JPH08185519A - Method and device for collating figure - Google Patents

Abstract

PURPOSE: To provide a method and device for collating figure with which an individual can be surely identified by improving the accuracy of collation for a previously registered pattern and a newly inputted pattern as the patterns at one part consisting of a human body. CONSTITUTION: The figure collating device 10 is provided with an input means 20 for picking up the images of patterns, control means 30 equipped with a first storage part 33 for storing the previously picked-up images of plural patterns respectively as basic and auxiliary registered images, second storage part 34 for holding the newly picked-up image of the pattern as an input measured image, and collation judging part 36 for judging whether the input measured image matches with the basic or auxiliary registered image or not, and collating means 40 for calculating the value of correlation between the basic or auxiliary registered image and the input measured image. In this case, when the matching of the input measured image with the auxiliary registered image is decided after the mismatching of the input measured image with the basic registered image is decided, the control means 30 updates the auxiliary registered image by using the auxiliary registered image and input measured image from which the minimum correlative value is removed.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an information processing technology for secretly using important information or valuables, and is based on pattern recognition of a part of a human body pattern, for example, a fingerprint. The present invention relates to a person verification method and apparatus for identifying a person.

[0002]

2. Description of the Related Art Conventionally, in a technique of performing fingerprint collation as one of techniques for performing person collation, a personal fingerprint that is registered in advance and a newly input personal fingerprint are recorded. Individuals are identified by matching. However, between the time of registering the fingerprint in advance and the time of newly inputting the fingerprint, the fingerprint of the fingerprint is dealt with in response to various factors related to the installation state of the finger, that is, the wetness of the finger, the arrangement, and the tension. Differences may occur in the arrangement, distortion, and breakage of the measurement image. That is, when the reproducibility of the fingerprint with respect to the measurement image is low, there is a problem that the collation accuracy of the fingerprint is significantly reduced.

One way to solve this problem is to
A guide for holding the finger arrangement in a fixed state is installed in the fingerprint input means. However, since the guide that does not correspond to the individual difference in the shape of the finger does not sufficiently compensate the arrangement of the finger, it is not possible to sufficiently improve the fingerprint matching accuracy.

Further, based on measured images of a plurality of fingerprints, individual fingerprints registered in advance, evaluation criteria for determining collation of fingerprints, etc. are determined. However, when collating with a newly input fingerprint, only one measurement image or one level is referred as an evaluation criterion for the registered fingerprint or collation, so that the collation accuracy of fingerprint can be sufficiently improved. Not done. Regarding such prior art, Japanese Patent Laid-Open No. 63-149776.
"And Japanese Patent Laid-Open No. 63-149777.

Furthermore, as a fingerprint of an individual registered in advance, a plurality of measurement images reflecting different installation states of fingers are held. Therefore, a newly input fingerprint that has been evaluated for higher matching is newly registered by adding or updating. However,
In this way, the measured images of fingerprints that are registered by adding or updating will converge to the tendency corresponding to the finger placement that appears as a personal habit, so if the finger placement is rarely different from the daily tendency, the fingerprint The matching accuracy of may be greatly reduced. In addition, such a prior art is described in detail in, for example, Japanese Patent Laid-Open No. 3-142685.

Therefore, the present invention has been made in view of the above problems, and improves the collation accuracy between a pattern registered in advance and a pattern newly input as a part of a pattern forming a human body. By doing so, it is an object of the present invention to provide a person verification method and an apparatus for surely identifying an individual.

[0007]

In order to achieve the above-mentioned object, a person collation method of the present invention is a person collation method for identifying an individual on the basis of collation for a part of patterns constituting a human body, (A) A first step in which the pattern is imaged a plurality of times in advance and the images of the patterns are divided and stored as either a basic registration image or an auxiliary registration image, and (B) the pattern is newly added. The second step of capturing an image of the pattern as an input measurement image and (C) matching the input measurement image with the basic registration image based on the correlation value calculated between the basic registration image and the input measurement image Alternatively, in the third step of determining the mismatch, (D) when the mismatch of the input measurement image with the basic registration image is determined in this third step, the correlation value calculated between the auxiliary registration image and the input measurement image is calculated. Based A fourth step of determining a match or mismatch of the input measurement image for the auxiliary register image,
(E) When the coincidence of the input measurement image with the auxiliary registration image is determined in this fourth step, the minimum correlation value is calculated based on the correlation values calculated by all the combinations selected from the auxiliary registration image and the input measurement image. Fifth updating and storing the auxiliary registration image and the input measurement image excluding those having
And the steps of.

In the first step, the pattern is imaged in advance and the image of the pattern is stored as a basic registration image, and then the pattern is newly imaged and the image of the pattern is held as an input measurement image for basic registration. When it is determined that the input measurement image coincides with the basic registration image based on the correlation value calculated between the image and the input measurement image, the input measurement image is stored as the auxiliary registration image separately from the auxiliary registration image. Alternatively, the process may be repeatedly performed until the existing number of auxiliary registration images matches a preset number.

Further, the auxiliary registration image may be an image of a plurality of patterns individually picked up.

Further, the pattern may be a fingerprint formed on the surface of a human finger.

