CN107949863A - Use the authentication device and authentication method of Biont information - Google Patents

Abstract

Realize a kind of high-precision and small-sized organism authentication apparatus that can easily operate.The authentication device that a kind of feature using organism is authenticated individual, it is characterised in that possess：Setting area, sets organism；Light source, to showing the organism irradiation light in the setting area；Image pickup part, shoots the light from the light source reflected by the organism；Image processing part, handles the image imaged by the image pickup part；Feature extraction unit, the living body feature of the organism is extracted from described image；Storage part, keeps the living body feature；And comparing part, compare the similar degree of the living body feature, the camera device and the configuration of the multiple light source are in the position faced with the organism, the multiple light source irradiates respectively different wavelength, and the camera device shoots multiple wavelength, unit with the distance for measuring the organism and the camera device, judge the posture of the organism, detection is optimal to be shown position and induces to the position, and the correction of the feature of the organism is carried out based on the posture of the organism.

Description

Use the authentication device and authentication method of Biont information

Technical field

The present invention relates to the Verification System being authenticated using organism to individual, more particularly to small-sized and convenience are excellent Different, high-precision authentication techniques.

Background technology

In various organism authentication technologies, it is known that fin- ger vein authentication can realize high-precision certification.Fin- ger vein authentication Due to the use of the vascular patterns of finger interior, therefore it can realize excellent authentication precision, and due to compared with finger print identifying It is difficult to forge and distorts, therefore can realizes the security of height.

In recent years, in pocket telephone, notebook type PC (Personal Computer, personal computer), intelligence electricity Organism authentication apparatus is carried in the equipment such as portable terminal, locker, safety box, the printer such as words or tablet terminal, it is ensured that The example increase of the security of each equipment.In addition, as the field using organism authentication, except entry and exit control, attendance pipe Beyond reason, login to computer etc., in recent years, organism authentication is also used in clearing etc..The particularly public life utilized Object authentication device realizes that reliable personal authentication is critically important.Moreover, in view of the popularization of plate portable terminal in recent years, The trend of wearable computing, while convenience is ensured as described above, the miniaturization of realization device also becomes important important document One of.Further it is provided that operational high authentication device enables user also to become important easily with authentication device.

As with carrying out the relevant technology of the miniaturization of the authentication device of personal authentication based on the shape of blood vessel, have special Sharp document 1.

In addition in patent document, a kind of following technology is disclosed, i.e. using colour TV camera to contactless state Shade includes the reflected image of red visible light source in the palm shooting of device, extracts vein and the palmmprint of palm to be recognized Card.

Citation

Patent document

Patent document 1：JP patents disclose 2013-33555 publications

Patent document 2：International Publication No. 13/136553

The content of the invention

The subject that the invention solves

In order to realize that small-sized and ease of use is good and high-precision person authentication device, the sensing narrow from area Device shoots the Biont information of wide scope and effectively corrects position deviation when showing organism or generating position Put induces correct position to become important in the case of deviation to user.

In the organism authentication apparatus described in patent document 1, the opening sensor for realization device miniaturization Oral area is set must be narrower than the width of finger, but due to showing finger contact, cannot shoot the profile of finger, thus produce Position correction in the case of the position deviation of finger becomes difficult, and do not refer to its solution it is relevant in Hold.Further, since needing for finger to be placed on the given position of device, therefore the Biont information of wide scope cannot be shot, and And due to needing to have finger mounting table, the miniaturization difficult of device.

In patent document, following technology is described, i.e. the colored shooting being equipped with using smart phone or tablet standard The light of machine and ambient light or liquid crystal panel, while palm vein and palmmprint are shot to be authenticated.The technology has and need not connect The advantages of touching device and can widely shooting the Biont information of palm.But it is suitable to stop at organism for user When position etc. guiding it is not open.

Therefore, it is an object of the invention to realize such a person authentication device, the person authentication device is in order to realize The good high-precision personal authentication of small-sized and ease of use, the free degree that can be when keeping showing organism higher Meanwhile obtain the information of the living body feature with many information with personal characteristics.

Means for solving the problems

The authentication device that a kind of feature using organism is authenticated individual, it is characterised in that possess：Setting area Domain, sets organism；Light source, to showing the organism irradiation light in the setting area；Image pickup part, shoots by the organism The light from the light source of reflection；Image processing part, handles the image imaged by the image pickup part；Feature carries Portion is taken, the living body feature of the organism is extracted from described image；Storage part, keeps the living body feature；And comparing part, Compare the similar degree of the living body feature, the camera device and the configuration of the multiple light source are faced with the organism Position, the multiple light source irradiates respectively different wavelength, and the camera device shoots multiple wavelength, has measure institute The unit of the distance of organism and the camera device is stated, judges the posture of the organism, detects and optimal shows position simultaneously Induction carries out the correction of the feature of the organism to the position based on the posture of the organism.

Invention effect

According to the present invention, in the organism authentication apparatus using finger, device case is small-sized and can shoot width The organism of scope, and can be by obtaining diversified living body feature from this so as to accurately be authenticated, in addition It can also be corrected in the case of the position deviation of finger is generated or detecting location, using the teaching of the invention it is possible to provide a kind of small-sized and just The authentication device of the excellent high authentication precision of profit.

Brief description of the drawings

Fig. 1 is the integrally-built figure for the biometrics authentication system for showing first embodiment.

Fig. 2 is the figure of the apparatus structure for the biometrics authentication system for illustrating first embodiment.

Fig. 3 is to implement to reflect an example of the process flow of non-contact certification using input unit.

Fig. 4 is the vein, fingerprint, the image example of fatty streak for making infrared light and green light while shining and shooting.

Fig. 5 is an example of the spectral sensitivity characteristic of each elements of RGB of colour TV camera.

Fig. 6 is the explanatory drawin of the detection method for showing position of finger.

Fig. 7 is the explanatory drawin of the detection method of the position of finger contours and finger tip.

Fig. 8 is the explanatory drawin of the method for the posture for showing detection finger.

Fig. 9 is the explanatory drawin of the method for the rolling anglec of rotation for showing detection finger.

Figure 10 is suitable for the position of finger of certification, an embodiment of the guide picture of posture to user's induction.

Figure 11 is the explanatory drawin for cutting out processing and three-dimensional correction processing of finger.

Figure 12 is the explanatory drawin of the structurally and operationally method of the authentication device for the induction that can easily carry out finger position.

Figure 13 is by tapping finger so as to easily induce the explanatory drawin of the operating method of the authentication device of position.

Figure 14 is the non-contact life of reflection formed as second embodiment, by infrared light supply and a thermal camera One embodiment of object authentication device.

Figure 15 is an example of the process flow in second embodiment.

Figure 16 is the explanatory drawin of the detection method of the pitch angle of the finger in second embodiment.

Figure 17 is one that the authentication device structure closer to the distance of subject is judged by the lightproof area of infrared light supply Embodiment.

Figure 18 is the figure of the apparatus structure for the authentication device for illustrating the 3rd embodiment.

Figure 19 is the figure of the apparatus structure for the authentication device for illustrating the 4th embodiment.

Figure 20 is the explanatory drawin of the bootstrap technique of the shooting finger in the 4th embodiment.

Figure 21 is the explanatory drawin that the bootstrap technique of finger is shot by one-handed performance in the 4th embodiment.

Figure 22 is the explanatory drawin of another embodiment of the bootstrap technique of the shooting finger in the 4th embodiment.

Figure 23 is the explanatory drawin of the structure of the authentication device in the 5th embodiment.

Figure 24 is the explanatory drawin of the structure of the authentication device in sixth embodiment.

Figure 25 is the explanatory drawin of another apparatus structure of the authentication device in sixth embodiment.

Figure 26 is the explanatory drawin of the structure of the authentication device in the 7th embodiment.

Embodiment

Embodiment 1

Fig. 1 is the integrally-built of the biometrics authentication system for the blood vessel for having used finger for showing first embodiment Figure.In addition, it goes without saying that the present invention can not also be configured to system, but be configured to carry all or part of structure In the device of housing.Device is configured to include the person authentication device of authentication processing, can also be configured to authentication processing and exist Carried out and blood-vessel image acquisition device, blood-vessel image extraction element dedicated for the acquirement of blood-vessel image outside device.In addition, It can also be the embodiment as terminal as described later.

The Verification System of embodiment 1 includes：It is input unit 2, authentication processing portion 10, storage device 14, display unit 15, defeated Enter portion 16, loudspeaker 17 and image input unit 18.

Input unit 2 includes the camera device 9 of the light source 3 and setting for being arranged at its housing inside housings.In addition, deposit By the part of the image processing function in authentication processing portion 10 or in the image processing function comprising image input unit 18 and The referred to as situation of image processing part.In short, authentication processing portion 10 is the general name performed with the processing unit of the relevant processing of certification, tool It is standby：Judge the judging part of the distance of organism (finger) and system or the posture of organism (finger), in display unit etc. from image Obtained with the state control unit of the amendment instruction of the distance of organism (finger) or the posture of organism (finger), from shooting Image remove the garbage removal portion of garbage (wrinkle, background etc.), from the obtained image zooming-out characteristic information of shooting Feature extraction unit, the control that is compareed the characteristic information extracted with the registration data pre-saved in storage device Portion etc..

Light source 3 is, for example, the light-emitting components such as LED (Light Emitting Diode, light emitting diode), to showing defeated Enter 1 irradiation light of finger on the top of device 2.Camera device 9 shoots the image shown in the finger 1 of input unit 2.Separately Outside, finger 1 can also be more.Image input unit 18 obtains the image photographed by the camera device 9 of input unit 2, and will Acquired image is input to authentication processing portion 10.

Authentication processing portion 10 includes：Central processing department (CPU：Central Processing Unit) 11, memory 12 with And various interfaces (IF) 13.

CPU11 is stored in the program of memory 12 by performing so as to carry out various processing.Memory 12 is stored to be held by CPU Capable program.In addition, memory 12 provisionally stores the image inputted from image input unit 18.

Authentication processing portion 10 and exterior device are attached by interface 13.Specifically, interface 13 and input unit 2, Storage device 14, display unit 15, input unit 16, loudspeaker 17 and image input unit 18 etc. connect.

Storage device 14 has prestored the registration data of user.Registration data is for compareing user Information, is, for example, image of finger vein pattern etc..In general, the image of finger vein pattern is will to be distributed mainly on the palmar side of finger Subcutaneous blood vessel (referring to vein) shooting for shadow pattern image.

Display unit 15 is, for example, liquid crystal display, is the output shown to the information received from authentication processing portion 10 Device.

Input unit 16 is, for example, keyboard, and the information inputted from user is sent to authentication processing portion 10.Loudspeaker 17 is The output device being transmitted with acoustic signal (for example, sound) to the information received from authentication processing portion 10.

Here, display unit 15 and loudspeaker 17 are as indicating organism using the user of the Verification System An example of the distance of (finger) and system, the modified device (instruction unit) of the posture of organism (finger), it is of the invention and unlimited Due to the device.

In addition, each processing unit of described above can both be carried out whole processing by a CPU, can also be by each processing unit Use CPU.

Fig. 2 is the figure of the structure of the input unit for the biometrics authentication system for illustrating first embodiment.

