a filter that transmits all wavelengths of light output from the light source 3 and blocks other bands, thereby blocking unnecessary stray light and improving image quality.

  With the above configuration, the input device 2 can shoot various biological features existing in the skin of the fingertip by irradiating a plurality of visible lights. Here, the living tissue in the skin will be described in detail.

  FIG. 3 is a schematic view of a skin cross section for explaining a living tissue near the body surface. The skin has a multi-layered structure, and is roughly classified into an epidermis 34, a dermis layer 35, and a subcutaneous tissue 36. The epidermis 34 is the outermost layer, and there are irregularities on the surface, which are observed as wrinkles on the skin. In particular, the unevenness at the fingertip is known as a fingerprint 31. In addition, there are wrinkles called joint prints and palm prints in the joints and palms of the fingers, but the epidermis of this part is thin, and the red color of the dermis and deeper blood is exhibited. Further, on the back side of the epidermis 34, there is a cellular tissue that generates melanin, and the skin color such as white, yellow, brown, etc. is formed, and an uneven pattern such as a blotch or a mole is formed.

  A layer existing behind the epidermis 34 is a dermis layer 35. The dermis layer 35 is formed of tissues such as fibroblasts, collagen, and elastin, and sebum lines, sweat glands, nerves, capillaries 32 and veins 30 are distributed therein, and hair roots also exist depending on the location. . The reddish color of the skin is due to blood carried by the capillaries 32 of the dermis layer 35.

  The deepest layer is the subcutaneous tissue 36. In the subcutaneous tissue 36, a relatively thick vein 30 and artery are present, and subcutaneous fat is present. Subcutaneous fat has a leaflet structure in which fat cells form one lump, and this lump is called a fat leaflet 33. The adipose tissue has a structure in which the fat leaflets 33 are stacked in multiple layers, and microvessels and veins 30 run through the gaps. That is, no blood exists in the portion where the fat leaflet 33 exists, and the blood is distributed only in the gap.

  In general, a vein pattern is observed as a dark shadow in an image captured by transmitting or reflecting infrared rays with respect to the skin. This is because the amount of light in the portion of the vein 30 is reduced because infrared rays are strongly absorbed by hemoglobin in the blood. On the other hand, capillaries 32 exist in layers shallower than the veins 30, and light absorption by the capillaries 32 actually occurs. Nevertheless, the pattern of the capillary vessel 32 is hardly observed, and only the pattern of the vein 30 is observed. This is because the vein 30 is relatively thicker than the capillaries 32 and the light absorption rate is accordingly larger than that of the capillaries 32.

  In contrast, skin images captured in color by transmitting or reflecting white light or other visible light include shadows due to irregularities on the surface of the skin, white or yellow-brown melanin in the epidermis, blood in dermal capillaries. Red and the blue or gray of the veins 30 present in the subcutaneous tissue 36 are observed in a state in which colors exhibited by various biological tissues existing inside and outside the skin are superimposed. In particular, as for the color of the shallow skin layer from the epidermis to the dermis, the portion where blood is largely distributed is red, and the other portion is white or yellow because the melanin pigment is dominant.

  Especially in the fingers and palms, the skin color due to blood and melanin is not uniform, and is observed as a non-uniform mottled pattern. This is mainly due to the uneven distribution of blood. There are two causes of non-uniform blood distribution: non-uniformity of blood vessel running and non-uniform thickness of the layer infiltrated with blood. Regarding the former, the capillaries 32 in the dermis are extremely thin, and the pattern is hardly observed due to the influence of scattering of irradiation light, and is observed as a substantially uniform red color. Therefore, the latter effect is the main factor.

  One possible reason for the unevenness of the blood infiltrating layer is the effect of the fat leaflets 33. As described above, the fat lobule 33 present in the subcutaneous tissue 36 is a mass of fat and does not contain hemoglobin in the inside thereof, so that strong light absorption does not occur. On the other hand, since the space between the fat leaflets 33 is filled with blood, it contains hemoglobin. Therefore, the portion where the fat leaflets 33 are present only absorbs light by the dermal capillaries, and the portion of the gap between the fat leaflets 33 absorbs light by the blood layers of both the dermis layer 35 and the subcutaneous tissue 36. Therefore, the uneven distribution of the fat leaflets 33 is observed as a difference in blood absorption rate. In addition, the fat leaflets 33 press the dermis layer 35 outward from the back of the skin, and the blood volume in the dermis where the fat leaflets 33 are present is also reduced. Similarly, the amount of light absorbed by the blood in the portion where the fat leaflets 33 are present is reduced, and as a result, it is observed as a non-uniform pattern. By combining these two factors, the pattern of the fat leaflet 33 is observed as the contrast between hemoglobin and melanin. Here, this contrast is called a fat pattern.

  The fat pattern is observed not only in a non-contact state where nothing touches the skin surface but also in a contact state. For example, when the fingertip is pressed against a transparent acrylic plate, the amount of blood changes due to the pressure applied, so the color pattern of blood and melanin changes, but the position of the fat leaflet 33 does not change, so Reproducibility is maintained. Therefore, if it is possible to absorb the change in blood flow due to the pressure force, it is possible to determine whether the patterns are the same.

  As described above, in the skin, biological features superimposed in multiple layers such as melanin distribution such as fingerprints and moles, fat prints, veins, and finger joints can be observed. These biometric features vary from those with high discrimination ability to those with low identification ability. However, if at least the recognition success rate of a single unit exceeds 0.5, authentication accuracy can be improved by fusing a plurality of biometric features. An authentication method that utilizes such multilayer biometric information for personal authentication is referred to as multi-layer biometric authentication. The advantages of this method are: (1) accuracy can be increased because a large number of different biological information can be acquired at the same time, and (2) it can be used as information for position correction because it is superimposed on each other, and further by utilizing the features of overlap. The accuracy is further increased, (3) fat prints and the like are dense feature amounts, and it is easy to realize downsizing of the apparatus, and (4) forgery becomes more difficult.

  Here, an embodiment of a measurement technique for acquiring fatprint information will be described. FIG. 4 is an example of a processing flow for measuring a fat pattern using the input device 2 shown in FIGS. 2A and 2B.

  First, it is detected that the user places the finger 1 on the placement area of the input device 2 (S401). As this detection method, a touch sensor may be provided in the opening 22 of the input device 2. As another detection method, when the light source 3 is blinked and an image is shot, and the brightness value of the image is repeated in synchronization with the blinking of the light source 3, it is determined that the subject exists above the apparatus. Also good.

  Next, it is determined whether the finger 1 is in the placement area of the input device 2 (S402). If it is determined that the finger 1 is present, the blue (green) and red (red) wavelengths of the LED (light source 3) are determined. The irradiation intensity of the light source 3 is adjusted while irradiating the finger with light (S403). Since the luminance value of the image may be saturated depending on the relationship between the camera sensitivity and the output intensity of the light source 3, irradiation and photographing are repeated so that the average luminance value of the image becomes substantially constant at each wavelength to converge the light amount value. When the light intensity values of the respective wavelengths converge, an image of each wavelength at that time is taken and a color image of the finger surface is acquired (S404). The image acquired by the input device 2 is input to the authentication processing unit 10 via the image input unit 18.

  The color information of the surface of the living body taken as a color image or the internal tissue in the vicinity of the living body surface includes information on the epidermis 34 such as unevenness of the skin surface such as fingerprints and wrinkles of the joints, as well as melanin pigment distributed in the epidermis 34, Includes reflection images of living tissue having various color characteristics such as capillaries 32 existing in the dermis layer 35 and fat leaflets 33 and veins 30 existing in the subcutaneous tissue 36 in the back thereof. ing. For example, on the ventral side of the finger, a red blood color caused by hemoglobin and a black pattern such as white, yellow, or mole caused by melanin are seen, and the relatively thick distributed in the dermis layer 35 or the subcutaneous tissue 36. The vein 30 appears blue or gray. In this way, by providing multi-wavelength visible reflected light, color information of various biological tissues can be superimposed and observed as one reflected image.