In order to achieve the above-mentioned object, the personal collation device of the present invention is a personal collation device for identifying an individual on the basis of collation with respect to a part of patterns constituting a human body,
(A) Input means for picking up the pattern, (B) First storage section for storing the images of the plurality of patterns picked up by the input means respectively as a basic registration image and an auxiliary registration image, and by the input means Control means configured to include a second storage unit that holds the image of the pattern captured as the input measurement image, and (C) the basic registration image or auxiliary registration image and the input measurement image input from the control means. And a collation unit that performs a correlation calculation on and to calculate a correlation value between the basic registration image or the auxiliary registration image and the input measurement image. Here, the control means, based on the correlation value calculated between the basic registration image or the auxiliary registration image and the input measurement image input from the matching means, whether the input measurement image matches or does not match the basic registration image or the auxiliary registration image. When the matching of the input measurement image with the basic registration image is determined and then the matching of the input measurement image with the auxiliary registration image is determined, the auxiliary input from the matching means is performed. Based on the correlation values calculated by all the combinations selected from the registration image and the input measurement image, the auxiliary registration image and the input measurement image excluding the one having the smallest correlation value are updated and stored as the auxiliary registration image. It is characterized by

The auxiliary registered image may be an image of a plurality of patterns that are individually picked up.

The pattern may be a fingerprint formed on the surface of a human finger.

Further, the collating means is configured to include an optical lens for performing two Fourier transforms on the basic registration image or auxiliary registration image displayed in parallel and the input measurement image. It may be a feature.

[0015]

In the person verification method and apparatus according to the present invention, first, a basic registration image in which an input measurement image for newly capturing and holding a part of a pattern forming a human body is previously captured and stored. Match against. At this time, based on the correlation value calculated between the basic registration image and the input measurement image, it is determined whether or not the input measurement image matches the basic registration image. Here, when the matching of the input measurement image with the basic registration image is determined, the individual who provided the basic registration image and the individual who provided the input measurement image are identified as the same person.

On the other hand, when it is determined that the input measurement image does not match the basic registration image, the input measurement image is subsequently collated with the auxiliary registration image in which the pattern is imaged in advance and stored. At this time, it is determined whether or not the input measurement image matches the auxiliary registration image based on the correlation value calculated between the auxiliary registration image and the input measurement image. Here, when it is determined that the input measurement image does not match the auxiliary registration image, it is determined that the individual who provided the basic registration image and the auxiliary registration image and the individual who provided the input measurement image are not the same person.

On the other hand, when the coincidence of the input measurement image with the auxiliary registration image is determined, the individual who provided the basic registration image and the auxiliary registration image and the individual who provided the input measurement image are identified as the same person. . At this time, based on the correlation values calculated by all the combinations selected from the auxiliary registration image and the input measurement image, the auxiliary registration image and the input measurement image excluding those having the smallest correlation value are updated as the auxiliary registration image. And store.

In order to preliminarily capture a part of the pattern constituting the human body and divide and store the images of a plurality of patterns as the basic registration image and the auxiliary registration image, first, the pattern is preliminarily imaged. The image of the pattern is stored as the basic registration image. Subsequently, the input measurement image newly capturing and holding the pattern is collated with the basic registration image. At this time, based on the correlation value calculated between the basic registration image and the input measurement image, it is determined whether or not the input measurement image matches the basic registration image. Here, when it is determined that the input measurement image matches the basic registration image, the input measurement image is stored as an auxiliary registration image separately from the auxiliary registration image. Such a process is repeatedly executed until the existing number of auxiliary registration images matches the preset number.

Further, when the auxiliary registration image is an image of a plurality of patterns individually picked up, the difference in the correlation value calculated by all the combinations selected from the auxiliary registration image and the input measurement image becomes large. The comparison between the image and the input measurement image becomes easy.

Further, when the pattern is a fingerprint formed on the surface of a human finger, the pattern has a personal characteristic, so that the individual can be surely identified.

Further, when the collating means includes an optical lens for performing the Fourier transform twice on the basic registration image or the auxiliary registration image displayed in parallel and the input measurement image, the correlation calculation is performed. Since the speed of does not depend on the gradation of the image of the pattern, collation with a pattern including breaks and distortions is also effective.

[0022]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS The structure and operation of one embodiment according to the present invention will be described in detail below with reference to FIGS. In the description of the drawings, the same elements will be denoted by the same reference symbols, without redundant description.

The person collation apparatus of this embodiment performs pattern recognition on a fingerprint, which is one of a part of patterns constituting a human body, and judges the collation result based on a certain evaluation standard. It functions as a fingerprint collation device for identifying an individual.

As shown in FIG. 1, the person collating apparatus 10 includes:
An input unit 20 for picking up a fingerprint, a control unit 30 for managing registration and search for a fingerprint image, for determining collation of the fingerprint image, and a collation unit 40 for performing an optical correlation calculation for the fingerprint image are provided.

As shown in FIGS. 1 and 2, the input means 20
Includes an operation system for the operator to actually operate, a display system for displaying the operation status to the operator, a control system for instructing an operation corresponding to the operation input by the operator, and a fingerprint of the operator. It is composed of an optical system for imaging.

The operation system of the input means 20 is such that the power is turned on and off.
A power switch 21a for setting the FF, a registration mode button 21b for selecting a registration mode for a fingerprint image, and a verification mode button 21 for selecting a verification mode for a fingerprint image.
c and a keyboard 21d for inputting an operator identification code
And a guide 21e for fixing and installing the operator's finger.

Further, the display system of the input means 20 includes a level indicator 22a and an LED (light emitting diode) lamp 22b for sequentially displaying the progress of the actual operation to the operator, and various operation instructions and fingerprints described later. A display 22c for sequentially presenting the operator with the processing results of image collation or registration is included.

Further, the control system of the input means 20 outputs the identification code input by the operator to the control means 30, outputs the processing result of registration or collation of the fingerprint image to the above-mentioned display system, and at the same time performs the above-mentioned operation. It includes a control unit 23 for controlling driving of an optical system, which will be described later, corresponding to the state of the system, and an adjusting unit 28 for adjusting the gain and contrast and outputting the fingerprint image input by the operator to the control unit 30. There is.