Input unit 2 shoots the surface or subcutaneous living body feature that are distributed in finger.Input unit 2 is by filling Put housing 21 to surround, be configured with 2 camera devices 9 inside it, multiple infrared light supplies 31 and visible light source 32 are being taken the photograph in addition As alternately being configured in annular shape around device 9, finger 1 can be equably irradiated via opening portion.Infrared light supply 31 irradiation infrared lights, it is seen that light source 32 irradiates visible ray.The wavelength of visible light source 32 can from substantially 450nm to 570nm degree, Arbitrarily selected into green wavelength from indigo plant.In addition, light source 31 and light source 32 each can be with arbitrary intensity illuminations.As An example of specific wavelength, light source 31 select the infrared light of 850nm, and light source 32 selects the green wavelength of 550nm.It is in addition, each The light-emitting component of wavelength can also be integrally formed.Acrylic panel 22 is embedded with opening portion, having prevents dust etc. from invading dress Put inside and the physically effect of the component inside protective device.If the further, since light that light source 31 and light source 32 irradiate Reflect then it cannot be seen that subject, therefore all light sources can be lighted in temporal sequence and continuous in acrylic panel 22 Shooting subject, by HDR (High dynamic range), technology is synthesized each image, so as to obtain areflexia component Clearly subject.In addition, light source 31 and light source 32 can not also be around camera devices 9, but imaged at two Circular either clathrate is configured between device 9 can also surround two camera devices 9 and be configured to circular or grid Shape.

Camera device 9 is colour TV camera, with multiple light of the wave band with sensitivity in visible ray and infrared light Element.Camera device 9 for example with the indigo plant three kinds of CMOS or CCD elements of (B), green (G), red (R) with light reception sensitivity, this A little elements are configured to clathrate as known as Bayer arrays.In addition, each element of RGB also has near infrared light There is sensitivity.The sensitivity of each photo detector, for example, by for indigo plant near 450nm, for it is green near 550nm, for red The sensor of the peak value with light reception sensitivity is formed near 620nm.In addition, set captured coloured image as RGB each The RGB color image form that can independently obtain of planes of color.In addition, it goes without saying that camera device 9 can also be set to have Have more than the multispectral video camera of the photo detector of 3 wavelength.In addition, possessing in camera device 9 makes all of the output of light source 3 The wavelength of light passes through and blocks the bandpass filter 33 of other frequency bands, blocks unwanted veiling glare and improves image quality.In this reality The light for applying the wavelength for only making 550nm and 850nm in example passes through.

Input unit 2 can shoot the various organisms in the skin for being present in finger tip by irradiating multiple visible rays Feature.For example, there are the pattern of the melanin such as fingerprint, the wrinkle of epidermis, the wrinkle in joint, vein, spot or black mole, the figure of blood The pattern (referred to here as fatty streak) of fat lobule etc. of sample, subcutaneous tissue.

One implementation of the process flow that Fig. 3 is authenticated while being the posture using stereo camera presumption organism Example.

First, system promotes user to show finger (S401).Then, irradiate infrared light and green light and two colours are taken the photograph Camera starts to shoot (S402) while obtaining synchronous.At this time, can in the case where observing luminance saturation because of exterior light etc. To implement exposure adjustment, it is set as the time for exposure that luminance saturation disappears.Then from the image that photographs implement subject away from From/three-dimensional shape measurement (S403).Then judge in the range of away from authentication device distance set in advance with the presence or absence of shot Body (S404), does not show organism if being considered as there is no if, is back to processing (S401) of beginning etc. until showing organism Untill.On the other hand, it is considered as in the case of having subject in the range of to set a distance and illustrates organism, implements the appearance of finger Gesture detection handles (S405).Then, result is judged using posture to carry out the induction of finger processing (S406).At this time in finger Position, angle with contemplate deviate or finger it is inadequately curved in the case of, shown using picture, guide LED, buzzer etc. User is induced, and posture is repeated and judges and induces, so as to appropriate shooting state.On the other hand, In the case where being determined as appropriate posture (S407), processing is cut out implement finger-image based on the pose information of finger (S408).In the pose information of finger, the position comprising finger tip, the root of finger, uses one of the object as certification Or the positional information of more fingers cuts out each image.Then, for position of the finger-image cut out according to finger, court Carry out the magnifying power of correction chart picture to, distance, and the posture of finger is standardized into (S409).Then, after posture is standardized Finger-image extraction living body feature (S410).Then, living body feature (S411) is compareed, is calculated similar with registration data Degree, judges whether user is registrant (S412).

Here, each processing module is described in detail.

First, the distance to subject, shape measure (S403) describe in detail.In the present embodiment, using two Colour TV camera, shoots the reflected image of near infrared light and green light, using the stereoscopic vision (Stereo matching) that make use of its parallax Technology implement range measurement.In addition, set the inner parameter and external parameter for converting the coordinate system between two video cameras It is known.In addition, set the synchronization that can also obtain the shooting between video camera.At this time, by using general stereoscopic vision Method, so that the arbitrary pixel of left video camera be associated with the arbitrary pixel of right video camera, can be obtained with pixel unit The distance of subject.As the associated method of the pixel between two video cameras, it can utilize and be based on having used parts of images each other Between template matches, the feature of SIFT (Scale-invariant Feature Transform, scale invariant feature conversion) Association of point etc..Thereby, it is possible to obtain the range information or shape information of the subject jointly shot by two video cameras, energy Enough grasp the stereochemical structure of subject.

In general Stereo matching, in the case where strong edge information is not present in subject, the pixel between two video cameras Association become difficult, it is difficult to obtain accurate range information.Particularly, for the organisms such as finger infrared light reflectogram Edge is weaker as, thus is difficult to obtain corresponding points.Therefore, in the present embodiment, make infrared light and green light and meanwhile shine and Shoot its reflected light, while obtain the information of the vein that can be shot by infrared light etc. and the edge that can be shot by green light Trickle fingerprint, fatty streak.

Fig. 4 be make infrared light and green light and meanwhile shine and photograph vein, fingerprint, fatty streak image an example. The 61 middle finger vein 62 of reflected image of infrared light by unintelligible is shot, the spy of the almost unobvious, edge etc. such as fingerprint of epidermis Sign point is fewer.On the other hand, it is barely perceivable in the reflected image 63 of green light and refers to vein 62, but is able to observe that finger Line 64, joint wrinkle 65, fatty streak 66.Since these characteristic quantities have strong edge, spy general SIFT etc. can be passed through Levy point extracting method and be readily available characteristic point, it is possible to increase the associated precision of the coordinate of two images in Stereo matching.

But fingerprint, joint wrinkle, fatty streak are many in the regional area situation similar with the edge of other positions.Cause This, when by general image procossing to ask for characteristic point and its characteristic quantity, in the area of the topography of extraction characteristic quantity It is possible to not to be identified with the characteristic quantity of other characteristic points in the case of less and detects the corresponding points of mistake.Therefore, exist In the case of line feature as fingerprint, joint wrinkle, end or the branch point of line are only extracted, and granular as fatty streak Feature in the case of, multiple edges of fingerprint or fatty streak comprising surrounding and the feature of the point is extracted from slightly wider scope Amount.The main reason for mistake thereby, it is possible to suppress as characteristic point corresponds to and commonly known opening problem.

Here, to by irradiate at the same time infrared light and green light so as to and meanwhile obtain an implementation of infrared and green reflected image Example is described.

If in the state of shade has organism while irradiating infrared light and green light, these light on the surface of organism or Absorbed or reflected under person surface, its reflected light reaches camera device.Thus, the intensity of reflected light near organism surface point Cloth is visualized.At this time, due to having irradiated infrared light and green light at the same time, in each photo detector of the RGB of colour TV camera Reflect the luminous intensity.

Fig. 5 is an example of the spectral sensitivity characteristic of each element of the RGB of colour TV camera.As shown in the drawing, than ripple The infrared light region of more long wavelength, each element of RGB become roughly the same sensitivity profile near long 850nm.On the other hand, It is higher for green light, the sensitivity of only G elements.At this time, R, G, B these three elements are asked for by following simultaneous equations Light income (be respectively Ir, Ig, Ib) and the LED to shine wavelength.

[numerical expression 1]

Ir=I₅₅₀*Wr(550)+I_85o* Wr (850) formula 1

[numerical expression 2]

Ig=I₅₅₀*Wg(550)+I_85o* Wg (850) formula 2

[numerical expression 3]

Ib=I₅₅₀*Wb(550)+I_85o* Wb (85O) formula 3

Wherein, reflected light component when I550, I850 represent the monomer of green light and infrared light respectively, Wr (λ), Wg (λ), Wg (λ) represents the sensitivity of each element of R, G, B under wavelength X respectively.By these equations, green light and infrared light can be asked for Reflected light component (I550 and I850).Wherein, it is two relative to the quantity of unknown number and the quantity of equation is three, due to Can ask for two unknown numbers by two equations, thus by will be taken out from three equations three kinds of results of two into Row is average, so that the error equalization in will be contained in the solution of unknown number.

In addition, as other embodiment, it can also use and infrared light and green light are alternately lighted and use respective shadow The method of picture, but compared with the manner, the frame rate meeting drop by half of shooting, and the subject shot in the image of two wavelength Position be possible to slightly offset from.Therefore, the mode of the present embodiment, can shoot infrared image and fatty streak, thus energy at the same time It is enough that subject is shot at high speed, and will not generation position deviation in the interframe of subject, it is thus possible to realize high-precision The Stereo matching of degree.

In addition, usually, it is range measurement the result is that with pixel based on stereoscopic vision and what is obtained by the above method What unit was implemented, but the influence of the noise due to image, there is also the situation that inaccurate result is produced with pixel unit.Cause This, can also be by pattern cut, the application of speckle wave filter, so as to spatially average to implement noise remove.

Then, the embodiment that the posture of the organism to implementing in step S405 detects describes.For finger Presentation method contemplate various situations.For example, having only makes a finger shade on device, opens five fingers screenings The relevant situation of posture that cover, makes digital flexion shade etc. and finger, the position of finger deviate video camera and do not appear before one's eyes suitably Deng with showing the relevant situation in position.Therefore, using above-mentioned subject range measurement handle as a result, estimating each finger Finger tip, the position of root, the posture for showing the fingers such as angle of finger, and obtain and be used for based on presumption result induction user So that show position, posture or the information being corrected on the image as appropriate.

Fig. 6 is an embodiment of the detection method for showing position of finger.First, as shown in Fig. 6 (a), to by distance map Integrated towards x-axis and obtain the projection brightness distribution of average value and 101 asked for as 100.Wherein, got over here, setting distance It is near then the brightness value of range image is higher, the right direction of image is defined as x-axis direction in addition, lower direction is defined as y-axis side To (paper the is inboard) direction that will be far from video camera is defined as z-axis direction.At this time, the projection brightness of the part of finger is observed The projection brightness distribution that distribution becomes the part that subject is not present in higher value, the gap of finger etc. to become relatively low value.This Outside, due to multiple fingers of appearing before one's eyes in the picture sometimes, thus there are multiple for the fluctuating of projection brightness.At this time, if such as Fig. 6 (b) institutes Show the peak value for asking for projection brightness like that, then can ask for the coordinate in the x-axis direction of multiple fingers.

If it will appreciate that the care shown position, next define centered on each finger substantially of multiple fingers Region 102.As shown in Fig. 6 (b), the Region Of Interest 102 of finger is following region, i.e. for 101 setting threshold of projection brightness distribution Value, using the center more than the maximal point of threshold value as x-axis direction, and has arbitrary width sufficiently wide compared with the width of finger Degree.In addition, the height in region is set to the height identical with range image 100.Three fingers are have taken at the same time in the present embodiment, Therefore three Region Of Interests are set.