  Next, the authentication unit 10a of the authentication processing unit 10 measures the distribution of pigment concentrations of melanin and hemoglobin using the photographed color image (S405). Here, the reason for measuring the pigment amount of melanin and hemoglobin in order to acquire fatprint information will be described. Biological features such as fingerprints and veins that can be used for personal authentication are composed of basic biological tissues such as hemoglobin and melanin. In addition, depending on the position where the living tissue exists, the thickness of the living tissue, and the like, the body features are different parts. For example, the capillaries 32 of the dermis layer 35 and the veins 30 of the subcutaneous tissue 36 are both formed of hemoglobin, but the capillaries 32 of the dermis layer 35 are red and the veins 30 of the subcutaneous tissue 36 appear blue, so that the biological features are You can see that they are different. Further, the joint crest is a portion where the epidermis is thin and reddish, and basically has the same hemoglobin color as the capillaries 32 of the dermis layer 35, but the thickness of the epidermis (unevenness) compared to other skin portions. Thin). Therefore, the joint pattern is regarded as a biological feature different from the capillary 32 of the dermis layer 35 in that the amount of melanin is small. Thus, it is important to analyze the intensity of the color of hemoglobin and melanin, especially when photographing and extracting biological features of the skin, and by measuring the pigment concentration distribution of melanin and hemoglobin, It is possible to obtain information in consideration of a state in which colors exhibited by a plurality of internal biological tissues are superimposed.

  Therefore, in this embodiment, the spatial density distributions of the melanin of the epidermis and the hemoglobin of the dermis are obtained independently from the color information of the reflected image, and the fat pattern image is reconstructed with high definition based on them. In the following example, the distribution of pigment concentration of melanin and hemoglobin is measured as the measurement of fat crest, but as described above, the measurement of pigment concentration of melanin and hemoglobin is performed using other biological features such as fingerprints and moles. It can also be applied to the measurement of stains, veins, joint prints, etc.

An example of a technique for independently acquiring the melanin concentration and the hemoglobin concentration is as follows. First, according to the Modified Lambert-beer rule, when the hemoglobin concentration at the image coordinates (x, y) is ρ _h (x, y) and the melanin concentration is ρ _m (x, y), these and the color information of the image It is known that the relationship with C ^log (x, y) is expressed by the following equation (1).

C ^log (x, y) = -ρ _m (x, y) σ _m l _m -ρ _h (x, y) σ _h l _h -p ^log (x, y) 1 + e ^log ... (1)

Here, C ^log (x, y) is a value obtained by logarithmizing the pixel value of each color element of the RGB color image at the image coordinates (x, y) as a three-dimensional vector, σ _h, σ _m Are the color vectors of hemoglobin and melanin, l _h and l _m are the average distance traveled by the light that reaches and reflects the layer where hemoglobin and melanin are present, respectively, and p ^log (x, y) is in coordinates (x, y) The intensity of the shadow, 1 is a three-dimensional unit vector having a norm of 1, and e ^log is the intensity of the irradiation light. The color vector is a vector of RGB luminance values obtained when a biological tissue of unit density is reflected and taken. For example, if the biological tissue is completely red, only the element indicating red in the three-dimensional vector is used. The blue and green elements are zero. According to this equation, it is shown that the luminance value of the observed image can be expressed by a linear combination of the hemoglobin concentration and the melanin concentration by taking the logarithm thereof.

  Here, a method for estimating the concentrations of hemoglobin and melanin from Equation (1) will be described. The assumption is that the color vectors of hemoglobin and melanin are known, and the average distance traveled by light reaching hemoglobin and melanin is the sum of the average dermis and epidermis layer thickness, the thickness of the epidermis layer, and the bias term If ignored, there are three unknowns: hemoglobin and melanin concentrations and shadow information. At this time, since the color space of the photographed image is RGB three-dimensional, three equations can be obtained by photographing the color image. Therefore, by solving these as simultaneous equations, three unknowns of hemoglobin, melanin concentration, and shadow information can be calculated for each pixel. However, since the bias term is ignored, the relative density distribution information is obtained.

  From the above, information on the relative distribution of the hemoglobin and melanin concentrations, that is, these patterns can be obtained. In addition, since it is possible to obtain shading information, it is possible to obtain a fingerprint pattern that can be observed by light attenuation or diffusion due to unevenness on the skin surface. However, since the spatial pitch of the fingerprint pattern is almost constant, it is not necessary to acquire the pattern from the above shadow parameters, and it is acquired by emphasizing by applying an edge enhancement filter such as a Gabor filter. May be.

5A to 5C are schematic diagrams for explaining the equation (1) for calculating the concentration of the biological tissue from the color space. In these figures, it is shown that the three-dimensional space that is the logarithm of the luminance value of the RGB observation image and the pixel value acquired as the luminance value of a certain coordinate are established by the relationship of Expression (1). In FIG. 5A, the pixel value C ^log of a certain pixel is plotted as one point in a three-dimensional space. Here, it is assumed that σ _{h and} σ _m are as shown in FIG. 5B. Here, when projected onto the plane formed by σ _h and σ _m along one vector indicating the direction of the shadow, the relative value of the hemoglobin concentration ρ _{h and} the relative value of the melanin concentration ρ _m can be read as shown in FIG. 5C. it can.

In the above method, the color vectors σ _h and σ _m of hemoglobin and melanin included in the equation (1) are assumed to be known. These values are determined by the light absorption / scattering characteristics of each living body, the wavelength of the light source, and the spectral sensitivity characteristics of the sensor. Therefore, it can be handled as a known parameter by examining in advance the light absorption / scattering characteristics of the living body and the characteristics of the camera.

  However, considering practical aspects, it is difficult to determine the color vector parameters after investigating the camera sensitivity and the color characteristics of the living body in advance. Therefore, it is assumed that the concentration distribution of hemoglobin and melanin is independent of each other, and independent component analysis (ICA) that performs blind signal separation without knowing both the color vector and concentration distribution of each biological tissue. A calculation method based on is proposed, and this method may be used. On the other hand, since this method is premised on that there is no correlation between the two concentration distributions, and there is no guarantee that a physically correct separation will be performed, the method is limited.

  Therefore, in this embodiment, color vectors are estimated based on actual measurement values of the living body using fingertip contact. 6A to 6D are explanatory diagrams of a method for estimating a color vector of a living tissue from changes in pixel values of a large number of coordinates in the color space shown in FIGS. 5A to 5C described above.

  When the fingertip is pressed against the acrylic plate 23 in the input device 2 of FIG. 2A, the skin surface that normally exhibits red color changes to yellow or white due to melanin. This occurs because only the red color exhibited by hemoglobin becomes thin because the blood volume is reduced by the pressure of contact, but melanin does not decrease even when pressure is applied. That is, when the change amount of the color information when the acrylic plate 23 is pressed with the fingertip is observed, the color vector possessed by hemoglobin can be calculated from the change amount.

FIG. 6A shows a change in the pixel value 60 before pressing the finger and the pixel value 61 after pressing at the same coordinates. By pressing, the blood flow is reduced, and it can be seen that many pixel values move in the same direction on the color space. Since this moving direction is affected by the reduction of hemoglobin, the color vector σ _h of hemoglobin can be estimated by obtaining an average value in the moving direction as shown in FIG. 6B.

Since the color vector of hemoglobin is known, the color vector σ _{m of} melanin, which is another biological tissue, is estimated. As one method, there is a method in which a predetermined ratio is extracted from the top of pixels with a large distance that fluctuates in the color space before and after pressing, and the result of linear approximation of the pixel value after pressing the data is directly used as a melanin color vector. is there. This method is an estimation method based on the premise that there is a place where hemoglobin has almost disappeared due to fingertip compression, and although there may be an error, a practically sufficient approximate solution can be obtained. FIG. 6C shows a state in which only the upper half of pixel values having large fluctuations are left, and FIG. 6D shows a state in which an approximate straight line is obtained from the pixel value 61 after pressing and a melanin color vector σ _m is obtained. Yes. Thus, since the color vectors of hemoglobin and melanin can be estimated, the respective concentrations can be calculated.

  Since the above-described estimation of the hemoglobin color feature amount uses a physical change, it is possible to perform estimation with higher reliability than the conventional ICA-based method. In addition, since the estimation method demonstrated above can be used if the variation | change_quantity of the color information when pressing the acrylic board 23 with a fingertip can be observed, when the fingertip is made to contact or press the acrylic board 23 of the input device 2 The images in at least two different states may be acquired.