Furthermore, the optical system of the input means 20 is incident from the light source 24 on the prism surface which is not in contact with the light source 24 for generating the illumination light based on the drive signal input from the control section 23 and the operator's finger. A prism 25 that totally reflects the illumination light and emits the illumination light scattered as a measurement light corresponding to the fingerprint pattern on the prism surface with which the operator's finger is in contact, and the measurement light emitted from this prism 25 is sequentially reflected. And four plane mirrors 26a to 26d that bend the optical path,
Refractive lens group 26e for converging the measurement light guided by these plane mirrors 26a to 26d and compensating for its various aberrations.
And a CCD (charge coupled device) 27 that receives the measurement light emitted from the refracting lens group 26e and detects a fingerprint image.

As shown in FIG. 1, the control means 30 is a CPU (central processing unit) for controlling the driving of various devices described later.
31 and an operator identification code input from the input unit 20 are output to the CPU 31, and a processing result of fingerprint image registration or collation described below is output to the input unit 20.
Input from the O (input / output) interface 32, a first storage unit that constantly stores the fingerprint image input from the input unit 20 as a basic registration image or an auxiliary registration image, that is, an HD (hard disk) 33, and the input unit 20. Secondly, the fingerprint image is temporarily stored based on the cache memory method.
Storage unit, PM (protect memory) 34, FM (frame memory) 35 for storing two fingerprint images input from the input unit 20, HD 33 or PM 34 simultaneously or in parallel over time, and collation unit 40. The correlation value calculated based on the light intensity of the input correlation image, which will be described later, is compared with the evaluation standard of the collation, and the collation determination unit, that is, the dedicated hardware 36, determines the collation of two fingerprint images. .

The HD 33 stores a memory area consisting of a basic registration area for storing a plurality of basic registration images and a auxiliary registration area for storing a plurality of auxiliary registration images based on an identification code. It is a device. This HD33
The transfer rate of one fingerprint image from the FM to the FM 35 is, for example, about 100 ms.

Further, the PM 34 is a storage device configured as a RAM (Random Access Memory) having a capacity of 4 Mbytes, for example, and storing 64 kbytes at the maximum by assigning 64 kbytes to one fingerprint image.
The speed of transferring one fingerprint image from the PM 34 to the FM 35 is, for example, about 40 ms.

The FM 35 has, for example, a capacity of 64 kbytes.
Or VRAM (video ram) with 128kbyte
It is a storage device configured to store one or two fingerprint images as gray scale data of 254 × 254 bytes.

Further, the dedicated hardware 36 uses the collation means 40 to perform the two-time Fourier transform on the two fingerprint images, and based on the light intensity of the primary light of the correlation image,
The correlation value between the two fingerprint images is calculated, and the correlation value is compared with a threshold value which is an evaluation criterion for collation to determine whether the two fingerprint images match or do not match.

Here, the fingerprint images sequentially input to the FM 35 from the input means 20 are P in the LIFO (last in, first out) system.
Sequentially stored in M34. Also, from HD33 to FM35
The basic registration image and the auxiliary registration image that have been called once are sequentially stored in the PM 34 by the normal LIFO method, and are sequentially stored in the HD 33 when the entire capacity of the PM 34 is used up. As a result, when calling the basic registration image or the auxiliary registration image to the FM 35, the PM 34 is searched first. Here, the HD 33 is subsequently searched for the basic registration image or the auxiliary registration image that has not been stored in the PM 34. Therefore, since the PM 34 functions as a so-called cache memory, the processing speed for the fingerprint image increases as the fingerprint image is frequently called by the FM 35.

As shown in FIGS. 1 and 3, the collating means 40
Is an LD that generates laser light as writing light or reading light for the fingerprint image and its Fourier transform intensity image.
(Laser diode) 41, a flat mirror 42a that reflects the laser light emitted from the LD 41 and bends its optical path, and a half that transmits the laser light guided by the flat mirror 42a to one side and reflects the other to the other side. The two fingerprint images input from the mirror 43a and the frame memory 35 of the control means 30 are displayed, and the half mirror 4
An LCD (Liquid Crystal Display) 44 is provided, which modulates the laser light transmitted through 3a and which corresponds to the two measurement images displayed on the liquid crystal, and transmits the modulated light.

The collating means 40 is a refraction lens group 45 which performs Fourier transform on the laser light emitted from the LCD 44.
a, three plane mirrors 42b to 42d for sequentially reflecting the laser light emitted from the refracting lens group 45a and bending the optical path thereof, and receiving the laser light guided by the plane mirrors 42b to 42d to obtain 2 The orientation of the liquid crystal corresponding to the Fourier transform intensity image of each measurement image is maintained, and the laser light transmitted through the half mirror 43b described later and guided is subjected to modulation corresponding to the Fourier transform image held on the liquid crystal and reflected. And an SLM (spatial light modulator) 46 that

The collating means 40 is a half mirror 43b.
The SLM 46 transmits the laser light reflected by the
Half mirror 43b that reflects the laser light reflected by
And a refracting lens group 45b that performs Fourier transform on the laser light guided by the half mirror 43b, and three plane mirrors 42e to 42g that sequentially reflect the laser light emitted from the refracting lens group 45b and bend the optical path.
And a CCD 47 for receiving the laser light guided by the plane mirrors 42e to 42g and detecting the intensity of the warped light.

The LCD 44 arranges the two fingerprint images input from the input means 20 in parallel (see FIG. 4A).