Alternatively, it is also possible to which the more fingers photographed not to be all set to the object of authentication processing, and only by specific one Root finger is set to deal with objects.Fig. 6 (c) is an embodiment of the processing for detecting a specific finger.If consider the mistake of lens Very in the picture the heart reduce, user by finger shade in the surface of authentication device, it may be considered that being most reasonably to reflect The finger-image at the center of present image is set to deal with objects.Therefore, the finger close to the center of image is only taken out.First, pin To the projective distribution 101 of x-axis, being multiplied by and being then worth bigger weight coefficient 103 closer to the center of image, be transformed to close to image The value of projective distribution at center become higher.Then, the maximum 104 of projective distribution is asked for, which is defined as near figure The position in the x-axis direction of the finger of inconocenter.Thereby, it is possible to detect to appear before one's eyes not to appear before one's eyes in figure in the finger at the center of image The finger of the end of picture.

Then, for each Region Of Interest, the profile detection of finger is implemented.Fig. 7 be detect finger profile and finger tip one The explanatory drawin of embodiment.The profile detection of finger can be according to the side such as the 2 value processing based on discriminant analysis or pattern cut Pixel value in each Region Of Interest is divided into finger and background to take out its border by method, in addition can also be by linking or tracking Mode take out strong edge.At this time, since adjacent finger is also possible to be included in Region Of Interest, thus can also use Closer to picture centre the weight at edge is set higher so that the total amount at edge maximumlly while makes the distance at edge, song The linking method that rate minimizes.Thus, the situation for mistakenly extracting the profile of adjacent finger is reduced, and can be reduced certification The effect of mistake.By the processing, as shown in fig. 7, finger contours 120 can be obtained.

If the finger contours 120 of each Region Of Interest can be detected, the center line 121 of finger is next asked for.Finger Center line 121 can carry out straight line approximation to ask for finger contours 120, or can also be by the finger contours extracted Inside region is defined as finger areas, implements the principal component analysis using each pixel of finger areas as key element, as obtained The direction (principal direction) of first principal component and ask for.Then, ask for referring to using the center line 121 of finger contours 120 and finger Point 122.The intersection point of the center line of finger and the contour line of finger can be defined as finger tip 122 by this, or can also be towards pass The y-axis in heart district domain 102 calculates projective distribution, from the downside of image towards the value of upside investigation Luminance Distribution, will be less than energy first Enough it is determined as being that the y-coordinate of the threshold value of background is determined as finger tip.Also it can determine that even if not detecting finger contours in the method The approximate location of finger tip, therefore processing simplification, high speed can be made.

Processing more than, can detect the x coordinate of each finger, y-coordinate, z coordinate and show position, be furthermore possible to carry Take the position of finger contours and finger tip.

Then, using finger distance/three-dimensional shape measurement result, to detect the posture of finger.The posture of finger is not The x, y, z coordinate of finger, the also bending comprising finger, warpage and the rotation centered on each axis of x, y, z axis are only included, Need to detect these states.In addition, finger is set to be divided into three sections by two articulations digitorum manus, and analyzes the central shaft of each section and be approximately The skeletal structure of the finger of straight line.In addition, in the coordinate system shown in Fig. 6 (a), by by x-axis, y-axis, z-axis each axis centered on Rotation be respectively defined as pitching, rolling, yaw.

Fig. 8 shows an embodiment of the posture of detection finger.As shown in Fig. 8 (a), in this embodiment, have taken for away from From the finger 1 that two joints 150 are included in the Region Of Interest 102 of image, finger is gently bent, also comprising rolling.According to being gone out The position of the finger shown, it is different from the figure of illustration, it is also contemplated that only to include a joint or do not include the situation in joint.This Outside, the region of the inner side of finger contours 120 is set to more bright, is the distance closer to video camera.First, it is right as shown in Fig. 8 (b) Range image carries out partial differential to calculate the normal line vector 140 of each coordinate.Since finger 1 is analogous to the shape of elliptical cylinder Shape, it may thus be appreciated that normal is radially distributed.Then, as shown in Fig. 8 (c), the pixel that the central shaft 121 of finger passes through is investigated Normal line vector.The normal line vector is shown in Fig. 8 (d).Since the middle section portion etc. of finger tip, finger is heaved, normal line vector is towards each Kind direction.At this time, since joint 150 is depression, thus the side of the direction direction intersection of the normal line vector of the position in joint is crossed over To.Therefore, angle formed by the central shaft 121 of these normal line vectors and finger is asked for, the position detection that its direction is inverted is pass The estimated position 151 of section.Finally, for the region respectively saved of 151 disjunction of estimated position by joint, by principal component analysis come The direction most orthogonal with all normal line vectors is asked for, is set to the central shaft 152 of section.Processing in this way, can obtain Obtain the skeletal structure each saved of finger.In addition, the pitching of finger can be to the finger interior zone of the central shaft 152 along section The distribution of distance value carries out straight line approximation, is asked for as its inclination angle.In addition yaw being capable of inclining from the central shaft 121 of finger Tiltedly obtain.

On the rolling of finger, the section of finger can be considered as ellipse, from the finger surface of the part that can be observed Coordinate estimates elliptical major diameter and minor axis, and roll angle is obtained from the major diameter direction and x-axis angulation.Specifically, first As shown in Fig. 9 (a), if definition passes through certain concerned pixel 163 and the finger orthogonal with the central shaft 152 of the above-mentioned section obtained Section 160, then curve formed by the section and finger surface become ellipse 161.At this time, ellipse 161 is not all can Observe, the part that multiple normal line vectors 140 that can only observe can obtain as image are distributed.On the part, Each coordinate of the x, y, z on the surface of finger is known.That is, the state for the part for being only able to observe that ellipse 161 is become.It is logical Cross and be directed to the partial trace, using such as Hough transform, so as to obtain oval 161 major diameter 162 and the length of minor axis with And direction.Then, by concerned pixel 163 is changed on the direction of the central shaft 152 of the section along finger to various positions and The major diameter 162 of many can be obtained, and by asking for calculated whole major diameters 162 and the average value at angle formed by x-axis, so that The average roll angle of the section of corresponding finger can be estimated.Finally, as long as the average roll angle all saved is estimated and put down , the average roll angle of the finger can just be obtained.

Can also be known by the pose information of finger as another embodiment for the pose information for obtaining above-mentioned finger Largely refer to image as teacher's data, based on machine learning such as Random forest, estimated from unknown range image The pose information of finger.

By the above, it will appreciate that the position of x, y, z axis of each finger, the bending of finger, warpage, rotating state.

In addition it is also possible to for each frame in temporal sequence the position to the x, y, z axis of each finger, finger angle of bend, Rotation angle is accumulated, and is added according to these information to calculate with the position of finger, the relevant translational speed of posture or movement Speed, the posture that they are used for finger judge.For example, image can be produced in the case where detecting finger and moving at high speed It is fuzzy, thus user can be induced not with high-speed mobile finger, or finger also has in the case where finger is hardly mobile May accidentally it be taken, thus can also be by promoting mobile finger it is conclusivelyed show the intention being authenticated.In addition, By only allowing to shoot in the case where leave the action of video camera, so as to shooting subject fuzzy tendency substantially The image of consistent image quality.Thus, unitize during registration with image quality during input, it is possible to increase authentication precision.

Then, an embodiment of the display (S406) of the guiding of the induction finger in detail shown in narration Fig. 3.If as described above Finger can be detected shows position, posture, then can induce captured finger to appropriate position.

Figure 10 is suitable for the position of finger of certification, an embodiment of the guide picture of posture to user's induction.Figure 10 (a) it is to guide 180 in the position display finger contours suitable for certification and show to promote the guiding message shown of finger 181 state.In the picture, the image overlap photographed by camera device is shown, if user moves right, finger causes User is easy to operate, then implements the left and right reversion processing of image so that the finger of image also moves to the right.In addition, show herein The image shown can be the range image for being synthesized into the image of left and right cameras, can also be set in left or right video camera The infrared light reflection image of any one or green reflection image.In addition, in order to hide Biont information, can also opponent Refer to region to be filled.

User is according to the display shade finger.But since the general shape of finger contours is not to be suitable for all hands The shape of finger, therefore not necessarily can be consistent with the shape of the profile of the finger of shade.In addition, not only need the position of two dimensional surface Alignment is put, is also required to be directed at appropriate position for video camera distance, thus it is even if consistent with shown finger contours guiding Ground shade finger, also not necessarily shade shows position in appropriate.In addition, the position for being directed at multiple fingers is not easy to, and Not necessarily all finger all comes the position suitable for certification.Induced while therefore, it is necessary to change guiding content successively.

Figure 10 (b) shows an example during three fingers of shade.First, the finger movement arrow of left direction is shown 182 so that be located at by above-mentioned processing and the finger detected in center among three fingers detecting in image The heart.Thereby, it is possible to implement the overall position alignment of finger.Then, as shown in Figure 10 (c), even if being moved for the finger in center To the state at the center of image, also due to position deviation occurs for the finger of left and right, thus notice makes the meaning that finger slightly closes up, And show that finger to the right or to the left moves arrow 182 to induce position near each finger.Similarly, in finger , can also be into the display for exercising the meaning that finger is slightly opened in the case of closing up artificially.

In addition, Figure 10 (d) shows that finger is deviated to the right and rotates and close to camera position and arthrogryposis State.Display is used to make the arrow that finger is moved to the left first, and then display is used for the video camera distance for pulling open video camera distance Indicate icon 183.Then, display is used for the rotation instruction icon 184 for restoring the rotation of finger.Finally, instruction makes finger Stretch the straight meaning in joint.In the case that if finger tip excessively goes to the lower section of picture at this time, it can also show the meaning.

If in this way, inducing the position up and down of finger, video camera distance first, rotation, the hands such as pitching are then induced What is referred to bends and stretches, then can be induced according to from the rough order for showing position to fine adjustment, it is possible to increase induction Efficiency.

As described above, based on above-mentioned posture detection as a result, it is possible to by finger show position and posture is induced to correct Position.

In addition, the shooting radical of finger appropriate in the above-described embodiment contemplates three, but as long as being one to five It can implement to shoot between root.In the case where the radical of finger is more, by carrying out hand of the induction positioned at the rightmost side Refer to and the mean place of finger positioned at the leftmost side becomes image center, the finger of shade integrally makes it flat so as to induce Weighing apparatus converges on field angle well.Alternatively, the radical of the finger of current shade can be also detected in real time, according to the every of each finger A radical, switches the display radical of above-mentioned finger contours guiding 180 and shows appropriate position., can by these processing Expeditiously show multiple fingers, can realize multiple finger certifications of higher precision.

Then, to finger, whether the determination step S407 in appropriate position is described.Although suitable for the hand of certification The position of finger, posture can determine uniquely, if but do not allow the deviation with this, user can spend in showing for finger Time-consuming, convenience declines.Therefore, for shoot the radical of finger, finger x, y, z position, rotate, bend and stretch the ginseng such as state Number, to setting allowed band from the bias of perfect condition, is determined as it being appropriate at the time point for being all introduced into allowed band Posture.In addition, the radical on shooting finger, such as in the case where being set to appropriate value by three, although during registration until Persistently induced untill including three in the picture to ensure data volume, but one can also be allowed during certification.This be because To have unanimously also can determine that it is substantially my situation on probability even if only one, in this way, so as to slow With the restriction of the posture of finger during certification, the operation of high convenience can be realized.

Then, the embodiment for cutting out processing step S408 and three-dimensional correction processing step S409 of finger is chatted State.