  As another method for estimating the color vector of the biological tissue, an estimation method based on the shape of the biological feature can also be used. In this method, a living body part is estimated from the shape of the living body, and the color vector of the living body tissue is estimated from prior information regarding the color information of the living body part and the density of the living body tissue of a general living body part. For example, joint wrinkles found in the finger joints have a thin epidermis, so that the red color caused by hemoglobin is stronger than the color of the surrounding epidermis, and the color caused by melanin is light. Therefore, it can be said that only the color characteristic of hemoglobin appears strongly and the color characteristic of melanin is small in the wrinkled part of the finger. Therefore, if the finger region is specified from the shape of the contour of the finger and the position is specified by using a reddish line perpendicular to the finger direction as the joint wrinkle, the color information of the joint wrinkle portion is directly used as the color vector of the hemoglobin. it can. As described above, the authentication unit 10a of the authentication processing unit 10 may specify a specific part of the finger from the color information in the image, and may regard the color information of the part as a color vector of a certain pigment. Based on this information, the authentication unit 10a of the authentication processing unit 10 may obtain information on hemoglobin concentration distribution.

  Since this method can be performed without pressing the living body like the above-described method, it has an advantage that it can be used when realizing non-contact biometric authentication. Note that the melanin color feature amount may be estimated by the same method as ICA, assuming that the hemoglobin color feature amount is known, as described above.

  From the color characteristics of the biological tissue acquired as described above, information on the proportion of hemoglobin and melanin present in each pixel can be obtained.

  Returning to the description of FIG. 4, finally, the authentication unit 10 a of the authentication processing unit 10 reconstructs a fatprint image (S <b> 406). In this example, a fatprint image is obtained as a biometric characteristic peculiar to an individual from the amount of acquired biomedical tissue.

  The pattern of hemoglobin and melanin obtained above can be estimated independently even when the living organisms are observed in multiple layers. Therefore, for example, a melanin pattern distributed over the capillary vessel 32 pattern is also obtained. Therefore, in the case of a biological feature to be originally acquired, in this case, a fat pattern, the distribution of hemoglobin and melanin can be directly acquired as fat pattern information, and an image reconstructed based on this information can be used for authentication. This is because, as described above, the distribution of hemoglobin and melanin becomes non-uniform depending on the portion where the fat leaflets 33 are present, and the pattern of the distribution can be used as a fat pattern. Similarly, with regard to the joint pattern, the hemoglobin distribution can be used as it is as the joint pattern information from the viewpoint that the amount of melanin in this portion is small. The joint crest here means a part of a finger joint, and a part where the epidermis 34 is thin and reddish.

  However, if the purpose is to distinguish and extract fat prints and joint prints and use them for authentication at the same time, the fat prints and joint prints can be distinguished based on prior knowledge that the melanin distribution of the joint print is less than the surroundings. . That is, among the patterns composed of portions having a high hemoglobin concentration distribution, a portion having a relatively high melanin concentration is distinguished as a fat pattern and a portion having a low concentration as a joint pattern. If the composition ratio of the basic biological tissue can be grasped in advance, not only for fat and joint patterns, but also for other biological features such as moles and spots, the living body can be determined based on the concentration of the basic biological tissue. The certainty that is a feature can be obtained, and an image reconstruction of a biometric feature with higher contrast can be realized. As described above, the authentication unit 10a of the authentication processing unit 10 uses the information on the hemoglobin and melanin concentration distribution measured in S405 to distinguish and extract a plurality of biological features (in this case, a fat pattern and a joint pattern) to obtain an image. May be reconfigured.

  Alternatively, if only the contrast between hemoglobin and melanin is to be photographed, blue or green visible light may be irradiated and the reflected light or transmitted light may be photographed by the color camera 9. The hemoglobin that has a red color in the dermis absorbs light strongly at the wavelength of green or blue, and the absorption rate decreases relatively in yellow and red, so the observation of the pattern of hemoglobin that looks red with visible light is Green or blue irradiation is sufficient. However, observation with transmitted light is difficult to transmit due to the relationship between the thickness of the living body and the light intensity, and it is necessary to irradiate with strong light except for a relatively thin part such as a webbed part or an ear at the base of a finger. On the other hand, it is possible to observe with a relatively weak light quantity when photographing with reflected light. However, there are many cases where the contrast of the obtained image is low, and by applying an enhancement filter such as an unsharp mask, biological features related to the capillary vessel 32 such as a fat pattern or a joint pattern can be clearly acquired. Similarly, the vein 30 existing in the deep part can be observed by irradiating the red wavelength. This is because short-wavelength light such as blue and green cannot reach deep veins.

  In this way, it is possible to observe a clearer biological tissue by separating each biological tissue that was originally mixed by image analysis of multiple wavelengths, but as shown here, it is monochromatic according to the biological tissue to be observed. Even if only visible light is irradiated, it is possible to shoot with constant image quality. This method eliminates the need for a complicated image reconstruction technique, so that the processing cost can be reduced.

  Next, a method for accurately collating biometric features captured by the above-described method using the input device 2 shown in FIGS. 2A and 2B will be described. Since the input device 2 is based on the premise that the fingertip is placed on the placement area (transparent acrylic plate 23), the fat pattern fluctuates due to the pressure on the fingertip. Therefore, it is indispensable for high accuracy to perform a collation process robust to this variation.

  FIG. 7A and FIG. 7B are diagrams for explaining a change in fat pattern due to a difference in pressing method. FIG. 7A is a diagram showing a pattern of fat prints when the finger 1 is in light contact with the placement area of the input device 2. In FIG. 7A, since the finger 1 is weakly pressed and blood disappears little, a portion of the blood pattern 71 in the fat pattern reflected in the fingertip image 70 can be observed relatively large.

  On the other hand, as shown in FIG. 7B, when the pressing force is stronger than that in FIG. 7A, the blood pattern 71 partially disappears and is divided into a blood pattern 72 that has not disappeared and a blood pattern 73 that has disappeared. The reason is that the capillaries 32 and the capillaries 32 that disappear easily even if pressure is applied to the capillaries 32 and the capillaries 32 that easily disappear due to the relationship between the position, shape, size, depth, etc. of the capillaries 32 and the fat leaflets 33. This is because of the distribution. On the other hand, the shape of the capillaries 32 and the fat leaflets 33 themselves does not change even when pressure is applied. Therefore, when the pressure is changed by repeatedly pressing the fingertip, the blood pattern 71 that has been visible until then disappears or conversely disappears. The blood pattern 73 may be observed again. In this way, the pattern itself is reproduced, but varies depending on how it is placed. In addition, depending on factors such as the position and shape of the fat leaflet 33, it is considered that there are places where blood patterns are always observed, and there are places where blood patterns tend to disappear even with light pressure. That is, it becomes possible to quantify the probability distribution in which a blood pattern is observed.

  Therefore, in this embodiment, the blood pattern observation probability distribution is calculated by taking statistics of the place where the blood distribution can be observed and the place where the blood distribution has disappeared so far, and this is registered as a matching weight table. To do. In the matching process, template matching is performed using this weight table.

  FIG. 8 is an example of a registration process procedure for fatprint registration data used in the collation process. Note that it is assumed that a plurality of color images are captured by the input device 2 and the images are input to the authentication processing unit 10 before the processing of FIG. 8 is performed. Since the weight table can be created if the color change when the acrylic plate 23 is pressed with the fingertip can be observed, images of at least two different states when the fingertip is brought into contact with or pressed against the acrylic plate 23 of the input device 2 are obtained. It only has to be acquired. The subject of the following processing is the registration unit 10b of the authentication processing unit 10.

  First, the registration unit 10b generates fatprint registration data (S801). The registration data is a threshold value process for the distribution of melanin and hemoglobin concentration obtained by the above-described method so that it becomes binary data of the blood portion of the fat crest and the non-blood portion. Part and other data are converted into binary values of the non-blood part. Further, the registration unit 10b creates registration data of a plurality of fat prints of the same finger using a plurality of images obtained by photographing a plurality of fingers. And the registration part 10b preserve | saves arbitrary one of these in the memory | storage device 14 as registration data. As the registration data, the first or last one may be selected. As another example, a round-robin collation may be performed between a plurality of registered data candidates photographed, and the result that has the highest similarity with other candidates may be selected as registered data. The registration data is not limited to a single image, and a plurality of images may be stored as registration data.