The SLM 46 includes a first substrate, a first transparent electrode layer, an optical address layer, a mirror layer, a first alignment layer,
A phase modulation SLM that is formed by sequentially laminating a light modulation layer, a second alignment layer, a second transparent electrode layer, and a second substrate, and that functions as a real-time spatial filter. In this SLM 46, a voltage is applied to the optical modulation layer based on the resistance distribution of the optical address layer that changes corresponding to the intensity distribution of the laser light, that is, the writing light, guided from the LCD 44 through the refraction lens group 45a. As a result, the refractive index distribution of the light modulation layer changes in accordance with the orientation of the molecular axes of the liquid crystal that constitutes the light modulation layer, so Fourier transform intensity images for the two fingerprint images displayed on the LCD 44 are recorded (Fig. 4 (b)). Therefore, when the laser light guided from the LD 41 through the half mirror 43b, that is, the reading light, travels back and forth through the light modulation layer by being reflected by the mirror layer, the refractive index distribution of the light modulation layer, that is, the writing light. Undergoes a phase modulation corresponding to the intensity distribution of.

Further, the CCD 47 causes the laser light guided from the SLM 46 via the refracting lens group 45b, that is, the correlation image of the two fingerprint images displayed on the LCD 44 to be 0.
Receives only one of the two primary lights that appear near the next light, and
The light intensity of the next light is detected (see FIG. 4 (c)).

The collating means 40 constructed in this way is
This is a JTC (joint Fourier transform correlator) that optically executes mutual or autocorrelation operations by arranging individual fingerprint images in parallel and performing Fourier transform twice.

As shown in FIG. 5, the capacity of the FM 35 is 12
In the case of 8 kbytes, the FM 35 is divided into two memory areas α and β having a capacity of 64 kbytes. Here, the data of the two fingerprint images read from the HD 33 or PM 4 is written in the FM 35 by the two scanning lines (illustrated by the solid line) whose access address is limited to one of the memory areas α and β. . Therefore, FM35
Data of two fingerprint images are simultaneously stored in the memory areas α and β, respectively (see FIG. 5A). Then, the data of the two fingerprint images read out from the FM 35 as usual is 1
Written on the LCD 44 by book scan lines (shown in solid lines). Therefore, the LCDD 44 simultaneously displays two fingerprint images (see FIG. 5B).

As shown in FIG. 6, the capacity of the FM 35 is 64.
If it is kbyte, the memory of LCD44 is 64kb in capacity.
It is divided into two memory areas α and β having yte. Here, the data of the two fingerprint images sequentially read from the HD 33 or the PM 34 is sequentially written in the FM 35 by one scanning line (illustrated by a solid line). Therefore, the FM 35 stores the data of the two fingerprint images over time (see FIG. 6A). Subsequently, the data of the two fingerprint images sequentially read from the FM 35 is written in the LCD 44 by one scanning line (illustrated by a solid line and a dotted line) whose access address is sequentially switched to one of the memory areas α and β. . Therefore, the LCD 44 simultaneously displays two fingerprint images (see FIG. 6B).

As shown in FIG. 7, when the FM 35 is composed of two fume memories α and β having a capacity of 64 kbytes as a double buffer, the memory of the LCD 44 has two memory areas α and β having a capacity of 64 kbytes. Is divided into Here, the data of the two fingerprint images sequentially read from the HD 33 or the PM 34 are written in the frame memories α and β by one scanning line (illustrated by the solid line and the dotted line). Therefore, FM35
Stores the data of two fingerprint images over time (FIG. 7).
(See (a)). Subsequently, the data of the two measurement images sequentially read from the frame memories α and β is processed by one scanning line (illustrated by a solid line and a dotted line) whose access address is sequentially switched to one of the memory regions α and β. It is sequentially written in the LCD 44. Therefore, the LCD 44 simultaneously displays two fingerprint images (see FIG. 7B). Where H
During transmission of one fingerprint image from the D33 or PM34 to one of the frame memories, one fingerprint image is transmitted from the other frame memory to one of the memory areas of the LCD44, so that the HD33 or PM34 transmits to the LCD4.
The transmission time of the fingerprint image up to 4 is shortened.

Next, the operation of this embodiment will be described.

The person collating apparatus 10 selects one of the basic registration processing for storing a fingerprint image as a basic registration image and the collation determination processing for collating the fingerprint image to identify an individual by the operator. It works.

The basic registration process includes a process for newly creating a basic registration image and a process for updating the basic registration image. The matching determination process includes a process of matching a fingerprint image with a reference image having a basic registration image, a process of matching a fingerprint image with a reference image having an auxiliary registration image, and an auxiliary registration process. It includes a process of newly creating an image and a process of updating the auxiliary registration image.

As shown in FIG. 4, first, the operator turns on the power switch 21a to start the operation of the input unit 20, the control unit 30, and the collating unit 40, and the process proceeds to step 100.

Then, in step 100, a message is displayed on the display 22c asking the operator what he / she wants in the registration mode, that is, whether to execute the basic registration process. Here, when the operator selects the registration mode button 21b to set it to the ON state, the process proceeds to step 110. On the other hand, when the operator rejects the registration mode, the process proceeds to step 160.

Then, in step 160, a message is displayed on the display 22c asking the operator's intention about the collation mode, that is, whether or not to execute the fingerprint collation processing. Here, when the operator selects the matching mode button 21c and sets it to the ON state, the process proceeds to step 170. On the other hand, if the operator rejects the collation mode, the process proceeds to step 100.

As shown in FIGS. 9 and 10, step 1
In step 10, basic registration processing is executed in steps 120 to 152.

First, in step 120, a message requesting the operator to enter the identification code is displayed on the display 22c. Here, the operator uses the keyboard 2
If the identification code is entered using 1d, step 12
Move to 1.

Then, in step 121, the identification code input by the operator is searched in the data stored in the HD 33. Here, if the identification code is not registered in advance by the administrator, the basic registration process itself ends. On the other hand, when the identification code is registered in advance by the administrator, the process proceeds to step 122.