In the case of the guiding of finger is implemented as described above, if considering the convenience of user, it is difficult to Realize the shooting under ideal position, therefore the posture that can include finger changes.In this case, due to the variation with posture The deformation of living body feature, the different characteristic quantity of pattern when can obtain from registering.In this regard, for example need by SIFT feature amount Implement to compare Deng the characteristic quantity for deformation robust, or the deformation to original image is corrected.In the present embodiment, with Premised on control based on the less template matches of processing cost, (here mean that finger tip using the pose information of finger, close Section, root be marked as finger three-dimensional shape state) testing result, to the organism image of captured two dimensional surface Geometry conversion is carried out, is corrected to the state of standardization.

Firstly, for the Region Of Interest 102 that the decision of range image 100 is directed to such as Figure 11 (a) Suo Shi, as shown in Figure 11 (b) The finger on the basis of by the above-mentioned fingertip location obtained is generated for infrared light reflection image and green reflection image 200 Cut out image 201.Thus, finger tip becomes and connects with cutting out the top of image 201, and the position deviation of x-axis and y-axis is by school Just.In addition, it is to be directed to by the pose information such as the central shaft each saved of the above-mentioned fingertip location obtained, finger contours, finger What range image 100 was obtained, there is the deviation of the coordinate system relative to the reflected image photographed by the video camera of left and right two. This, if the deviation of the coordinate system between range image 100 and reflected image 200 can calibrate, can by coordinate transform come Ask for.In the case where coordinate transform produces error, the finger contours that can also make to be obtained by range image are on reflected image The edge total amount of whole pixels existing for finger contours is measured while moving in parallel, the edge total amount in finger contours is most Place re-start position correction.In addition, reflected image 200 can be the image of any video camera in left and right or will The image that the image of left and right both sides is synthesized into.It is set to no matter coordinate transform can be carried out in the case of any.

Then, as shown in Figure 11 (c), for the profile for the finger appeared before one's eyes in image 201 is cut out, to the size of image into Row zooms in or out so that the distance value of the finger obtained by range image becomes fixation.The calculating of magnifying power by it is optimal away from Determined from value and the ratio of the distance value of current finger.Thus, though there are z-axis direction deviation, can also make finger Size keep fix.At this time, become by the distance value for the subregion for zooming in or out finger by image section It is fixed, it can also eliminate the inhomogeneities of the distance caused by the pitch angle of finger.

Then, as shown in Figure 11 (d), carry out rotation correction and cause the central shaft 152 each saved of finger towards image Top.Since central shaft 152 has that gradient is different for each saving for finger, each section can also be asked for The average value of central shaft, and rotation correction is carried out so that it is towards the top of image.

Finally, as shown in Figure 11 (e), in the ellipse 161 and its major diameter 162 and image of the section by the above-mentioned finger obtained X-axis become projective transformation carried out to the coordinate of two dimensional surface on parallel direction.The x-axis after projective transformation is illustrated as herein X ' axis, but x-axis is also referred to as later.Thus, even if the shape of the two dimensional image of finger is deformed because of rolling, due to being transformed to The veutro of finger can obtain the image of no deformation towards positive projection image.

Processing more than, can implement the coordinate of x, y, z axis and the correction of direction of rotation of finger.In addition, herein The correction shown is implemented also obtain same effect after feature extraction described later processing.

Finally, feature extraction processing (S410), control treatment (S411) and determination processing (S412) are described.It is special Sign extraction process extracts characteristic quantity respectively for infrared external reflection image and green reflected image this two images.Due in infrared external reflection The information such as vein are able to confirm that in image, are able to confirm that fingerprint, joint wrinkle, fatty streak etc. in green reflected image in addition, because This implements extraction process, the extraction process of textured pattern of general vein pattern, so as to obtain living body feature.Especially for Sharpening line pattern, can also implement passivation mask process as pretreatment.In addition, image has the two of the video camera of left and right It is a, but the image of a video camera can also be used only, in addition the image that can also use both sides implements feature extraction, will be last Obtained pattern is synthesized, and can not also be synthesized in addition and be carried out flexible utilization respectively as independent pattern.In short, due to double The image of side has parallax, and can be utilized as different information, therefore by using the image of multiple video cameras, can Expect that precision improves.

On control treatment, the pattern extracted is sentenced by the method for general template matches or Feature Points Matching Determine similar degree.Moreover, become certification success in the case where obtained similar degree is less than the threshold value determined in advance.Though in addition, Multiple living body features are so obtained by using the image of infrared image, green image, multiple video cameras, but can also be passed through Average similar degree, the average similar degree of the high multiple results of similar degree during to all compareing, to implement the judgement of certification. The redundancy of data can be increased by calculating multiple similar degrees, help to improve precision.

The difference of abductive approach when then, to registration and abductive approach during certification describes in detail.In registration, It is expected to shoot organism in the state of stablize as far as possible.Therefore, organism is caused to preferably showing position in successive induction After shooting state becomes good, registration data is generated.For example, induction finger to meet that captured organism is reflected as far as possible Present center, shoot scope as wide as possible, the less and face video camera of rotation of finger, included in set video camera In the depth of field, finger do not bend.

However, it is possible to be susceptible to be difficult to the position for making finger strictly vacantly alignment, cannot be resonable for user That thinks shows the static situation in position, thus convenience deteriorates.Therefore, derived need has a degree of surplus.

In addition, in the case of geometry correction has been carried out to the image of shooting as described above, actually hand There occurs the image that same image quality might not can be also shot during change for the position of finger.Therefore, obtaining becomes various Multiple registration datas of the posture of finger, carry out obtaining multiple similar degrees, base with compareing for whole registration datas in control Implement final certification in these similar degrees to judge.In the method, since registration data has various finger appearances The change of gesture, the certification that thus be accordingly used in the variation progress robust to posture is effective.At this time, the induction of the finger in registration When, the shooting of fixed finger gesture can be implemented all the time if the benchmark of the induction of finger is identical, therefore, it is difficult in finger gesture In comprising change.

Therefore, initial one carries out induction so that in the highest position shooting of image quality as described above, is then slightly proximal to Video camera, is shot in the higher position of magnifying power.Finally, shot in position of the finger away from video camera.If registration This three images, though then in certification the finger of shade generated relative to normal place it is close or remote with video camera Change, the similar degree of a certain also can higher.As a result, it is possible to the control of robust is realized in the variation to magnifying power.

Similarly, it goes without saying that various combinations can be obtained for the posture of finger, such as shooting is induced to normal place Situation, induction to situation about keeping left than normal place and induction to situation these three situations kept right than normal place etc..Examine Consider the variation of organism according to the characteristic of video camera, the shape of device the finger posture become much larger, if therefore induction to become The posture of dynamic larger finger then can be more efficient.Induction in this way, so as to realize Shandong to the postural change of finger The certification of rod.

On the other hand, in certification, due to requiring stringent position alignment, the operating time is elongated or position alignment occurs It is miscellaneous, thus convenience decline.Therefore, it is big when the allowance of induced devia is obtained than registration, pass through the appearance based on finger Gesture information is corrected to absorb the position deviation.At this time, if can captured in real-time image and implement control system structure, As long as the somewhere during hand shade then is being become the posture close to the finger of registration data, certification will succeed, thus Also being capable of correctly certification even if the larger position deviation of permission.Therefore, the situation of appropriate position is not in even in finger Under also implement authentication processing, can also exception processes flow cause picture position deviate the meaning.

By the above, the organism of high-quality can be shot during registration, during certification can by the operation of high convenience come Implement authentication processing.

Figure 12 is the embodiment for showing easily to induce the structurally and operationally method of the authentication device of the position of finger Figure.In non-contact shooting is implemented, due to shade finger in the air, any part is have taken as video camera, if not confirming Picture is then also difficult to know, is also difficult to grasp the appropriate distance away from video camera in addition.Therefore, as shown in Figure 12 (a), in device Housing 21 sets finger tip leader label 220 and finger root leader label 221.Leader label is by triangular representation, triangle The direction on vertex show finger tip side.Thereby, it is possible to show device and finger direction it is relational.In addition, these marks As long as showing direction and the direction of finger, shape is not limited to triangle.

As shown in Figure 12 (b), the finger tip of finger 1 is placed in finger tip leader label 220 by user, and placement refers to root and makes it Into between two finger root leader labels 221.Then finger is made away from device as shown in Figure 12 (c).If in this way, carry out initial Make finger contact device case 21 and make the remote operation of finger from there, then by the optical axis center of video camera, away from video camera The possibility of optimum distance increase, can realize the shooting of the organism of higher quality.In addition, in the present embodiment will guiding Mark configuration at the center of housing, but can also according to the position of video camera, captured finger radical etc., carry out configuration so that Finger is easily moved to appropriate camera site.In addition, leader label can also be set to the shape with finger on device case 21 The depression that shape is matched and formed, makes finger be placed in the place.

An example of the process flow of the non-contact authentication device described below that leader label is configured with device.First, it is System guides user " please place a finger on device ".Then, light source igniting is made, continuously shot images whiles waits Organism is shown.If organism is placed in finger mounting table, image due to light source strong reflection and saturation.Due to image Whether luminance saturation by the distance of light source and finger to determine, therefore finger be placed in device case 21 by pretesting when Average brightness value, and determine be used for determine whether to be placed with the threshold process of finger, so as to judge the setting of finger.If User places finger according to leader label, then can be judged to placed finger by the setting determination processing of finger.

Then system guides user " please leave finger ".Correspondingly, as shown in Figure 12 (c), if user Make intensity reduction of the finger away from device, then reflected light, it is known that finger is separated with device.Then, surveyed if implementing above-mentioned distance Amount processing, can detect finger and be located at appropriately distance, then from the characteristic quantity of image zooming-out organism at this time.If registration The living body feature is then registered in processing, if authentication processing then carries out and registration data using the living body feature as input Control.

According to the present embodiment, by the way that finger to be positioned over to the leader label of finger, so as to make finger will not be from visual field Angle significantly deviates or shoots organism with significantly deviateing from the appropriate shooting distance away from video camera, therefore can obtain Following effect, i.e. user does not make finger static in the air, and optimal shooting image can be determined in a series of operation.

As shown in the embodiment of abductive approach during alternatively, it is also possible to register described above, stepped on according to acquisition is multiple Numeration according to when registration number of attempt make the changes of threshold of optimal shooting distance so that distance difference is included in registration data Multiple registration datas.Thus, the distance of finger changes when certification, is also able to maintain that similar with registration data Property.

As another operating method of the authentication device for the position for easily inducing finger, there are the method for tapping device. Above-mentioned makes the remote operation of finger if not device is fixed or is then difficult to carry out with the state that a hand is kept.Therefore, take In generation, makes the remote operation of finger, carries out the tapping of above-below direction as follows, i.e. with singlehanded holding meanss, from by same hand Finger tip be arranged on the state of given position and leave a certain distance and return again to.Thus, just can only be clapped by singlehanded operation Take the photograph the organism image as optimal video camera distance.

As shown in Figure 13 (a), device case 21 is held first with person, such as forefinger 240 is placed in the guiding mark of finger tip Note.If sending the instruction of the meaning of tapping finger from system, user starts to touch.The direction of finger such as Figure 13 during tapping (b) become inclined shown in like that to show, therefore the finger-image photographed also to produces the state of pitching and is taken.But As long as finger generally converges on the appropriate angle of the field angle of image, and registration data is preserved with same posture, that It can be used for certification.Therefore, the operation made above and below finger is repeated several times, the timing implementation that field angle is converged in finger is several The shooting opened.Then generate the characteristic of organism from these images and registered.If same behaviour is also carried out during certification Make, then as close to registration when state fingers posture moment and registration data similarity raising so that certification into Work(.