  Next, the registration unit 10b initializes the weight table (S802). The weight table is a distribution of blood observation probabilities, and is used in collation processing described later. A specific initialization method first superimposes all of the plurality of registered data candidates photographed above at the position where the similarity is the highest, and holds the frequency with which the pixel value of each coordinate is classified as blood as a probability. For example, when 5 registered data candidates are acquired, all images are collated and overlapped, and if a pixel at a certain coordinate becomes a blood part with 3 data, the coordinate stores a value of 3 in the weight table. It will be. In this way, the blood part frequency number is filled in all the pixels of the weight table, and this is used as the initial value of the weight table. In addition, the total number of statistical data, which is the number of data obtained when creating the weight table, is also initialized. For example, when 5 pieces of registered data are used, the total number of statistical data is initialized to 5. That is, dividing the weight table value by the total number of statistical data gives the blood observation probability at that pixel. Therefore, the weight table is data in which each pixel in the image is associated with the blood observation frequency at that pixel. Finally, the registration unit 10b stores the initialized weight table in the storage device 14.

  Next, an embodiment of a fatprint matching process procedure will be described with reference to FIG. FIG. 9 is an example of a collation procedure between input data obtained from a user and registered data.

  First, the user's finger placed in the placement area is photographed by the input device 2 shown in FIGS. 2A and 2B (S901). The captured image is input to the authentication processing unit 10 via the image input unit 18. Subsequently, the authentication unit 10a of the authentication processing unit 10 acquires a fat pattern from the input image based on the above-described method (S902). After that, the authentication unit 10a creates fatprint input data. Here, as with the generation of the registration data, it is converted into a blood part and other binary data.

  Next, collation of fat pattern input data and fat pattern registration data is performed. Here, weight collation using a weight table is performed (S904 to S905).

  Here, one method of collating fatprint templates by weighting will be described with reference to FIGS. 10A to 10C. Here, the registration data 100 in FIG. 10A is taken in a state where the finger contact pressure is low, and the input data 101 in FIG. 10B is taken in a state where the finger contact pressure is high. FIG. 10C is a diagram showing the distribution of the observation frequency stored in the weight table described above in shades on a two-dimensional plane.

  In general template matching, both the registration data 100 and the input data 101 having a slightly strong pressing pressure are overlapped to find the most matching place while shifting the position and rotation angle, and the similarity at that time is calculated. As a method of determining the most coincident place, when both data are overlapped, the total number of pixels of different pixels where one blood part 102 and the other non-blood part 103 overlap is summed, and this is calculated as the blood part of both data. The value obtained by dividing the total number of pixels is regarded as the dissimilarity, and this is based on a method of searching for an overlapping position where the value becomes the smallest. The value obtained by subtracting the difference degree from 1 at this time represents the similarity degree between the two data.

  On the other hand, in the template matching of the present embodiment, the difference from the conventional template matching is that the values of the number of different pixels and the total number of pixels of the blood part 102 are converted based on the weight table 104. . First, the authentication unit 10a calculates a weighted total value of the number of different pixels described above using the weight table 104 (S904).

  As shown in FIG. 10C, the weight table 104 represents a portion 105 where blood is frequently observed in a dark color, a portion 106 where the blood is frequently observed in a light color, and a portion where no blood appears in white. , Its value varies continuously spatially. First, the registration data 100 and the input data 101 are overlapped at a certain position, and the weight table 104 is overlapped so as to match the registration data 100. Then, in a state where the registration data 100 and the input data 101 are superimposed, the number of different pixels in which one is a blood pixel and the other is a bloodless pixel is counted. At this time, when a specific value obtained by dividing the value stored in the weight table 104 corresponding to a certain coordinate by the total number of the statistical data is defined as Wi, conventionally, it is counted as one pixel. And count. In addition, since the value of Wi represents the appearance probability of the blood part at that position, 0 ≦ Wi ≦ 1.

  Next, the authentication unit 10a calculates the degree of difference using the above-described weighted total value of the different pixels (S905). Similarly, when counting the total number of pixels of the blood part, the appearance probability of the blood part is calculated from the value of the weight table of the same coordinates, and this value is added. Finally, the degree of difference is calculated by dividing the weighted number of different pixels by the total number of pixels in the weighted blood part.

  Next, the authentication unit 10a determines whether the input data 101 and the registration data 100 match according to the degree of similarity (S906). The similarity is, for example, a value obtained by subtracting the difference from 1. The determination as to whether they match can be made, for example, based on whether the similarity exceeds a predetermined threshold.

  If it is determined that the degree of similarity is high, authentication is successful (S907), and a predetermined authentication process is performed. Then, the authentication unit 10a updates the weight table 104 (S908) and ends the authentication process. The weight table 104 is updated using the current weight table 104, the total number of statistical data, and the input data 101 when authentication is successful. That is, at the overlapping position where the degree of similarity is highest with the registered data 100, the value of the weight table 104 corresponding to the pixel that becomes the blood part 102 of the input data 101 is incremented by one, and the total number of statistical data is also incremented by one. Thereby, the appearance frequency of blood is updated. The updating of the weight table 104 may be performed by capturing a plurality of moving images that are being pressed during input in a moving image, and calculating the blood observation frequency for each pixel.

  On the other hand, if the degree of similarity is smaller than the predetermined threshold value, they do not match, and the authentication unit 10a determines a time-out after starting finger imaging (S909). If it is not timed out, the process returns to finger shooting again. If timed out, authentication fails (S910), and the authentication process is terminated.

  In this way, by considering the weight of the observation probability of blood in the number of different pixels, the difference in the portion that is likely to fluctuate due to the pressing pressure can be almost ignored, and the difference in the pattern that is observed stably is evaluated as a larger penalty. The verification process can be performed as described above. Thereby, it becomes possible to evaluate the similarity of the shape of a pattern, absorbing the presence or absence of the variation of the pattern by pressing.

  Moreover, although the above-mentioned collation method has the registration data 100 and the weight table 104 separately, since the weight table 104 is the information reflecting the blood pattern, it does not hold | maintain the registration data 100 but a weight table. It is also possible to collate by considering 104 as registered data 100. However, since the weight table 104 is gradually updated, if the pattern can be slightly changed intentionally, there is a possibility that it will eventually be replaced with another pattern. Therefore, as described above, it is safe to have the registration data 100 separately from the weight table 104 and collate it based on this. When the total number of statistical data becomes a sufficiently large value, if the registration data 100 is a non-blood part but there are pixels in the weight table 104 that a blood part appears frequently, this is the case for the registration data 100. The non-blood part of the pixel may be changed to a blood part. This is because the blood part may be missing when the registration data 100 is generated, and this change makes it possible to create the registration data 100 that reflects the true pattern and improve the authentication accuracy. .

  In the above-described embodiment, the matching process based on template matching has been described. However, for example, feature point matching is used in which end points, branch points, and inflection points are extracted, and matching is performed based on information such as luminance gradients around them. It is needless to say that even in the matching method described above, processing that is robust against fluctuations in pressing can be realized using the same weight table.

  According to the present embodiment, it is possible to acquire various biological feature information characterizing a person from a narrow range and perform authentication with high accuracy. Since it is possible to extract information in consideration of a state in which colors presented by a plurality of biological tissues are superimposed from a narrow range such as a fingertip, it is possible to provide a highly accurate biometric authentication device that is small and easy to use.

[Second Embodiment]
FIG. 11 shows an example of a biometric authentication apparatus using a fat print and a finger vein according to the second embodiment. The casing 21 of the input device 2 is provided with a finger rest 113 for placing the finger 1. In this embodiment, an image is acquired in a state where the finger is not brought into contact with the acrylic plate 23 of the opening 22.

  Further, the light source support 112 is provided on the casing 21 of the input device 2. The light source support 112 is provided so as to extend above the placement area of the finger 1, and the near-infrared transmitted light source 110 is attached to the light source support 112. The near-infrared transmission light source 110 is installed so as to irradiate the back side of the hand of the finger 1.

  A visible light reflection light source 111 is provided inside the housing 21, and the reflection light source 111 irradiates the palm side of the finger 1. The imaging device 9 is a general color camera, and near infrared light is received by a light receiving element sensitive to red of the color camera 9. As a result, a finger vein pattern is obtained from the image transmitted through the light from the transmissive light source 110, and a fat pattern is obtained from the image reflected from the light from the reflected light source 111. However, it goes without saying that the biometric features obtained from the reflected image are not limited to fat prints, but can be combined with fingerprints, joint prints, melanin distribution, and the like.