Then, in step 122, the basic registration image attached to the identification code input by the operator is searched in the data stored in the protect memory 34 or the HD 33. Here, if the basic registration image is not stored, the process proceeds to step 124, and the process of newly creating the basic registration image is executed. On the other hand, if the basic registration image is stored, the process proceeds to step 140, and the process of updating the basic registration image is executed.

Subsequently, in step 124, the guide 21
A message requesting the operator to place a finger on the prism 25 corresponding to the shape of c is displayed on the display 22c, and the light source 24 is turned on. Here, when the operator places the finger on the prism 25, the process proceeds to step 126.

At this time, the illumination light emitted from the light source 24 is totally reflected on the non-fingerprint contact portion on the prism surface of the prism 25 and scattered on the fingerprint contact portion, so that the interference corresponding to the fingerprint pattern of the operator occurs. As the received measurement light, C is transmitted through the plane mirrors 26a to 26d and the refraction lens group 26e.
Detected by CD27. The input measurement image photoelectrically converted by the CCD 27 is output to the FM 35 after being adjusted by the adjusting unit 28 regarding the gain and contrast of the two-dimensional intensity distribution.

Then, in step 126, the number of times of picking up the fingerprint of the operator is compared with the preset number of basic registration images. Here, when the number of times the fingerprint is captured is smaller than the number of basic registration images, the process proceeds to step 124. On the other hand, if the number of times the fingerprint has been captured matches the number of basic registration images, the process proceeds to step 128.

Incidentally, in step 124 after the transition again,
By displaying a message requesting the operator to sequentially change the installation state of the finger on the prism 25 on the display 22c, a plurality of input measurement images reflecting different installation states of the finger are sequentially stored in the PM 34 via the FM 35. To do.

Subsequently, in step 128, a combination of two input measurement images is selected from the data stored in the PM 34, and the process proceeds to step 130.

Subsequently, in step 130, the combination of the two measurement images is presented to the frame memory 35 at the same time or in a parallel arrangement with time, and the process proceeds to step 132.

Subsequently, in step 132, the two input measurement images are output from the FM 35 to the LCD 44 and displayed in a parallel arrangement at the same time, and after performing an optical correlation calculation described later, the process proceeds to step 134.

At this time, from the LD 41 to the plane mirror 42a.
The laser light transmitted through the LCD 40 via the half mirror 43a is displayed on the liquid crystal forming the LCD 40.
After receiving the modulation corresponding to the individual input measurement images, Fourier transform is performed by the refracting lens group 45a, and the plane mirror 4
Recorded by SAL 46 via 2b-42d.

Further, from the LD 41 to the plane mirror 42a,
The laser light reflected by the SLM 46 via the half mirrors 43a and 43b receives interference corresponding to the Fourier transform image recorded in the liquid crystal forming the SLM 46, and then is again reflected by the refraction lens group 45b via the half mirror 43b. Undergo a Fourier transform. Then, only one of the first-order lights appearing near the 0th-order light of the correlation image is the plane mirrors 42e to 42e.
It is detected by the CCD 47 via g.

The light intensity of the primary light photoelectrically converted by the CCD 47 is output to the dedicated hardware 36. The dedicated hardware 36 calculates the correlation value of the correlation image corresponding to the combination of the two input measurement images based on the light intensity of the primary light, and stores the averaged correlation value of each input measurement image.

Then, in step 134, it is determined whether or not all the input measurement images stored in the PM 34 are selected as a combination of two input measurement images. Here, if all the input measurement images have not been selected yet, the process proceeds to step 128. On the other hand, if all the input measurement images have already been selected, the process proceeds to step 136.

Subsequently, in step 136, the averaged correlation values of the respective input measurement images are compared, and the process proceeds to step 138. At this time, the input measurement images are sorted according to the order of the correlation values, and the priority order is determined as the basic registration image.

Subsequently, at step 138, each input measurement image stored in the PM 34 is set as a basic registration image in the HD 33.
To be stored. At this time, the basic registration image stored in the HD 33 is divided corresponding to the identification code and holds the array corresponding to the priority order. Here, the process of newly creating the basic registration image ends, and the basic registration process itself ends.

The process of newly creating such a basic registration image is executed as follows, for example. Note that FIG.
As shown in (a), in the HD 33 and the PM 34, the data area assigned to one identification code is 3
It is assumed that the basic registration areas R1 to R3 store individual basic registration images according to the priority order, and the auxiliary registration areas C1 to C3 store three auxiliary registration images according to the priority order.

Here, when the input measurement images A, B and C are sequentially picked up as shown in FIG. 15A, FIG.
As shown in, the correlation calculation is sequentially performed on the selected combination between the input measurement images, and the average value of the autocorrelation value and the cross-correlation value is calculated as the average correlation value of the input measurement images. That is, the averaged correlation values of the input measurement images A, B, and C are 83.0, 84.3, and 81.3, respectively.
Is. After that, each input measurement image is stored in the basic registration area of the HD 33 and the PM 34 according to the priority order determined based on the order of the averaged correlation values. That is, the priorities of the input measurement images A, B, C are second and first, respectively.
The input measurement images A, B, and C are stored in the basic registration areas R2, R1, and R3, respectively.

Then, in step 140, the guide 21
A message requesting the operator to place a finger on the prism 25 corresponding to the shape of c is displayed on the display 22c, and the light source 24 is turned on. Here, when the operator places the finger on the prism 25, the process proceeds to step 142. The input measurement image is output to the FM 35 in the same manner as in step 124 described above.