In this way, by carrying out tapping operation, finger shows position and can converge on certain scope, thus and registration data Reappearance rise, and without in the air implement finger position alignment just can easily shoot aerial finger, and then Can be with implementing certification while being hold by one hand device, therefore the convenience of device improves.

Embodiment 2

Figure 14 is to be formed as second embodiment by carrying out luminous light source and a thermal camera to infrared light Reflection non-contact biological realization card device an embodiment.

It is arranged with multiple infrared light supplies 31 around camera device 9, and is equipped with and only makes in the lens section of camera device 9 The bandpass filter 33 that the light near wavelength that infrared light supply 31 is sent passes through.Camera device 9 is thermal camera, can be shot The light reflected by finger 1.

Figure 15 is an embodiment of the process flow in the embodiment.Processing shown in this Fig. 3 with the above embodiments 1 Flow is roughly the same.But it is following aspect with the difference of above-described embodiment 1：Not two using a thermal camera Platform colour TV camera；The range measurement based on stereoscopic vision is not implemented；Not multiple wavelength using the image of single wavelength.With Under, illustrate the summary of process flow.

First, system urges showing (S501) for finger to user.Then irradiate infrared by one while infrared light Video camera starts to shoot (S502).Also implement exposure adjustment in the case of exterior light is stronger at this time.Then according to what is photographed Image implements distance/three-dimensional shape presumption (S503) of subject.Judge in the scope away from authentication device distance set in advance It is interior to whether there is subject (S504), it is considered as does not show organism in case of absence, is waited until showing life Untill object.On the other hand, it is considered as in the case of having subject in the range of to set a distance and is shown with organism, implements finger Posture detection processing (S505).Then, result is judged using posture to carry out the guiding of finger processing (S506).At this time in hand The position of finger or angle deviating contemplate or finger it is inadequately curved in the case of, LED, buzzing are shown or guided using picture Device etc. induces user, and posture is repeated and judges and induces, so as to appropriate shooting state.The opposing party Face, in the case where being determined as appropriate posture (S507), place is cut out based on the pose information of finger come implement finger-image Manage (S508).In the pose information of finger, the position comprising finger tip, the root of finger, uses one of the object as certification The positional information of root or more fingers cuts out each image.Then, the position for the finger-image cut out out according to finger Put, direction, distance carry out the magnifying power of correction chart picture, the posture of finger is standardized into (S509).Then, after posture is standardized Finger-image extraction living body feature (S510).Then, living body feature is compareed (S511), calculating and registration data Similar degree, judge user whether be registrant (S512).

Here, to Figure 15 as the present embodiment and the processing moulds that have differences of Fig. 3 as above-mentioned first embodiment Block is described in detail.

First, the irradiation to infrared light and by video camera shooting processing (S502) describe.In practice biology , can there are various objects around filming apparatus in the scene for realizing card device.Therefore, in infrared external reflection image In shooting, the situation for photographing unnecessary subject beyond finger is also very much.Since unnecessary subject becomes sometimes For correctly detecting the noise of finger, therefore in order to carry out the certification of higher precision, it is necessary to remove unnecessary subject.Make For one of which method, as shown in Figure 14, there is the side for carrying the bandpass filter 33 for only passing through the wavelength of light source 31 Method.On the other hand, in the environment there are the reflected light of identical wavelength region in the case of can not remove the component.Therefore in this reality Apply in example, background removal is carried out by the difference lighted image and extinguish image of light source.

The image for extinguishing light source 31 is shot first.And followed by shoot the image for lighting light source 31.Then, if The image photographed when extinguishing is subtracted from the image photographed when lighting, then can retain the image of shade finger not Become and remove the component for the unnecessary subject appeared before one's eyes brightly irrelevantly with light source 31.Thereby, it is possible to exterior light Influence independently to implement cutting out for finger exactly.In addition, in the case where exterior light component is irradiated in finger surface, pass through difference Point calculating and cancel, therefore can also reduce the malfunction caused by the exterior light of posture described later detection processing.

In addition, in the case where the image for extinguishing light source is extremely bright, it is known that consumingly irradiation has exterior light.At this time, By making light source consumingly irradiate according to its brightness and controlling to shorten the time for exposure of video camera, so as to have even in irradiation Also the measurement of organism can be realized in the environment of strong exterior light.In the case where adjusting exposure, due in finger described later Required optimal parameter change is handled in posture detection processing etc., therefore asks for being directed to each time for exposure in throughout managing in advance Optimal parameter.

In this way, unnecessary background is removed by the background difference of the flicker based on light source in the present embodiment, can Improve the accuracy of detection of finger.But in the method, the frame rate for shooting can halve, therefore be insufficient in frame rate In the case of height, the removal of the unnecessary background based on bandpass filter 33 can also be implemented, or applied in advance in large quantities Obtain finger and background separated infrared external reflection image and by the machine learning of Random forest etc. come separation finger With the method for background.

Then, the measurement (S503) to distance/three-dimensional shape of subject describes in detail.The present embodiment is due to cannot Embodiment described above implements the range measurement based on stereoscopic vision like that, it is therefore desirable to a thermal camera come implement away from From measurement.

As the method estimated with a thermal camera into row distance, exist using the luminous intensity with the reflected light of irradiation The method of relevant inverse square law.Light source is arranged on the periphery of video camera, thus substantially can be considered point light source, so towards hand Refer to the light sent to extend in spherical.The surface area of ball and the radius of ball square are becoming proportionately larger, thus per unit area The energy of light and square being inversely proportional for radius.Therefore, the brightness of subject by subject reflectivity and absorptivity and by The distance of body and video camera is taken the photograph to determine.

The light quantity of irradiation is set to Ii, reflection/absorption coefficient of skin is set to μ, D will be set to the distance of subject, The light quantity of light is set to Io.In addition, the reflected light of finger surface is Lambertian reflection, if assuming, reflected light isotropically expands Dissipate, then following approximate expression is set up.

[numerical expression 4]

Here, due to can in the average reflection intensity μ of investigation irradiation light intensity I i, skin in advance, by away from Solved and will appreciate that with a distance from each pixel of image from D.Wherein, the brightness of irradiate light quantity and light light quantity as image It is worth and is detected, if but directly measured using Ii as brightness value, due to usual luminance saturation, thus by attenuation rate for The wave filter known makes irradiate light quantity Ii shoot the image in the state of physically decaying, and implements the amount for making light quantity Ii increase decay Deng calibration.Alternatively, can also be configured to, carry in the device to irradiation light Ii speculums directly reflected etc., Neng Goujing By above-mentioned attenuation rate irradiation light Ii is directly observed for known wave filter.In this way, can be from the finger of institute's shade Brightness value obtains range image.

In addition it is also possible at the same time set the video camera of mensuration distance and the camera device of the present embodiment to clap exactly Finger is taken the photograph, collects the learning data that the normal solution label of distance is distributed to each pixel in large quantities, based on such as Random Forest Deng machine learning, the range image of finger is asked for from infrared external reflection image.

But actually due to the influence of exterior light etc., it is also possible to the possibility than contemplating brighter image can be photographed Property.In this case, error is included in the absolute value of distance D.Therefore, by asking for at the arbitrary position of image Subject distance be 1, relative distance distribution, also will appreciate that the stereochemical structure of subject.In addition, image is generated in advance Brightness value and distance value rough map table, according to the brightness in the pixel of the subject at the arbitrary position Value reads the distance value, by the way that the distance value is multiplied with the relative distance value of image entirety, can also obtain it is rough away from From image.

Then, (S505) is detected for the posture of finger, narration and the difference of embodiment 1.Here, it can also use above-mentioned The method of acquisition range image implement processing same as Example 1, but as easier method, showing will be infrared Reflected image is transformed to range image and carries out the embodiment of the posture detection of finger.

First, infrared external reflection image is considered as range image, is concerned about by the method shown in Fig. 6 of embodiment 1 Region 102.Wherein, the brightness value of reflected image is more high, and it is nearer to be considered as distance.

, such as embodiment 1 as shown in Figure 6, can be from will be infrared anti-primarily with respect to the position detection in the x-axis direction of finger The coordinate for penetrating the x-axis direction that image is considered as range image and determines to obtain during Region Of Interest obtains.In addition the y-axis side on finger To finger tip position detection, can be obtained by the same method shown in Fig. 7.In addition on the distance (z with video camera The position detection of axis) because inverse square law is set up between the brightness of reflected image and light source distance, exist close between the two Connection, so by the average brightness and the relation of video camera distance of the finger when pretesting makes certain light value shine, can Generation is from average brightness to the map table of video camera distance.Then finger is asked for when implementing the contour detecting of inputted finger Internal average brightness, and using the map table video camera distance is calculated from the average brightness value.Thus, it is excessively bright in brightness value In the case of can detect finger and get too close to video camera, otherwise can detect finger distance in the case where brightness value is excessively dark and take the photograph Camera is too far.Although the value changes due to the influence of exterior light, as long as image when being extinguished by photographic light sources is slapped The brightness caused by exterior light is held, just can relax the influence of exterior light by subtracting the value.

On the detection of yaw, in the same manner as the above-mentioned method shown in Fig. 7, ask for the central shaft 121 of finger and implement Rotation correction so that the axis is parallel with image y-axis.

In addition the detection on pitching, figure 16 illustrates an embodiment of detection method.In this embodiment, be located at finger tip from There occurs pitching rotation on the remote direction of video camera.First, figure is cut out for the finger shot in infrared external reflection image As 300, average brightness projective distribution 301 is calculated towards y-axis.It is herein Iave by the average brightness projective distribution token of coordinate y (y).Because finger tip, away from video camera, Iave (y) is that the value near finger tip is relatively low.Further, since reflected image is with hand Brightness change caused by the trickle bumps that refer to, the presence or absence of blood vessel, therefore trickle fluctuating is also observed in Iave (y). Then, the relative distance Dr (y) shown in following formula 7 is calculated.

[numerical expression 5]

Air line distance ratio value D ' (y)=Wa (1/Sqrt { lave (y) }), (formula 5)

[numerical expression 6]

Vertical distance proportion value D (y)=D ' (y) ^2-Wb (y-yc) ^2, (formula 6)

[numerical expression 7]

Relative distance Dr (y)=D (y)-D (yc), (formula 7)

Wherein, yc is the picture centre coordinate of the y-axis for cutting out image 300 of finger, and Wa is arbitrary proportionality coefficient, and Wb is Calibration factor described later, Sqrt { x } are the square root of x.In addition, by with from light source to advancing on the incline direction of coordinate y D ' (y) is set to apart from proportional value, the value proportional to the distance of the height from light source to the vertical of coordinate y is set to D (y) (with reference to Figure 16 (b)).And at each coordinate y that relative distance Dr (y) is when by the vertical distance definition of coordinate yc being zero Relative distance.

The distance D (y) or D ' that Iave (y) is the coordinate y of less value are estimated as farther out.But as shown in Figure 16 (b) Like that, if before video camera flatly holding plane plate, the D (y) calculated from the average brightness projective distribution 302 of surface plate Should be unrelated with y and as fixing, but brightness value is dimmed at away from light source y-coordinate farther out, therefore D ' (y) changes.Herein If distance according to y-coordinate and if changing it is very inconvenient, therefore distance D ' (y) is transformed to by distance D (y) by formula 6.In formula 6 Comprising parameter Wb, and its D ' (y) when determining to cause by horizontal surface plate shade before video camera in advance is unrelated with y as solid Definite value.