  FIG. 12 is a processing flow of an embodiment of authentication using fat prints and finger veins in the authentication processing system of the present embodiment. As a premise of the following authentication processing, the storage device 14 stores a user's finger vein pattern (for example, the pattern in FIG. 14A) and a fat pattern (for example, the pattern in FIG. 14B) as registered data. 14 is stored.

  First, when the user places the finger 1 on the placement area of the input device 2, the transmitted light source 110 is turned on and the reflected light source 111 is turned off. Then, an infrared transmission image is taken by the color camera 9 to obtain a finger vein image (S1201). Subsequently, the transmission light source 110 is turned off, the reflection light source 111 is turned on, and a reflected image of visible light is captured by the color camera 9 to obtain a fatprint image (S1202). The finger vein image and fat print image obtained here are input to the authentication processing unit 10 via the image input unit 18.

  The reason why these light sources are not turned on at the same time is that both infrared transmitted light and visible reflected light react to the light receiving element sensitive to the red wavelength. This is because it becomes impossible to separate the infrared light and the red light that have reached 1. Therefore, these light sources are alternately turned on instantaneously. That is, the light receiving element having sensitivity to the red component when the transmitted light is turned on and when the reflected light is turned on captures infrared transmitted light and visible reflected light, respectively. In this way, the infrared transmission image and the visible reflection image at substantially the same time can be received by one color camera 9 by receiving the light in a time division manner while instantaneously switching the lighting of the infrared light and the red light. Since the images are almost the same time, it can be considered that there is no positional deviation of the finger in both images.

  If the color camera 9 is replaced with a sensor equipped with a sensor having sensitivity to infrared wavelengths, or a visible light / infrared camera that receives light by splitting the wavelength for each wavelength with a prism or the like, infrared light is simultaneously transmitted. Needless to say, visible light can be received.

  Next, the authentication unit 10a of the authentication processing unit 10 extracts a finger vein pattern and a fat pattern from the captured infrared transmission image and visible reflection image (S1203). Regarding finger vein extraction, as a generally known method, a vein pattern is acquired by, for example, a method of emphasizing a blood vessel with a matched filter or a Gabor filter, or a filter process for detecting the center position of the blood vessel. In addition, the method described in the above-described embodiment can be used for the fat pattern. In addition, not only fat prints but also fingerprints and joint prints may be extracted simultaneously from the visible reflection image. In addition, when the finger contour can be photographed, the finger position deviation amount may be normalized based on the finger contour. By using such other biometric features, it is possible to perform a collation that is robust against misalignment and to improve authentication accuracy.

  Subsequently, the authentication unit 10a collates the input finger vein pattern and the fat pattern with the registered finger vein pattern and the fat pattern (S1204). And the authentication part 10a calculates the difference degree of registration data and input data from the collation result of both biometric features (S1205). Specific examples of the collation processing and the difference calculation will be described later.

  Then, the authentication unit 10a compares the degree of difference with a preset threshold value, and determines whether both data can be regarded as matching (S1206). If it can be considered that it matches the registered data, the authentication is successful (S1207), and the process at the time of successful authentication is performed and the process ends. On the other hand, if it is determined that they do not match, it is determined whether or not a timeout has occurred since the start of shooting of the finger 1 (S1208). Otherwise, the process returns to shooting of the finger 1 by the input device 2. If timeout occurs, authentication fails (S1209), and the authentication process is terminated.

  Here, an example of collation processing of finger veins and fat prints and calculation of the degree of difference will be described. FIG. 13A to FIG. 13C show schematic diagrams of finger vein and fat pattern patterns respectively taken.

  FIG. 13A is an infrared transmission image 130 of a finger. The infrared transmission image 130 of a finger shows a finger through which light from the transmission light source 110 is transmitted, and a finger vein 132 is shown here. FIG. 13B is a visible light reflection image 131 of the finger. The finger when the reflected light source 111 is irradiated is reflected in the visible light reflection image 131 of the finger, and the fat crest 133 is distributed therein.

  The fat pattern 133 has a fine spot-like pattern, and the finger vein 132 has a linear pattern. Further, since the fat crest 133 and the finger veins 132 are biological features of different layers in the skin, a composite image 134 of the finger veins 132 and the fat crest 133 is created as shown in FIG. 13C. As shown in the composite image 134, the fat crest 133 and the finger vein 132 may overlap each other. In that portion, the spotted pattern of the fat crest 133 is observed overlapping the finger vein 132 line. Since both of these patterns can be acquired independently, it is also possible to collate each independently. However, it is possible to obtain higher authentication accuracy by collating using the overlapping method. Here, two examples relating to the technique will be described.

  The first embodiment is a method of extracting a feature point from the position of the pattern of the fat pattern 133 when the finger vein 132 is collated based on the feature point.

  For finger vein patterns, robust matching against displacement and deformation can be achieved by using feature points such as SIFT (Scale-invariant feature transform) or information corresponding to fingerprint minutia. A portion that is characteristic as the shape of the vein 132 is automatically detected. As characteristic portions of the finger vein 132, end points, branch points, and bending points can be detected stably. However, it is relatively small compared to the number of bifurcation points and end points of biometric features constituted by fine line patterns such as fingerprints. In addition, the area of the non-distributed portion of the vein is relatively large and often has a sparse pattern. Therefore, a large number of feature points cannot be extracted from the finger vein pattern alone, and there is a tendency for authentication accuracy to deteriorate. Therefore, if the spotted pattern of the fat crest 133 is utilized, the feature amount of the finger vein 132 can be relatively increased.

  The specific processing procedure is as follows. The subject of the following processing is the authentication unit 10a as described above. 14A to 14E are schematic diagrams illustrating a feature point matching method in which the feature amounts of the fat crest 133 and the finger vein 132 are combined.

  First, as illustrated in FIG. 14A, the authentication unit 10 a creates a finger vein pattern image 140 from a transparent image of the finger vein 132. Further, as illustrated in FIG. 14B, the authentication unit 10 a creates a fat pattern image 141 from the reflected image of the fat pattern 133.

  Next, as shown in FIG. 14C, the authentication unit 10a detects a branch point, end point, inflection point, etc. of the vein from the finger vein pattern image 140, and the finger vein 132 shown by a solid circle in the figure. The feature point 143 is obtained. Similarly, as illustrated in FIG. 14D, the authentication unit 10a detects the center position of the spot pattern of the fat pattern from the fat pattern image 141, and the feature point 145 of the fat pattern indicated by a dotted circle in the drawing. Get.

  Finally, as shown in FIG. 14E, the finger vein pattern image 140 obtained from the feature point 143 and the fat print image 141 obtained from the feature point 145 are superimposed to synthesize the feature points of the finger vein and the fat print. An image 146 is obtained. These points are regarded as feature points, and feature point matching based on the above-mentioned SIFT and minutiae is performed. At this time, as a classification of feature points, in addition to the finger vein feature point 143 and the fat print feature point 145, a pattern in which the finger vein pattern overlaps a fat print dot pattern is regarded as the third feature point. Also good. Generally, in the feature point matching by the minutiae, miscorrespondence is reduced by classifying whether the feature point is an end point or a bifurcation point. Or different labels such as fat on a vein. More specifically, in this embodiment, the feature points are classified in terms of the types of biometric features and feature points, for example, end points on veins, branch points on veins, bend points on veins, fat prints. , Can be divided into categories like fat on the veins.

  If the feature points can be acquired, the feature amounts around each feature point can be described using veins and fat prints and collated. As the feature amount, a luminance gradient feature of an image such as SIFT can be used. At the time of collation, the authentication unit 10a calculates the similarity of feature amounts between feature points classified into the same category. And the final similarity of both data can be acquired with the ratio of the number of the feature points which can be determined to be similar among all the feature points.

  With the above processing, feature points can be obtained even in a portion that cannot be obtained conventionally as a feature point of a finger vein (for example, a portion where a vein runs in a straight line). The accuracy of authentication can be improved.

  In the second embodiment, when the fat crest and the finger vein are collated independently, a positional deviation amount between patterns detected by one collation is used for the other collation.