Subsequently, in step 142, the basic registration image attached to the identification code input by the operator is searched in the data stored in the PM 34 or the HD 33, and then 2
A combination of individual basic registration images or input measurement images is selected, and the process proceeds to step 144.

Subsequently, in step 144, a combination of two basic registration images or input measurement images is stored in the FM 35 at the same time or in a temporally parallel arrangement, and the process proceeds to step 146.

Subsequently, in step 146, two basic registration images or input measurement images are output from the FM 35 to the LCD 44 and displayed in parallel at the same time, and optical correlation calculation is performed in the same manner as in step 132 described above. After that, the process proceeds to step 148.

Then, in step 148, P is set as a combination of two basic registration images or input measurement images.
It is determined whether or not all the basic reference images and input measurement images stored in M34 or HD33 have been selected. Here, if all the basic reference images and input measurement images have not been selected yet, the process proceeds to step 142. On the other hand, if all the basic reference images and input measurement images have already been selected, the process proceeds to step 150.

Subsequently, in step 150, the averaged correlation values of the basic registration image and the input measurement image are compared, and the process proceeds to step 152. At this time, the basic registration image and the input measurement image are sorted according to the order of the correlation values, and the priority order is newly determined.

Subsequently, in step 152, the basic registration image or the input measurement image is set as a new basic registration image in the PM 34.
And the HD 33, and the previous basic registration image is updated.
At this time, only when the number of basic registration images matches the preset number, a new basic registration image is selected except the one having the smallest correlation value. Further, the basic registration image stored in the hard disk 33 is divided corresponding to the identification code and holds the array corresponding to the priority order. Here, the process of updating the basic registration image ends, and the basic registration process itself ends.

The processing for updating the basic registration image is executed as follows, for example. Here, in FIG.
As shown in (d), the basic registration images A, B, and C are the basic registration areas R2, R1, and R of the HD 33 or PM 34, respectively.
18 and the input measurement image J is captured in FIG.
As shown in (b), the correlation calculation is sequentially performed on the selected combination between the basic registration image and the input measurement image, and the autocorrelation value and cross-correlation are obtained as the average correlation value of the basic registration image and the input measurement image. Calculate the average of the values. That is, the averaged correlation values of the basic registration images A, B, C and the input measurement image J are 84.3, 83.8, 82.3, 8 respectively.
It is 6.3.

Thereafter, the basic registration image and the input measurement image excluding the one having the smallest accumulated correlation value are HD-recorded in accordance with the priority order determined based on the rank of the averaged correlation values.
33 and PM 34 in the basic registration area. That is, the basic registration images A, B, C and the input measurement image J have the second priority, the third position, the fourth position, and the first priority, respectively. J is stored in each of the basic registration areas R2, R3 and R1.

As shown in FIGS. 11 to 14, in step 170, the collation determination process is executed in steps 180 to 228.

First, in step 180, a message requesting the operator to enter the identification code is displayed on the display 22c. Here, the operator uses the keyboard 2
If the identification code is entered using 1d, step 18
Move to 2.

Subsequently, in step 182, the basic registration image attached to the identification code input by the operator is searched in the data stored in the PM 34 or the HD 33. Here, if the basic registration image is not stored, the matching determination process itself ends. On the other hand, if the basic registration image is stored, the process proceeds to step 184, and the process of collating the input measurement image with the reference image in which the basic registration image is set is executed.

Subsequently, in step 184, the guide 21
A message requesting the operator to place a finger on the prism 25 corresponding to the shape of c is displayed on the display 22c, and the light source 24 is turned on. Here, when the operator places the finger on the prism 25, the process proceeds to step 186. Note that the input measurement image is stored in the frame memory 35 of the control means 30 in the same manner as in steps 124 and 140 described above.
Is output to

Subsequently, in step 186, the basic registration image attached to the identification code input by the operator is searched in the data stored in the PM 34 or the HD 33, and then the basic registration image selected according to the priority order is referred to. The image is set as an image, and the process proceeds to step 188. Here, by referring to the basic registration image according to the priority order thereof, the number of collations with respect to the input measurement image is suppressed to the minimum.

Subsequently, in step 188, the combination of the input measurement image and the reference image is presented to the FM 35 at the same time or in a temporally parallel arrangement, and the process proceeds to step 190.

Subsequently, in step 190, the input measurement image and the reference image are output from the FM 35 to the LCD 44 and displayed at the same time in a side-by-side arrangement.
After performing the optical correlation calculation in the same manner as 46, the process proceeds to step 192.

Then, in step 192, the averaged correlation value of the input measurement image is compared with a preset threshold value. Here, when the correlation value of the input measurement image is larger than the threshold value, the process proceeds to step 194, and the process of newly creating the auxiliary registration image is executed. On the other hand, when the correlation value of the input measurement image is less than or equal to the threshold value, the process proceeds to step 200.

Subsequently, in step 194, it is determined that the operator matches the fingerprint with the person who has previously registered the fingerprint, based on the success of the matching between the input measurement image and one of the basic registration images, and the process proceeds to step 196. To do. At this time, the dedicated hardware 36 outputs a coincidence signal indicating success of fingerprint collation to an external device (not shown).

Subsequently, in step 196, the auxiliary registration image attached to the identification code input by the operator is searched in the data stored in the PM 34 or the HD 33. Here, when the existing number of the auxiliary registration images is smaller than the preset number, the process proceeds to step 198. on the other hand,
When the existing number of the auxiliary registration images matches the preset number, the process of collating the input measurement image with the reference image having the basic registration image ends, and the collation determination process itself is performed. finish.

Subsequently, in step 198, the input measurement image stored in the PM 34 is stored in the HD 33 as an auxiliary registration image. At this time, the auxiliary registration image stored in the HD 33 is divided corresponding to the identification code. Here, the process of newly creating the auxiliary registration image ends, and the collation determination process itself ends.