Figure 16 (c) shows the flow of calculating.First, D ' (y) is obtained from Iave (y) as shown in Equation 5 (not recording in figure vertical Axis).Then Dr (y) is transformed it into using formula 6, formula 7.Then, straight line is carried out to obtained Dr (y) by least square method Approximation, obtains the presumption pitching straight line 303 of finger.Finally ask for the gradient of the straight line.Gradient at this time, which becomes, to be estimated Pitch angle, the angle be more than fixed value in the case of, the deviation to pitching is alerted or by being described in detail in embodiment 1 The method of pitch correction infrared external reflection image is standardized.

Finally with respect to rolling, the presence or absence of rolling and its rotation are estimated according to the direction in the joint of finger, the shape of profile Direction.Specifically, the rotation image that the anglec of rotation is known finger is collected in large quantities, for example, by convolutional neural networks to this The roll angle of unknown images is estimated after being learnt.Using as a result, being sent in the case where rotation angle exceedes fixed value Warning, or be standardized infrared external reflection image by the method for the roll correction being described in detail in embodiment 1.

Method more than, without obtaining the range image of subject, just can estimate the posture of finger, ask for From the bias of standard gestures, and user can be guided or pattern is corrected using its result.

Figure 17 is the authentication device structure closer to the distance for being determined as subject by the lightproof area of infrared light supply One embodiment.On the top for the camera device 9 for being equipped on input unit 2, shadow shield 320 is configured with so that surrounding camera device 9 Periphery.Shadow shield 320 is arranged to hide the inner side of the infrared light supply 31 of configuration circumferentially, and a part for infrared light is carried out Shading.At this time, if finger 1 produces the region 321 that light does not reach as illustrated close to input unit 2.If observation figure at this time Picture, then can photograph dark and brighter around it image of the immediate vicinity of camera device 9.Therefore, define using image center as The different dual circle of the radius at center, the region surrounded by the circle to inner side with the circumference by inner side and the circumference in outside it is flat Method that equal brightness is compared etc., the center for carrying out detection image is dark and situation that its periphery is brighter.Detecting the situation In the case of can determine that excessively near for finger, user is alerted.In addition, the area by measuring the dark region in center Or radius, it can also estimate the distance of finger and video camera.

Then, the feature extraction processing (S510) in embodiment 2 is described in detail.In the present embodiment due to only obtaining To the reflected image of infrared wavelength, therefore cannot implement shown in the above embodiments at such image for multiple wavelength Reason.Therefore, the vein pattern easily appeared before one's eyes in the reflected image of near-infrared, joint reticulation patterns, finger contours shape are extracted in As characteristic quantity.First, in order to extract vein, using the line of the long side direction of finger is strengthened strongly and removes noise It is passivated mask.Thus strengthen the stronger vein pattern of the tendency advanced on the long side direction of finger, further obscure others Information.By implementing from the image as the prior art and the extraction process of finger vein pattern that generally utilizes, so as to from The efficient and stable finger vein pattern of infrared external reflection image extraction.Then, using the side orthogonal with long side direction to finger To line strengthen strongly and remove the passivation mask of noise so that enhancing joint reticulation patterns.If then similarly extract Pattern, then can extract the line pattern of particularly joint wrinkle.Due to the shape of the line pattern, the first joint and second joint it Between distance have individual, therefore authentication information can be acted effectively as.In addition finger contours information can pass through above-mentioned reality The finger contours detection process shown in example is applied similarly to obtain.By the processing, vein, joint wrinkle, finger can be obtained The living body feature of multiple mode (modality) of profile.

Obtain these information and be registered as multiple organism patterns, separately compareed or made their position Relation has to be compareed while restriction, finally by the technology of general multi-modal organism authentication by the control of each mode As a result merged and obtain final authentication result.Thus, it is high-precision using can also be realized while the image of single wavelength The certification of degree.

In addition, as the relevant other embodiment of range measurement with subject, can also implement continuously to subject The method that the stereochemical structure of subject is measured while shooting.First, when starting shooting, system draws user Lead so that finger is moved along various directions towards input unit, and continuously shoot the image.If by template matches, with The Feature Points Matching that SIFT starts in the image photographed subject how to have carried out movement be detected, then can measure The 3D shape of the subject.This is the technology of commonly referred to as Structure from motion (SfM), thus, it is possible to obtain Obtain the stereochemical structure on subject surface.But, it is necessary to carry out implementing the feature corresponding to non-rigid in the case of finger flex Point control etc. is corresponding.

In this way, in the method for moving organism, even if being the life at the back side that can not be observed under certain specific posture Object features, solid can be also obtained from the information of the amount of movement and three-dimensional structure if shooting is made it possible to by mobile Living body feature, therefore stronger control can be realized in the movement of organism.

Embodiment 3

Figure 18 is the reflection formed as the 3rd embodiment, the light source by sending visible ray and a colour TV camera Non-contact biological realizes an embodiment of card device.

Multiple visible light sources 341 are arranged with around colour TV camera 340, it is red to be equipped with blocking in the lens of video camera The low-pass filter 342 of outer light.The wavelength of visible light source 341 for example for indigo plant near 450nm, for it is green near 550nm, For the red peak value near 620nm with luminous intensity, the luminous intensity of each wavelength can be individually adjusted.

Authentication processing process is identical with the above embodiments 2.As the difference with embodiment 2, it is following aspect：Use Colour TV camera and not thermal camera；Captured image not single ripple using the visible reflectance image of multiple wavelength Long infrared external reflection image.Here, processing distinctive to the present embodiment describes in detail.

In the present embodiment, it is different from foregoing embodiment, organism can be clapped with the colour TV camera of RGB Take the photograph.Visible ray is different from infrared light, has the characteristic in the deep for being difficult to reach organism surface, can handle but then more A wavelength, therefore many Biont informations can be shot according to the optical characteristics of the bio-tissue of skin surface.

In skin histology, there are the blood in fingerprint, the wrinkle of epidermis, the wrinkle of articulations digitorum manus, melanin, capillary The various living body features such as liquid, artery, vein, subcutaneous fat, the absorptivity of light, reflectivity are respectively different.Therefore, by thing The reflection characteristic of preceding each bio-tissue of investigation, irradiates multiple visible rays and is carried out at the same time observation, can independently extract each life The information of object tissue.As its method, can use black to separate by the independent component analysis of three images for RGB The method of pigment and blood, multi-layer biological bulk measurement technology.In addition, vein is easily observed in red image, green or blue Image in easy observation articulations digitorum manus, thus carry out by the difference of two images or with arbitrary ratio the group of division arithmetic etc. Close, articulations digitorum manus component can be removed from red image so as to only obtain purer vein image, therefore can more clearly measure Living body feature.

In this way, in the reflection non-cpntact measurement using the light source of 3 wavelength and colour TV camera, can be obtained by infrared light It must be difficult to the Biont information observed, therefore can obtain improving the effect of authentication precision.

In addition, it is not only that can utilize in the case that visible ray can also shoot the video camera of infrared ray, it goes without saying that In addition to the light source of RGB, additionally it is possible to add infrared light supply to be shot.Thereby, it is possible to shoot to be present in organism surface The vein in deep etc. information, by increasing capacitance it is possible to increase can be used in the living body feature of certification and improve authentication precision.

In addition, can also apply in the present embodiment shown in above-mentioned second embodiment based on inverse square law The range measurement of subject.But there are three wavelength for light source in the present embodiment, it is thus possible to independently calculate it is respective away from From.For example, the light of three wavelength for irradiation, is asking for the reflection characteristic of average skin respectively in advance, imaged with colour Machine shoots respective reflected light at the same time.At this time, as shown in Equation 4 by decay and the decay of the distance based on subject based on reflection Relation, the relation of the luminous intensity of the luminous intensity that is radiated in each wavelength with observing is described.If the finger table based on reflection The average attenuation of light on face is it is known that then only the distance of subject becomes unknown number.Thus, ask for being shot for three wavelength The distance of body.The result of range measurement under three wavelength is different due to the influence of exterior light, measurement error etc., but substantially It is contemplated that as near normal solution value.In order to equalize its error, obtained three distance values are averaged and It is set to final result.Thus, compared with the distance value only estimated by infrared light as described above, the influence of error can be relaxed, Thus it can realize more accurately range measurement.

Embodiment 4

Figure 19 is to realize card device as the reflection non-contact biological in the 4th embodiment, smart phone or tablet Structure an embodiment.Smart phone 360 possesses interior shooting camera 362 in the top of color liquid crystal panel 361 and in its back of the body Face possesses outer shooting camera 363, shoots the flash of light projection white light being equipped near camera 363 for shooting of glistening outside Source 364.Interior shooting camera 362 and outer shooting camera 363 are the colour TV cameras that can shoot visible ray, are equipped with inside Block the optical filter of infrared light.In addition, also it is equipped with focusing adjustment function, exposure adjustment function.

Authentication processing process in the embodiment is substantially identical with the above embodiments 2.As the difference with embodiment 2 It is following aspect：There is a situation where that reflection source cannot be utilized；In the presence of sensory characteristic, the light that cannot grasp video camera in advance The situation of the wavelength spectrum in source.Here, processing distinctive to the present embodiment describes in detail.

First, the shooting processing to subject and range measurement, the posture detection of finger describe.On subject The extraction process of range measurement, posture detection and living body feature, in the same manner as the method shown in above-mentioned 3rd embodiment, energy It is enough to improve the method for the precision of range measurement, multiple living body features inside extraction skin using using colour TV camera Method.It is further possible to the color of the finger of palmar side, the inspection by the color effective for finger areas are individually registered in advance Survey.

On the other hand, by when the outer shooting camera 363 of the white light source of flash of light can be utilized to shoot organism, in thing Before will appreciate that the luminescent spectrum of white light source in the case of will appreciate that organism characteristic quantity measurement in calibration, distance The luminous intensity of each wavelength components in measurement, therefore the detection, range measurement, the feature of organism of finger areas can be improved The precision of extraction etc..

In addition in the case where using interior shooting camera 362, since liquid crystal panel 361 is located at the organism in face of institute's shade Position, therefore liquid crystal panel can also be used as organism light source for shooting.Thus, by with the above embodiments 3 The same method of the device of the visible light source using RGB shown, can realize the range measurement of subject, biological bulk measurement. In addition, liquid crystal panel can control shining for planar, therefore light source position can be arbitrarily set, such as only make the upper half of panel Millet cake is bright to be shot, and then only makes lower half millet cake bright to shoot.Because luminous position is known, the present invention's The three-dimensional shape of subject can also be obtained in structure by the technology of photometric stereo vision.

But the luminous intensity of liquid crystal panel is general weaker compared with the light-emitting components such as LED, it is contemplated that cannot be with sufficient The possibility of intensity illumination subject.Therefore, it is necessary to implement to shoot using the ambient light around authentication device is distributed in. In this case, since ambient light not can determine that the position of light source, therefore, it is difficult to implement the distance of the intensity based on image brightness values Measurement.In addition the spatial distribution of light is different also according to environment, thus is difficult with colouring information and obtains vein, melanin, fat The Biont informations such as fat line.Therefore, taken finger-image is decomposed into the image of each planes of color of RGB, for each A image asks for average brightness, does not utilize the image of the planes of color in the case of darker than given threshold value and gives up, for It is determined as the image of bright enough planes of color, implements enhancing filtering, the enhancing of articulations digitorum manus filtering of vein etc. as described above, from And multiple living body features can be obtained.