  In actual operation, the finger position during registration and the finger position during authentication often differ slightly. Therefore, in general finger vein verification processing, the registration data and the input data are overlapped and the patterns are compared while shifting the position, and the position where the pattern is most similar is the amount of positional deviation of the finger. presume. Thereby, the collation between the data containing a position shift can be performed correctly.

  However, when matching different patterns, it will also look for a place where the similarity of the pattern will increase, so there is a tendency for the similarity to increase even when matching different finger patterns that have low similarity. It was. This occurs because the amount of positional deviation is estimated based on the similarity of patterns instead of performing essential positional deviation correction of the finger. On the other hand, there is a method of using information that can specify the position of the finger, such as finger outline information, in order to detect a true finger position shift amount. However, in this method, the outline of the finger may not be captured depending on the authentication device. The finger contour information is acquired as two-dimensional information. However, the actual finger misalignment may occur in three dimensions, and the finger misalignment may occur even when the contour position as the two-dimensional information is adjusted. Cannot always be corrected correctly. For this reason, it is difficult to specify the absolute position of the finger from the contour of the finger.

  In order to solve this problem, in this embodiment, the fat pattern and the finger vein are independently collated, and after calculating the amount of misalignment with the highest similarity in one pattern, the degree of similarity is calculated in the other pattern. Determine. This method will be described with reference to FIG. The subject of the following processing is the authentication unit 10a of the authentication processing unit 10 as described above.

  First, the authentication unit 10a superimposes the registered finger vein pattern 150 and the input finger vein pattern 151, and searches for a positional deviation amount that most closely matches the vein pattern. Next, the authentication unit 10a uses the search position and superimposes the fat pattern 152 registered in the position shift amount state with the input fat pattern 153 and collates them to determine the similarity of the fat patterns. calculate. However, considering that there is an error in the search for the amount of misalignment at this time, the position is slightly shifted when comparing the fat prints, and the highest similarity among the results is obtained as the final fat print similarity. You may earn as

  Thereafter, the processing with the finger vein and the processing with the fat crest are interchanged. As described above, the authentication unit 10a superimposes the registered fat pattern 152 and the input fat pattern 153, and obtains a positional deviation amount that maximizes similarity in the fat pattern. Next, the authentication unit 10a calculates the finger vein similarity by superimposing the registered finger vein pattern 150 and the input finger vein pattern 151 on the basis of the positional deviation amount. Finally, the authentication unit 10a synthesizes the two similarities of the finger vein and the fatprint by a method such as averaging the similarity of the finger vein and the similarity of the fatprint or linearly combining them according to a predetermined weight. , To obtain the final similarity or dissimilarity.

  In the method of this embodiment, the feature amount used for estimating the positional deviation amount of the finger and the characteristic amount for calculating the similarity are different information. Thus, it is possible to correctly determine the similarity between fingers that are misaligned while avoiding the problem that the degree of similarity decreases.

  FIGS. 16A to 16C are diagrams illustrating a process of coding and authenticating an overlapping portion between the finger vein 132 and the joint pattern.

  A joint crest is a skin wrinkle in a joint such as a hand, and the epidermal layer and subcutaneous tissue are thinner than the surroundings. Since the position of the joint pattern is almost unchanged, and the shape of the joint pattern varies depending on the finger, it is one of the biometric features that can be used for authentication robust against displacement. In the following description, it is assumed that the joint pattern picked up is a wrinkle at a joint portion closest to the fingertip on the palm side of the finger.

  The joint crest is a part where the color of hemoglobin is superior to melanin because the epidermis is thin. Since this is a part exhibiting the color of hemoglobin, the authentication apparatus using reflected light and transmitted light shown in FIG. 11 can photograph not only a fat pattern but also a joint pattern. At this time, in the method of acquiring the concentration distribution of hemoglobin and melanin using reflected light, a location where the hemoglobin concentration is higher than the melanin concentration can be specified as the position of the joint pattern. Also, a finger vein 132 is running on the back side of the joint pattern, and this pattern can also be photographed using the transmitted light in FIG.

  FIG. 16A shows a superimposed image 160 of finger veins and joint patterns acquired by the input device 2. In the superimposed image 160, the finger vein 132 is dominated by blood vessels that run along the longitudinal direction of the finger, while the joint crest 161 runs in a direction perpendicular to the longitudinal direction of the finger. And the joint crest 161 are generally orthogonal.

  Therefore, as illustrated in FIG. 16B, the authentication unit 10a of the authentication processing unit 10 extracts a vein pattern and a joint pattern by the above-described method, and creates an image 162 in which both patterns are superimposed. At this time, an intersection 163 between the finger vein 132 and the joint pattern 161 becomes a characteristic amount unique to the finger. In addition, the joint pattern 161 can be clearly observed as a pattern on the surface of the finger, and since it exists at almost the same position with most fingers, it is easy to specify the position uniquely, and it is detected even when the position of the finger is shifted. It is easy to use. The finger vein 132 also has the advantage that the joint pattern 161 can be clearly observed because the epidermis is thin. However, there may be a plurality of joint patterns 161 or bend. Therefore, the authentication unit 10a linearly approximates the joint pattern 161 by the least square method. Further, if a plurality of joint prints are observed, the authentication unit 10a uniquely uses the joint print closest to the fingertip or obtains the approximate print lines of all joint prints. To decide.

  Next, the authentication unit 10 a converts the approximate straight line of the joint pattern 161 and the intersection 163 of the finger vein 132 into a one-dimensional binary code 164. Along the approximate straight line of the joint crest 161, it is checked pixel by pixel whether the point intersects the finger vein 132. If it does not intersect, 0 is assigned, and if it intersects, 1 is assigned as a code. The code is filled from the top to the bottom, and the binary code 164 is created as a whole. The white part of the binary code 164 represents a part intersecting the finger vein 132, and the black part represents the other part. In this example, the black portion of the binary code 164 is a bit string assigned 0, and the white portion 1. The longer the distance between two intersections, the longer the 0 bit string inserted between them. In addition, the bit string of 1 becomes longer as the width of the vein at the intersecting point is wider.

  The length of the bit string of the binary code 164 may be assigned as one bit per pixel corresponding to the pixel of the image taken as described above, and the bit string is converted to a scale that is enlarged or reduced by a certain coefficient. Also good. However, since the width of the finger vein 132 may fluctuate, stability is enhanced by normalizing the vein width. In other words, the center line of the finger vein 132 is detected, the center line is thickened to a certain width, and the intersection with the approximate straight line of the joint pattern 161 is acquired as the intersection 163. In this way, the length of one bit string is the same for all intersections. As described above, since the relative position between the joint pattern 161 and the finger vein 132 is invariant, if the joint pattern can be detected, a difference depending on the scale occurs, but this binary code is almost unchanged in position. It can be acquired as information.

  An example of the collation process of the binary code 164 is as follows. FIG. 16C is a diagram for explaining a binary code matching process. The authentication unit 10a compares the binary code 164 obtained from the input image with the binary code 166 that is registered data registered in advance by the same method as described above, and calculates the similarity. In general, since a method based on a Hamming distance or Euclidean distance can be applied to a method for calculating the similarity between binary codes, the distance between the registered binary code and the input binary code can be applied. Can be regarded as the degree of difference.

  However, when the finger rotates three-dimensionally with respect to the central axis, the binary code 164 may shift in the vertical direction in the figure. Therefore, as shown by the arrows in FIG. 16C, for example, the binary code 165 that is the input data is shifted in the vertical direction bit by bit, the binary code 166 that is the registered data is compared, and both binary codes 165, 166 obtains the most similar result as the final similarity. In addition, since the enlargement ratio of the finger may be different, the scale of the binary code 166 of the registered data is expanded or contracted in several ways, and the result with the highest similarity is obtained after collation with all data. Final result. Alternatively, a collation process that is robust to fluctuations in enlargement may be performed by a DP (Dynamic programming) matching method that can handle expansion and contraction.

  When the finger contour can be photographed, the binary code is obtained after normalizing the enlargement ratio of the image so that the intersection of the extension line and the finger contour line when the joint pattern 161 is linearly approximated becomes a certain number of pixels. It may be generated. If this is possible, there is no need to consider the expansion and contraction of the scale of the binary code, which can simplify the process and contribute to speeding up and high accuracy.

  According to the present embodiment, even when a finger misalignment occurs, the position can be corrected or detected, and a small biometric authentication device that is highly convenient and has high authentication accuracy can be provided.