The process of newly creating such an auxiliary registration image is executed as follows, for example. Here, FIG.
As shown in FIG. 5B, the basic registration images A, B, and C are the basic registration areas R2, R1, and R of the HD 33 and the PM 34, respectively.
3 and the input measurement image D is successfully collated with the basic registration image A, the input measurement image D is stored in the HD 33 and the PM 3
4 in the auxiliary registration area C1.

Then, in step 200, it is determined whether or not all the basic reference images stored in the PM 34 or the HD 33 are selected as the reference images. Here, if all basic reference images have not been selected, step 1
Move to 86. On the other hand, when all the basic reference images have already been selected, the process proceeds to step 202, and the process of collating the input measurement image with the reference image for which the auxiliary registration image is set is executed.

Subsequently, in step 202, the auxiliary registration image attached to the identification code input by the operator is searched in the data stored in the PM 34 or the HD 33. Here, if the auxiliary registration image is not stored, the matching determination process itself ends. On the other hand, if the basic registration image is stored, the process proceeds to step 204, and the process of matching the input measurement image with the reference image for which the auxiliary registration image is set is executed.

Subsequently, in step 204, after the auxiliary registration image attached to the identification code input by the operator is searched in the data stored in the PM 34 or the HD 33, the auxiliary registration image selected according to the priority order is referred to. The image is set as an image, and the process proceeds to step 206.

Subsequently, in step 206, the combination of the input measurement image and the reference image is stored in the FM 35 at the same time or in a temporally parallel arrangement, and the process proceeds to step 208.

Subsequently, in step 208, the input measurement image and the reference image are output from the FM 35 to the LCD 44 and displayed in a parallel arrangement at the same time, and the above-mentioned steps 132, 1 are performed.
After performing the optical correlation calculation in the same manner as 46 and 190, the process proceeds to step 210.

Subsequently, in step 210, the averaged correlation value of the input measurement image is compared with a preset threshold value. Here, when the correlation value of the input measurement image is larger than the threshold value, the process proceeds to step 216, and the process of updating the auxiliary registration image is executed. On the other hand, if the correlation value of the input measurement image is less than or equal to the threshold value, the process proceeds to step 232.

Subsequently, in step 212, it is determined that the operator matches the fingerprint with the person who has previously registered the fingerprint, based on the success of the matching between the input measurement image and one of the auxiliary registration images, and the process proceeds to step 214. To do. At this time, the dedicated hardware 36 outputs a coincidence signal indicating success of fingerprint collation to an external device (not shown).

Subsequently, in step 214, the basic registration image and the auxiliary registration image attached to the identification code input by the operator are searched in the data stored in the PM 34 or the HD 33, and then the basic registration image or the auxiliary registration image is searched. 1 and one of the auxiliary registration image and one of the input measurement images are selected, and the process proceeds to step 216.

Subsequently, in step 216, a combination of one of the basic registration image or auxiliary registration image and one of the auxiliary registration image or one of the input measurement images is stored in the FM 35 at the same time or in a parallel arrangement with time, and in step 218. Move to.

Subsequently, in step 218, one of the basic registration image or the auxiliary registration image and one of the auxiliary registration image or the input measurement image is output from the FM 35 to the LCD 44 and displayed in a parallel arrangement at the same time. , 14
After performing the optical correlation calculation in the same manner as 6, 190, 208, the process proceeds to step 220.

Subsequently, in step 220, PM34 or H is set as a combination of one of the basic registration image or the auxiliary registration image and one of the auxiliary registration image or the input measurement image.
It is determined whether or not all the basic registration images, auxiliary registration images and input measurement images stored in D33 are selected. Here, if all the basic registration images, auxiliary registration images and input measurement images have not been selected yet, the process proceeds to step 214. On the other hand, if all the basic registration images, auxiliary registration images, and input measurement images have already been selected, the process proceeds to step 222.

Then, in step 222, the averaged correlation values of the auxiliary registration image and the input measurement image are compared, and the process proceeds to step 224. At this time, the auxiliary registration image and the input measurement image are sorted according to the order of the correlation values, and the priority order is newly determined.

Subsequently, in step 224, the auxiliary registration image or the input measurement image is set as a new auxiliary registration image in the PM 34.
And the HD 33 to update the previous auxiliary registration image.
At this time, only when the existing number of auxiliary registration images matches the preset number, new auxiliary registration images are selected except those having the smallest correlation value. Also, HD3
The auxiliary registration image stored in No. 3 is divided according to the identification code and holds the array corresponding to the priority order. At this point, when the process of updating the auxiliary registration image ends,
The collation determination process itself ends.

The processing for updating the auxiliary registration image as described above is executed as follows, for example. Here, FIG.
As shown in (c), the basic registration images A, B, C and the auxiliary registration image D are stored in the basic registration areas R2, R1, R3 and the auxiliary registration area C1 of the HD 33 and PM 34, respectively, and input to the auxiliary registration image D. When the matching of the measurement image E is successful, as shown in FIG. 17B, the correlation calculation is performed on the selected combination of one of the basic registration image and the auxiliary registration image and one of the auxiliary registration image and the input measurement image. The sequential execution is performed to calculate the average value of the autocorrelation value and the cross-correlation value as the averaged correlation value of each of the auxiliary registration image and the input measurement image. That is, the average correlation values of the auxiliary registration image D and the input measurement image E are 60.8 and 73.0, respectively.

After that, the auxiliary registration image and the input measurement image are stored in the auxiliary registration area of the HD 33 or the PM 34 according to the priority order determined based on the order of the averaged correlation values. That is, since the priority order of the auxiliary registration image D and the input measurement image E is second and first, respectively, the auxiliary registration image D
And the input measurement image E are stored in the auxiliary registration areas C2 and C1, respectively.