In addition, in the case where being able to detect that the joint of finger, its position can observe hemoglobin denselyer Color, so by obtaining the joint of finger and the difference of surrounding color, can extract the color of hemoglobin.At this time, thing The preceding color that general hemoglobin is observed in various ambient lights, as color spaces such as rgb space, HSV spaces Information shows.Then, to the color space information of hemoglobin that will appreciate that from current image with prepare in advance it is each The color space information of kind hemoglobin is compared, and can be estimated as with the associated ambient light of most similar color space information The spectrum of current ambient light.Ambient light can also be estimated by this method, and effective for foregoing finger region detection, life Object measures.

In addition on finger posture detect, in above-mentioned second embodiment with from close to the position of video camera irradiation have Premised on reflected light, the position of finger can be estimated, but in the present embodiment, particularly is implementing to have used interior shooting camera Do not have light source in the case of 362 shooting, thus the premise can not be utilized.Therefore, the reality that the posture as finger detects Example is applied, profit is with the following method, i.e. the region for the finger for shoot in detection image using the marginal information of colouring information, image.

First, range sensor of pose information with learning data calculating range information and finger can be used as etc. one Rise, finger is shot using the video camera of the smart phone of the embodiment.Thus, to being photographed by the video camera of smart phone The normal solution value of image, marking path information and pose information.At this time, the number of species of finger, the number of species of posture, image Number obtains as much as possible.Then, using general machine learning, implement to be used to only estimate hand by the image of smart phone The range information of finger, the parameter learning of pose information.For example, the range information of the pixel of certain coordinate, pose information in order to obtain, While the intensity of the flesh color component of its pixel value and neighbouring image border amount are used for evaluation of estimate, learned based on random forest Practise.Then, by the image of smart phone put into study after random forest and traveled through, by each pixel classifications to background Any one of region and finger interior region, and if then then judging referred to herein as point, root or articular portion in finger Deng.Again while also estimate distance value.The presumption result of distance value since the influence of noise etc. can produce deviation by each pixel, Therefore processing can also be implemented so that spatially smoothened.

, can be as above-mentioned by by background parts blacking in this way, if interior zone and the background area of hand can be separated Second embodiment shown in like that application be not accompanied by range measurement the brightness value based on reflected image finger posture detection Method.Therefore, by combining the extracting method of the living body feature shown in above-mentioned 3rd embodiment, it can realize and recognize Card processing.

Figure 20 is an embodiment of the bootstrap technique for using the outer shooting camera shooting finger in the present embodiment.User's handle The hand 380 for holding device side holds smart phone, and the finger 1 of opposite side is shot.At this time, retouched on liquid crystal panel 361 Finger contours guiding 381 is painted with, user opponent's fingering row is shot so that consistent with the guiding.System calculates finger in real time Pose information, is shot at the time point for entering allowed band, is authenticated based on the image zooming-out living body feature amount. At this time, if expressing the position of the root of finger in finger contours guiding 381 as shown in the figure, video camera and finger are easily adjusted Rough distance.Such display can be applied to the arbitrary embodiment of the present invention.

In the case where generating position deviation, arrow can also be passed through as shown in the above-described embodiment Carry out guide position bias direction Deng sense of vision means, or using loudspeaker, the vibrating elements for being equipped on smart phone, pass through sound Sound, vibrating function notify the deviation of finger.Especially with regard to video camera and the distance of organism, it is difficult to carry out drawing on picture Lead, if the guiding sound that the vibratory output of vibrating function is set stronger or will be sent from loudspeaker if more big by position deviation The volume of sound sets bigger etc. to guide, then can intuitively enforcing location be aligned.

Figure 21 is an embodiment of the bootstrap technique with outer shooting camera shooting finger that can be implemented by one-handed performance.Profit User holds smart phone with one hand as shown in the figure, and system instruction first blocks outer shooting camera 363 with finger tip.User's hand Refer to 1 and block outer shooting camera 363.If Verification System detects this case, then finger contours guiding is shown in picture 381, prompt the recurved meaning of finger so that finger is farther out with shooting.User holds smart phone 360 as shown in figure 21 While according to guiding make finger leave it is outer shooting camera 363 and it is curved backward.

At this time, as shown in liquid crystal panel 361, finger 1 is appeared before one's eyes in a slanted state.If detecting the state, system is real Apply shooting processing, authentication processing.By applying the bootstrap technique, so as to also can easily implement the position of finger even if one hand Alignment, can realize the organism authentication described in the present invention in smart phone.

In addition it is also possible to induce keep 1 recurved state of finger constant and jiggle the behaviour of smart phone 360 Make.At this time, if shooting consecutive image, the finger 1 of shade remains unchanged at identical position detection and arrives, and mirrors in image Unnecessary background 400 move with rocking action.Therefore, movement is determined whether by each pixel, by without mobile area Domain is considered as finger areas, other regions are considered as background area, the region detection thus, it is possible to implement finger.According to this method, Finger areas can be detected exactly by simple image procossing, it is possible to increase authentication precision.

Figure 22 be make use of it is interior shooting camera in the case of bootstrap technique an embodiment.Interior shooting camera is mostly set It is set to and easily face is shot in many cases, and is difficult to shoot the finger near liquid crystal panel.In this case can Finger is carried out shade separated by a distance in the air, since liquid crystal panel is blocked by finger, thus be difficult to confirm.Therefore, guide For finger to be placed in the given position of liquid crystal panel, then to aerial mobile finger.

First, if starting certification, the guiding 420 for the meaning that finger is set is shown in picture such as shown in Figure 22 (a) and used In the finger position guiding 421 for placing finger.In the present embodiment place finger direction be set to it is left tiltedly on direction, but Direction can be changed according to custom hand, in addition can also be along positive horizontal direction shade.Placed herein followed by person's alignment Finger.Liquid crystal panel 361 is touch panel, if thus detect finger be placed with specified position, then such as Figure 22 (b) It is shown, show finger to the guiding 422 for lifting the meaning opened in the air.At this time, placed since finger is tiltedly upper to the left, in order to hold Easily see guiding and shown finger in picture lower-left to the guiding 422 for lifting the meaning opened in the air.But there is also picture by hand Refer to the possibility of covering, thus the guiding based on sound can also be used at the same time.User is according to the guiding by finger to aerial It is mobile.Then, as shown in Figure 22 (c), finger enters the field angle of video camera.Implement to clap if system detects this case Take the photograph.

Start setting up the guiding 421 of finger position have finger is induced to the effect for the position for being easily accessible field angle Fruit.In addition, the guiding 421 can also be set to the finger contours of multiple fingers in the case where wanting the multiple fingers of shooting.This When, if carrying out induction so that finger to be opened for appropriate angle, it can obtain easily while shoot the effect of multiple fingers. Bootstrap technique in this way, can also realize the authentication department of above-described embodiment in the case of interior shooting camera is used Reason.

As other bootstrap techniques, induction is carried out so that touching interior shooting camera or interior shooting camera repeatedly with finger tip Neighbouring liquid crystal panel, and tapping is detected by the variation of the brightness of touch panel or interior shooting camera.Also can at this time Shot with successive frame, the distance of finger is detected by the distance measuring unit of the measurement of the width based on finger contours etc. Appropriate situation, from the finger-image photographed in appropriately distance, carries out the feature extraction of organism and implements certification.

Abductive approach based on such tapping also has following and other effects in addition to being fairly simple operation：Energy Enough positions for easily inducing finger；Without making finger static in the air, thus position alignment becomes easy；And organism with The distance of video camera changes all the time, thus there are the timing that organism reaches the distance suitable for shooting, can be with appropriate focusing Shoot organism.

It is equipped with addition, the smart phone shown in the present embodiment is referred in the widely available interior shooting picture of current point in time The situation of machine, outer shooting camera.It is self-evident, as long as being equipped with smart phone or tablet can shoot positioned at liquid crystal panel The video camera of neighbouring finger, thermal camera, multispectral video camera, apart from video camera etc., it is possibility to have effect is taken the photograph using these Camera embodiment described above implements range measurement, the posture detection processing of finger, authentication processing like that.In addition, the present embodiment Smart phone is not limited to, also can similarly be implemented even if for tablet, portable game machine.

In addition it is also possible to towards the user of aerial position alignment difficulty, have and maintained a certain distance with video camera And the finger mounting table of finger can be loaded, the pedestal of smart phone or tablet can be embedded into be authenticated.Such as Administrative services window or bank window etc., are set fixed as the terminal that ordinary user can simply operate sometimes Tablet.In this case, portability is not special problem, thus by combination settings foregoing pedestal and tablet, so that not Need dedicated authentication device, additionally it is possible to which certification is implemented by shirtsleeve operation.

Embodiment 5

Figure 23 shows the implementation that the authentication device shown in above-mentioned first to 3rd embodiment is applied to video-game Example.

Figure 23 (a) sets the posture for the body for being used for detecting player to be caught to operate the movement of game on the top of TV 440 Device 441 is caught, wherein possessing the input unit 2 of certification.Thus the finger 1 of user is shot, implements certification.User uses Motion capture device 441 or the game console that in addition prepares etc. operate game, but wherein input unit 2 detects user Finger and implement certification.It can show and guide in TV 440 during certification, or loudspeaker etc. can also be used to carry out sound and drawn Lead.

Figure 23 (b) is to be equipped with the embodiment of input unit 2 in the controller 442 of video game.User holds control Device 442 carries out game operation, is provided with the input unit 2 of certification herein, can be used in game control as described above, right The clearing of the charging of project.

Figure 23 (c) is mounted in an embodiment of portable game machine.Possess in the middle body of portable game machine 443 The input unit 2 of certification.Authentication method is identical with the example of the carrying to smart phone shown in the above embodiments 4.

In above-mentioned each game machine, authentication application as shown below can be implemented.First, can simply realize as follows Content, i.e. user and automatically reading when game is implemented in identification preserve data, either implement on game machine or network Content access control, or manage and limit the object for appreciation playtime of game.Furthermore it is possible to game is made according to the result of certification The various parameters change of role, or the charging progress Automatic-settlement to payment project etc..Particularly, additionally it is possible to according to registrating number The characteristic of game role, such as the translational speed of increase role or enhancing attack are controlled according to the similar degree with input data Power, the interface that can be used as game control utilize.

If in addition, registering multiple fingers, a finger is only shown in certification, then can be according to the finger shown Species carrys out partition functionality.For example, if action RPG game, then be the attack of percussion system when showing forefinger, show middle finger When for the attack based on magic, be then physical recovery based on magic etc. if the third finger.Such action cuts under present situation Changing is made choice in menu screen, or the instruction of complexity is inputted using multiple input key, through this embodiment being capable of letter Change these operations.If in addition, will individually it be stepped on along the Biont information of the periphery of the finger of the central shaft of the finger shown Note, then, can respectively and the roll angle with showing by showing finger in input unit and finger is rolled around axis center Corresponding registration data is spent to be matched.Hereby it is possible to according to matched registration data detect the roll angle of finger, because The operation can be acted effectively as the interface to game by this.For example, can realize makes car race game according to the roll angle of finger In engine rotating speed simulation ground change etc..

Embodiment 6

Figure 24 be while walking only shade finger just can be authenticated walk through gate inhibition (Walkthrough Gate an embodiment of the authentication device of the present invention is applied in).

User makes finger 1 close to input unit 2 as shown in the figure when close to certification gate inhibition 460.Input unit 2 is continuously Filmed image, the posture for detecting finger while, implement certification at high speed.If certification success, baffle 461 are opened so that are utilized Person can pass through.The guiding of finger can be implemented by liquid crystal panel 462 or loudspeaker 463.In liquid crystal panel 462, The guiding of finger is not only, can also show result of charging implemented when authentication result, discrepancy gate inhibition etc..