  Further, according to the present embodiment, it is possible to perform authentication that is invariant to the positional deviation of the finger. Furthermore, it is known that the technique for generating a code that is invariant to misalignment can be applied to the construction of a secure biometric authentication system and the management of biometric information, such as calculating the similarity between patterns in an encrypted state. For example, the present invention can be applied to a template public biometric infrastructure (PBI) technology for realizing a public key infrastructure with biometric data.

  In the example of FIG. 12 described above, only the authentication process that does not use the weight table has been described. However, the contents of the first embodiment may be incorporated into the present embodiment and the authentication process using the weight table may be performed.

[Third embodiment]
17A and 17B show an embodiment of a biometric authentication apparatus that is the third embodiment. The biometric authentication device of the present embodiment has a configuration applicable to the first embodiment and is mounted on a mobile terminal such as a smartphone or a tablet. In this example, the mobile terminal includes a visible light source for reflected light irradiation and a color camera as an input device.

  The input device 2 described above is mounted on the housing of the smartphone 170. The input device 2 also serves as a main button of the smartphone 170, and various operations of the smartphone 170 can be performed by pressing the input device 2.

  When performing biometric authentication, the user places the finger 1 in contact with the placement area at the top of the input device 2. At this time, it can be displayed on the screen 171 that a finger is placed. Regarding the imaging of biological information, in view of the size of the input device 2 that can be mounted on a small terminal such as the smartphone 170, the opening 22 of the input device is often smaller than the fingertip, and thus the area of the living body that can be captured is limited. (FIG. 17B). For example, if the size of the opening 22 of the input device 2 is a circle having a diameter of 10 mm, the range of the fingertip having a diameter of 10 mm can be photographed. However, the area of the fingertip is often larger than that, and information on the entire fingertip cannot be captured. On the other hand, if the entire fingertip is photographed in a non-contact state with the fingertip floating in a hollow state, the entire fingertip can be projected at the angle of view of the camera.

  As a general method of presenting the fingertips in a non-contact manner, for example, the target finger position is displayed on the smartphone screen while photographing the finger position with a camera, and the user fine-tunes the fingertip accordingly. There is a way. However, with this method, operability is likely to be degraded, such as difficulty in one-handed operation, fatigue due to the fact that it is difficult to perform alignment intuitively, and it takes time for alignment.

  Therefore, in this embodiment, the operability is enhanced by an interface in which the finger 1 is first placed in a predetermined placement area and the finger 1 is tapped at that position. 18A and 18B show an example of the operation of the fingertip authentication device for the smartphone 170 that performs authentication by tapping the finger 1.

  In the present embodiment, the components such as the image input unit 18 and the authentication processing unit 10 described above are implemented as software installed in the smartphone 170. Note that these components may be implemented by an external authentication device different from the smartphone 170, and the image acquired by the input device 2 in the smartphone 170 is transmitted to the external authentication device via the network. Authentication processing may be executed by an external authentication device.

  First, the authentication unit 10a displays “Place your finger on the sensor” on the screen 171 of the smartphone 170. In accordance with this, the user places the thumb 180 on the placement area of the input device 2. However, in this embodiment, since the input device 2 is located at the position shown in FIGS. 17A and 17B or FIGS. 18A and 18B, the thumb 180 is placed in the placement area. However, the present invention is not limited to the thumb. For example, the same operability can be realized by providing the input device 2 on the back side of the smartphone 170 and placing the index finger or the middle finger on the input device 2.

  Next, the authentication unit 10a displays a guidance “Please start tapping” on the screen 171 of the smartphone 170. In accordance with this, the user taps the position (mounting area) of the opening 22 of the input device 2 with the thumb 180. The tap operation is an operation of tapping a predetermined position with a fingertip. At this time, the fingertip moves up and down as it floats in the air and contacts the device. In this embodiment, the opening 22 of the input device 2 is tapped several times. FIG. 18B depicts a thumb 181 floating in the air. The user repeats the operation of bringing the thumb 180 into contact with the opening 22 of the input device 2 (FIG. 18A) and the operation of lifting the thumb 181 from the opening 22 of the input device 2 (FIG. 18B).

  As an advantage of the tap operation, the place where the fingertip is first placed on the input device 2 can be easily grasped so that it is difficult to get lost in the place. The tap operation has a limited trajectory because the range of movement of the finger is limited. There is that it does not shake greatly, and furthermore, since the finger length is constant, the distance when the finger is moved farthest is substantially constant, so that the enlargement rate is stable and one-handed operation is possible.

  During the tap operation, the reflected light of the thumb is photographed in a moving image by the color camera 9 while continuing to irradiate the visible light source 3 in the input device 2 toward the thumb. In the present embodiment, the wavelength of the visible light source is limited to green and only the green plane of the photographed color image is used so that the fat pattern can be observed relatively clearly and the processing is simplified. . This advantage not only makes it easy to observe fat prints, but also makes it easier to identify finger regions because the color can be limited to green. The brightness of the entire screen is saturated when the finger is in contact with the camera, but when the finger is moved away from the camera, the entire fingertip enters the angle of view, and the surrounding background image is also reflected. When the tapped finger is farthest away, the fingertip appears to be the smallest, and approaches the camera again, and finally returns to the state of covering the camera with the finger. A moving image while tapping is obtained while repeating this state several times. This moving image is input to the authentication processing unit 10 via the image input unit 18. The authentication unit 10a of the authentication processing unit 10 extracts an image of a fingertip that can be determined to be appropriate from the moving image.

  The image of the fingertip that can be determined to be appropriate is preferably in a state where the fingertip is less shaken and the entire fingertip can be photographed. This state is considered to be a state in which the fingertip is farthest from the input device 2 during the tap. It can be estimated that the finger is farthest away when the area on the image when the fingertip is irradiated with the illuminating green light source 3 is small. Therefore, for the image of each frame of the moving image, the minimum value when calculating the number of pixels of spatially continuous pixels among the pixels with strong green component pixel values and arranging them in time series Can be acquired as an image when the finger is farthest away. In the state where the finger is farthest away, the moving speed of the finger is the slowest, so that there is little blurring of the image and a stable image can be obtained. Alternatively, an image obtained when the number of spatially continuous pixels with strong green components is within a predetermined range may be acquired as the optimum image. In addition, since several taps are urged, a plurality of images determined to be appropriate are obtained. Here, all of these are used for authentication.

  When photographing of the finger image is completed, the authentication unit 10a cuts out the fingertip region from the image. Since the fingertip is irradiated with the green light source 3, it can be implemented by a method in which a lump of continuous pixels in green is regarded as the fingertip. Alternatively, the fingertip region may be determined based on segmentation such as graph cut.

  Subsequently, the authentication unit 10a extracts a fatty pattern existing in the fingertip region and performs collation. As described above, this process can use fat pattern emphasis processing by unsharp mask processing and fat pattern matching processing techniques. However, since a plurality of fingertip images are obtained, the authentication unit 10a performs a brute force comparison between the plurality of registered images and the plurality of input images, and obtains the highest similarity as the final similarity. Is calculated. If the similarity exceeds a predetermined value determined in advance, the authentication is successful assuming that the registered finger is presented. As a result, for example, it is possible to automatically perform identity verification when unlocking the smartphone 170 or shopping online.

  It should be noted that not only the fat prints on the fingertips but also other biological features such as fingerprints may be extracted from the reflected image and used together. Thereby, authentication accuracy can be improved more.

  According to the present embodiment, it is possible to provide a biometric authentication device that performs high-precision personal authentication that is easy to use in a small portable terminal such as a smartphone. In particular, highly accurate personal authentication is possible while maintaining a high degree of freedom when presenting a living body.

  The present invention is not limited to the above-described embodiments, and includes various modifications. The above embodiments have been described in detail for easy understanding of the present invention, and are not necessarily limited to those having all the configurations described. Also, a part of the configuration of one embodiment can be replaced with the configuration of another embodiment. Moreover, the structure of another Example can also be added to the structure of a certain Example. Further, with respect to a part of the configuration of each embodiment, another configuration can be added, deleted, or replaced.