Registration can both have been implemented in certification gate inhibition 460 and carry CPU and memory inside gate inhibition and preserve registrating number According to can also be set in different places be equipped with the registered terminal of input unit 2 in addition, registration data is arrived via network transfers Certification gate inhibition 460.In addition, registration data can also be stored in server in addition set etc..In addition, certification both can be in certification Implement inside gate inhibition, image conveyer to server can also be implemented on the server.In addition it is also possible in certification gate inhibition 460 The equipment for carrying wireless communication, user carry Wireless IC tag etc. when passing through gate inhibition, will be stored in Wireless IC tag Registration data, which is transferred on certification gate inhibition 460, implements certification.

It is thick while walking due to being the non-contacting authentication device of reflection according to the structure of the present embodiment Slightly shade can be also authenticated, it is possible to increase the handling capacity of gate inhibition.In addition, by application above-described embodiment in based on light The method of the difference image of the flicker in source and exposure adjustment function, so that even if without can also implementing to shoot, can expand The application range of device.

In addition, situation in bigger devices such as the admission gate inhibitions for allowing to set extensive facility, setting in important management , can also in addition to the authentication device of the reflected light mode of the present invention in applying etc. strictly in the case of management security etc. It is set to be combined with the authentication device of the hybrid power of the general device through light mode.

Figure 25 illustrates an embodiment of the apparatus structure of transmission/reflection both sides' formula.Figure 25 (a) is in gate inhibition's certification Device 480 is vertically configured with the structure through light source 481.If user is from shade finger 1 on the right side of attached drawing, through light source 481 and reflection source 482 at the same time irradiation light.Infrared light is irradiated through light source 481, reflection source 482 irradiates green and red visible Light.It is respectively 550nm, 620nm, 850nm to set green, red and infrared wavelength herein.Camera device 9 is with infrared sensitivity RGB color video camera, while the light to passing through and reflecting carries out light.At this time, according to the spectral sensitivity characteristic of video camera, on Stating the luminance components of three wavelength can describe as follows.

[numerical expression 8]

Ir=I₅₅₀*Wr(550)+I₆₂₀*Wr(620)+I₈₅₀* Wr (850) formula 8

[numerical expression 9]

Ig=I₅₅₀*Wg(550)+I₆₂₀*Wg(620)+I₈₅₀* Wg (850) formula 9

[numerical expression 10]

Ib=I₅₅₀*Wb(550)+I₆₂₀*Wb(620)+I₈₅₀* Wb (850) formula 10

In addition, the definition of mark is identical shown in 1~formula of formula 3 of embodiment 1.By being solved to the simultaneous equations, The I550 and I620 of reflected light component and the I850 through light component can be obtained, certification can be implemented using each image. Alternatively, it is also possible to respectively respectively carry a video camera make it possible to individually shooting pass through light and reflected light, additionally can with divide When mode switch the light source of irradiation, but the former is there are the increased problem of cost, the latter there are frame rate decline as class Topic.Therefore as being shown in the present embodiment with a video camera equally shoots the light of different wave length and carries out separating treatment Method can be inexpensive and be shot at high speed.In addition in this embodiment due to the place of shade finger be it is open, Also can simply it be operated even if for first user.

In addition Figure 25 (b) is the mode that space in the device shows finger.Finger 1 is put into infrared light supply 481 with taking the photograph As between device 9, light is shone through from top, and reflected light is irradiated from below, image and reflectogram are passed through thus, it is possible to shoot Picture.Finger can be inserted on the right side of the attached drawing to the left, but as shown in attached drawing hypomere from device side can also move finger and into Enter.The movement so that without prejudice to finger is opened wide in device side at this time.In addition, in the side inputted, mutually somewhat wrong The position opened is provided with multiple photosensitive sensors 483.It can be reacted when the photosensitive sensor is immediately below finger passes through, from diagram Distance between the deviation and sensor of the timing in the reaction time of four sensors, is capable of translational speed, the finger of sensing finger Angle etc..This situation is received, prediction central time of the finger by video camera moment, camera device 9 is suitable for shooting Timed shooting finger.At this time, the time for exposure is obtained extremely short so that the shake of the image caused by the movement of finger will not be produced, In addition light quantity is also correspondingly strongly irradiated in the short time.In this way, only by making finger at high speed by just can Instantaneously certification.

In the either type of the present embodiment, the living body feature of the both sides through light and reflected light can be obtained, therefore It can expect that precision improves the raising with antifalsification.In addition, clearly referring to vein by that can be obtained through light, pass through reflected light Even in organism shade also can stably be realized certification in the case of the position that cannot be irradiated through light, therefore also can The effect of authentication device can more simply be utilized by accessing user.

Embodiment 7

Figure 26 is the embodiment that the authentication device of the present invention is applied to mouse.Left button is equipped with mouse 500 501 and right button 502, in addition roller 503 is equipped with central portion.And the input unit 2 of above-mentioned certification is in roller 503 Near, it is arranged on the central portion of mouse 500.The shape of the form fit finger of mouse 500 and become curve, from finger tip side to The inclination of mouse central portion is opposite with the inclined gradient from mouse central portion to palmar side.When holding mouse 500.Finger 1 and input unit 2 distance it is close, if passing through the inclined relational and if making finger 1 stretch as the crow flies as shown in the figure of mouse The shooting distance with video camera needed for shooting is readily obtained, enough scopes can be shot.In addition, food is stretched in figure Refer to, but it is self-evident, by stretching other fingers such as middle finger, these fingers can be shot.In addition input unit 2 can also be taken It is loaded in the left button 501 or right button 502 of mouse.

The operation of the stretching, extension finger can simply carry out in the state of mouse is held, therefore various when operating PC Situation such as PC log in when self acknowledging, settlement on networks when self acknowledging, each application software user's access right really In recognizing etc., the convenience of user can not be damaged and realized in the operation of natural PC.

In addition, above-mentioned all authentication devices are recorded by taking finger as an example, but as long as being that the skin of organism can To be arbitrary position, such as to can be applied to face, ear, palm, the back of the hand, wrist, arm, head, foot etc. arbitrary Position.Handled in this case, it is necessary to carry out optimal posture detection according to each position, but it is self-evident, by a large amount of Method that the progress machine learning afterwards of teacher's data is collected on ground etc. can be realized.

Description of reference numerals

1 refers to, 2 input units, 3 light sources, 9 video cameras, 10 authentication processings Portion, 11 central processing departments, 12 memories, 13 interfaces, 14 storage devices, 15 is aobvious Show portion, 16 input units, 17 loudspeakers, 18 image input units, 21 device cases, 31 infrared light supplies, 32 visible light sources, 33 bandpass filters, the reflectogram of 61 infrared lights Picture, 62 finger veins, the reflected image of 63 green lights, 64 fingerprints, 65 joint wrinkles, 66 fatty streaks, 100 range images, the distribution of 101 projection brightness, the care area of 102 fingers Domain, 103 weight coefficients, the maximum of 104 projective distributions, 120 finger contours, 121 hands The center line of finger, 122 finger tips, 140 normal line vectors, 150 joints, the presumption in 151 joints Position, the central shaft of the section of 152 fingers, the section of 160 fingers, 161 is oval, and 162 is ellipse Round major diameter, 163 concerned pixels, the guiding of 180 finger contours, 181 guiding message, 182 Refer to mobile arrow, 183 video cameras distance instruction icon, 184 rotation instruction icons, 200 is anti- Image is penetrated, 201 cut out image, and 220 finger tip leader labels, 221 refer to root leader label, 240 forefingers, 300 fingers cut out image, 301 average brightness projective distributions, 302 planes The average brightness projective distribution of plate, the presumption pitching straight line of 303 fingers, 320 shadow shields, 321 light The region not reached, 340 colour TV cameras, 341 visible light sources, 342 low-pass filters, 360 smart phones, 361 color liquid crystal panels, 362 interior shooting cameras, 363 shoot picture outside Machine, 364 flash of light projection white light sources, the hand of 380 grasping device sides, 381 finger contours are drawn To lead, 400 unnecessary backgrounds, 420 fingers set guiding, and 421 finger positions guide, 422 fingers, which are lifted, opens guiding, 440 TVs, 441 motion capture devices, 442 controllers, 443 portable game machines, 460 certification gate inhibitions, 461 baffles, 462 liquid crystal panels, 463 loudspeakers, 480 gate inhibition's authentication devices, 481 pass through light source, 482 reflection sources, 483 photosensitive sensors, 500 mouses, 501 left buttons, 502 right buttons, 503 rollings Wheel.

Claims (10)

1. a kind of authentication device, it is characterised in that possess：

Storage part, it keeps and the relevant characteristic information of organism；

Instruction unit, it indicates the amendment of the distance of the organism and/or the posture of the organism：

Light source, it is to the organism irradiation light；

Image pickup part, it shoots the light reflected by the organism；

Judging part, it judges and the distance of the organism and/or institute according to the image imaged by the image pickup part State the posture of organism；

State control unit, it is based on the judgement, the instruction unit is carried out and the distance of the organism and/or described The amendment instruction of the posture of organism；

Feature extraction unit, it extracts the characteristic information from described image；With

Comparing part, it is similar with the characteristic information that is stored in the storage part that it compares the characteristic information extracted Degree.

2. authentication device according to claim 1, it is characterised in that

The light quantity that the state control unit is irradiated according to the light source and the light quantity received by the image pickup part, to the instruction The amendment that portion carries out the posture with the distance of the organism and/or the organism indicates.

3. authentication device according to claim 2, it is characterised in that

The relative distance for being used as 1 positioned at the distance of the arbitrary position of image is distributed by the state control unit by asking for, right The instruction unit carries out the amendment instruction of the posture of the organism.

4. the authentication device according to any one of claims 1 to 3, it is characterised in that

Possess：Garbage removal portion, it removes garbage from described image.

5. authentication device according to claim 4, it is characterised in that

The garbage removal portion passes through to lighting the light source and the image imaged and the figure for extinguishing the light source and imaging As being compared, to remove the garbage.

6. authentication device according to claim 4, it is characterised in that

The light source has multiple, the light of the respectively different frequency of the plurality of light source irradiation,

The garbage removal portion is described to remove by being compared to the image imaged with respectively different frequency Garbage.

7. the authentication device according to claim 5 or 6, it is characterised in that

The judging part, the state control unit, the feature extraction unit, the comparing part and the garbage removal portion, Performed by a processing unit.

8. authentication device according to claim 7, it is characterised in that

The instruction unit is the amendment of the posture for the distance and/or organism that the organism is indicated by image Display unit or the modified of the distance of the organism and/or the posture of the organism is indicated by sound Sound generating unit.

9. authentication device according to claim 8, it is characterised in that

The characteristic information is and the fat of subcutaneous tissue or fingerprint or vein or articulations digitorum manus or finger contours Relevant information.

10. a kind of authentication method, it is characterised in that possess：

Irradiating step, to the organism irradiation light；

Image pickup step, shoots the light reflected by the organism；

Judgment step, according to and the image that images, judged by the image pickup step with the distance of the organism and/ Or the posture of the organism；

Based on the judgement, the distance of user's instruction to authentication device and the organism and/or the organism The step of amendment instruction of posture；

Characteristic extraction step, characteristic information is extracted from described image；With

Step is compareed, compares the characteristic information extracted and the similar degree of the characteristic information kept in advance.