  For example, the present invention may be implemented as an image capturing apparatus that does not include an authentication unit, from the viewpoint of acquiring information related to a plurality of biological tissues that are superimposed. The image capturing apparatus includes a placement area on which a finger is placed, at least one light source that irradiates light on the finger, and an imaging unit that captures light from the light source to acquire at least one image. The imaging unit may acquire a plurality of images in different states when the finger is brought into contact with or placed on the placement area. As a result, it is possible to acquire information on a plurality of living tissues that are superimposed. Similarly to the above-described embodiment, the imaging unit may be configured to acquire the first image by the transmissive light source and the second image by the reflected light source. Further, as in the above-described embodiment, the imaging unit may be configured to acquire a plurality of images during an operation of hitting the placement area with a finger. Note that some or all of the functions of the above-described image processing unit other than the authentication processing may be appropriately implemented in the image capturing apparatus.

  Each of the above-described configurations, functions, processing units, and the like may be realized by hardware by designing a part or all of them with, for example, an integrated circuit. Each of the above-described configurations, functions, and the like may be realized by software by interpreting and executing a program that realizes each function by the processor. Information such as programs, tables, and files for realizing each function can be stored in a recording device such as a memory, a hard disk, an SSD (Solid State Drive), or a recording medium such as an IC card, an SD card, or a DVD.

  In the above-described embodiments, the control lines and information lines are those that are considered necessary for explanation, and not all the control lines and information lines on the product are necessarily shown. All the components may be connected to each other.

  INDUSTRIAL APPLICABILITY The present invention can implement a highly accurate biometric authentication device that is small and excellent in convenience, and is useful as a personal authentication device.

DESCRIPTION OF SYMBOLS 1 ... Finger, 2 ... Input device, 3 ... Light source, 9 ... Imaging device (color camera), 10 ... Authentication processing part, 10a ... Authentication part, 10b ... Registration 11, CPU (Central Processing Unit), 12 Memory, 13 Interface, 14 Storage device, 15 Display unit, 16 Input unit, 17 Speaker, 18 ... image input unit, 21 ... housing, 22 ... opening, 23 ... acrylic plate, 30 ... vein, 31 ... fingerprint, 32 ... capillary, 33 ... Fat leaflets, 34 ... skin, 35 ... dermis layer, 36 ... subcutaneous tissue, 60 ... pixel value before pressing finger, 61 ... pixel value after pressing finger , 70 ... fingertip image, 71 ... blood pattern, 72 ... blood pattern that has not disappeared, 73 ... blood that has disappeared 100 ... registration data, 101 ... input data, 102 ... blood part, 103 ... non-blood part, 104 ... weight table, 105 ... part with high blood observation frequency, 106: a portion where blood observation frequency is low, 110: a transmission light source, 111: a reflection light source, 112 ... a transmission light source support, 130 ... an infrared transmission image of a finger, 131 ... Visible light reflection image of finger, 132 ... finger vein, 133 ... fat pattern, 134 ... composite image of vein and fat pattern, 140 ... finger vein pattern image, 141 ... fat pattern image, 143: Finger vein feature points, 145 ... Fat print feature points, 146 ... Finger vein and fat print feature point composite image, 150 ... Registered finger vein pattern, 151 ... Input finger Vein pattern, 152 ... registered fat crest, 153 ... input Fat pattern, 160 ... superimposed image of finger vein and joint pattern, 161 ... joint pattern, 162 ... superimposed image of vein pattern and joint pattern, 163 ... intersection of finger vein and joint pattern, 164, 165, 166 ... binary code, 170 ... smartphone, 171 ... screen, 180 ... thumb, 181 ... thumb floating in the air

Claims (15)

A placement area on which a finger is placed;
At least one light source for illuminating the finger;
An image capturing unit that captures light from the light source to acquire at least one image;
An authentication unit for processing the image;
A storage unit that stores registration information related to at least one of a plurality of biometric features superimposed on the inside of the finger;
With
The authentication unit
From the image, obtain information on the distribution of pigment concentration related to a plurality of biological features superimposed on the inside of the finger,
Obtaining input information relating to at least one biological feature from the distribution of the pigment concentration;
Calculating the similarity using the input information and the registration information;
A biometric authentication device that performs authentication based on the similarity. The biometric authentication device according to claim 1,
The biometric authentication apparatus characterized in that the information on the distribution of the pigment concentration is information on the distribution of the concentrations of melanin and hemoglobin. The biometric authentication device according to claim 2,
The biometric authentication device, wherein the input information and the registration information are information relating to a fat pattern represented by a contrast between the melanin and the hemoglobin. The biometric authentication device according to claim 1,
The imaging unit acquires images in a plurality of different states when the finger is brought into contact with or pressed against the placement area,
The biometric authentication device, wherein the authentication unit calculates information on the distribution of the dye concentration based on a change amount of color information in the plurality of images in different states. The biometric authentication device according to claim 1,
The biometric authentication device, wherein the authentication unit specifies a specific part of the finger from color information in the image, and calculates information on the distribution of the pigment concentration from the color information of the part. The biometric authentication device according to claim 1,
The biometric authentication apparatus, wherein the authentication unit distinguishes and extracts a plurality of biometric features included in the image using information on the distribution of the pigment concentration, and acquires the input information. The biometric authentication device according to claim 3.
The input information is a first image related to the fat print,
The registration information includes an observation probability distribution representing a distribution of blood observation probabilities in a second image related to the fat print and images in a plurality of different states when the finger is brought into contact with or pressed against the placement area. Information and
The authentication unit calculates the similarity by converting the number of different pixels when the first image and the second image are overlaid using the observation probability distribution information. A biometric authentication device. The biometric authentication device according to claim 7, wherein
A registration unit that creates the observation probability distribution information from the plurality of images in different states, and stores the observation probability distribution information and the second image selected from the plurality of images in different states in the storage unit; A biometric authentication device further comprising: The biometric authentication device according to claim 1,
The light source includes a near-infrared transmission light source located above the finger, and a visible light reflection light source located below the finger,
The imaging unit acquires a first image from the transmissive light source and a second image from the reflected light source,
The registration information includes registration information of a first biometric feature and registration information of a second biometric feature superimposed on the inside of the finger,
The authentication unit
Obtaining the input information of the first biometric feature from the first image, obtaining the input information of the second biometric feature from the second image,
The input information of the first biometric feature and the input information of the second biometric feature are collated with the registration information of the first biometric feature and the registration information of the second biometric feature, and authentication is performed. A biometric authentication device. The biometric authentication device according to claim 9, wherein
The authentication unit includes a first feature point of the first biometric feature, a second feature point of the second biometric feature, and a point where the first biometric feature and the second biometric feature overlap. A biometric authentication device that performs the collation using a third feature point. The biometric authentication device according to claim 10, wherein
The first biological feature is a finger vein;
The biometric authentication device, wherein the second biometric feature is a fat print expressed by a contrast between melanin and hemoglobin. The biometric authentication device according to claim 9, wherein
The authentication unit, during the verification,
Obtaining a first displacement amount when the input information of the first biometric feature and the registration information of the first biometric feature are superimposed;
In the state of the first displacement amount, the first similarity is calculated by comparing the input information of the second biometric feature and the registration information of the second biometric feature,
Obtaining a second displacement amount when the input information of the second biometric feature and the registration information of the second biometric feature are superimposed;
The second similarity is calculated by comparing the input information of the first biometric feature with the registration information of the first biometric feature in the state of the second displacement amount,
A biometric authentication device, wherein a final similarity is calculated from the first similarity and the second similarity. The biometric authentication device according to claim 9, wherein
The authentication unit, during the verification,
Creating coded input information that encodes an intersection when the input information of the first biometric feature and the input information of the second biometric feature are superimposed;
Collating the coded input information with coded registration information obtained by coding an intersection when the registration information of the first biometric feature and the registration information of the second biometric feature are overlapped. A biometric authentication device. The biometric authentication device according to claim 13,
The first biological feature is a finger vein;
The biometric authentication device, wherein the second biometric feature is a joint pattern represented by a contrast between melanin and hemoglobin at a joint portion of the finger. The biometric authentication device according to claim 1,
It further includes a portable terminal having a display unit,
At least the placement area, the light source, and the imaging unit are mounted on the mobile terminal,
The authentication unit displays on the display unit the operation of hitting the placement area with the finger,
The imaging unit acquires a plurality of images during an operation of hitting the placement area with the finger,
The biometric authentication device, wherein the authentication unit acquires the input information from the plurality of images.

Images