Here, as shown in FIG. 15D, the basic registration images A, B, C and the auxiliary registration images D, E are the basic registration areas R2, R1, respectively of the HD 33 and the PM 34.
When all of the input measurement images F stored in the R3 and the auxiliary registration areas C2 and C1 are unsuccessful in collation, the basic registration area and the auxiliary registration areas of the HD 33 and the PM 34 are stored in both the basic registration area and the auxiliary registration area. The measurement image F is not stored.

As shown in FIG. 16A, the basic registration images A, B and C and the auxiliary registration images D and E are HD3, respectively.
3 and the PM 34 are stored in the basic registration areas R2, R1, R3 and the auxiliary registration areas C2, C1. When the input measurement image G is successfully collated with the basic registration image C, as shown in FIG. An autocorrelation value is calculated as an averaged correlation value of each of the auxiliary registration image and the input measurement image by sequentially performing the correlation calculation on the combination of one of the registration image and the auxiliary registration image and one of the auxiliary registration image and the input measurement image selected. And calculate the average of the cross-correlation values. That is, the auxiliary registration images E and D
The average correlation values of the input measurement image G are 68.0, 60.3, and 53.0, respectively.

After that, the auxiliary registration image and the input measurement image are stored in the auxiliary registration area of the HD 33 or PM 34 in accordance with the priority order determined based on the order of the averaged correlation values. That is, since the priority orders of the auxiliary registration images E and D and the input measurement image G are the first, second, and third, respectively,
The auxiliary registration images E, D and the input measurement image G are stored in the auxiliary registration areas C1, C2, C3, respectively.

Here, as shown in FIG. 16B, the basic registration images A, B, C and the auxiliary registration images D, E, G are respectively the basic registration area R2 of the HD 33 or PM 34.
When the input measurement image H stored in R1, R3 and the auxiliary registration areas C2, C1, C3 is successfully collated with the basic registration image B, the input measurement image is displayed in both the basic registration area and the auxiliary registration area of the HD 33 and PM 34. Do not store H.

Further, as shown in FIG. 16C, the basic registration images A, B, C and the auxiliary registration images D, E, G are respectively the basic registration areas R2, R1, R of the HD 33 or PM 34.
3 and the auxiliary registration areas C2, C1 and C3, and the collation of the input measurement image I with the auxiliary registration image D is successful,
As shown in FIG. 18A, the correlation calculation is sequentially performed on the selected combination of the auxiliary registration image and the input measurement image,
The average value of the autocorrelation value and the cross-correlation value is calculated as the averaged correlation value of each of the auxiliary registration image and the input measurement image. That is, the averaged correlation values of the auxiliary registration images E, D, G and the input measurement image I are 56.7, 67.0, 6 respectively.
It is 0.0 and 66.5.

Thereafter, the auxiliary registration image and the measurement image excluding those having the smallest averaged correlation value are HD33 according to the priority determined based on the averaged correlation value order.
Alternatively, it is stored in the auxiliary registration area of the PM 34. That is,
Since the auxiliary registration images E, D, G and the input measurement image I have the fourth priority, the first position, the third position, and the second priority, respectively, the auxiliary registration images E, D, G and the input measurement image I. Are stored in the auxiliary registration areas C4, C1, C3, C2, respectively.

Then, in step 226, it is determined whether or not all the auxiliary registration images stored in the PM 34 or the HD 33 are selected as the reference images. Here, if all the auxiliary registration images have not been selected yet, step 2
Move to 04. On the other hand, if all auxiliary registration images have already been selected, the process proceeds to step 232.

Subsequently, in step 228, it is determined that the operator does not match the fingerprint with the individual who has previously registered the fingerprint, based on the unsuccessful matching of the input measurement image with all of the basic registration image and the auxiliary registration image. The process of collating the input measurement image with the reference image for which the auxiliary registration image is set ends, and the collation determination process itself ends. At this time, the dedicated hardware 36 outputs an inconsistency signal indicating an unsuccessful fingerprint collation to an external device (not shown).

When such a person matching device 10 is used for managing entry / exit to / from a secret room, for example, the dedicated hardware 36 to an external device (not shown) is used.
The auto-lock of the door leading to the secret room is unlocked based on the coincidence signal output to. On the other hand, based on the coincidence signal output from the dedicated hardware 36 to the external device (not shown), the automatic lock of the door leading to the confidential room is maintained in the locked state.

In the present embodiment, as a technique for performing correlation calculation based on Fourier transform by optical means as a pattern matching method, that is, so-called optical correlation calculation, reference "Opt. Eng., Vol. 30, no. 12, 1991 "and so on. According to this optical correlation calculation, since the speed of the correlation calculation does not depend on the gradation of the pattern image, the pattern image can be treated as multivalued information. On the other hand, according to the electronic correlation calculation using an electronic computer such as a computer, the speed of the correlation calculation is restricted by the multivalued information. Therefore, the system that performs the optical correlation calculation is more effective than the system that performs the electronic correlation calculation even with respect to patterns including cuts and distortions as a fingerprint image.

Here, the present invention is not limited to the above embodiments, but various modifications can be made.

For example, in the above embodiment, the identification of the individual is performed by designating the fingerprint as a part of the pattern forming the human body. However, as a part of the pattern forming the human body, it is also preferable to specify another pattern such as a retinal image as long as it has personal characteristics.

Further, in the above embodiment, a so-called pattern matching method, which is a method of performing a correlation operation based on Fourier transform, is applied as a pattern matching method.
However, it is also preferable to apply a so-called feature extraction method, which is a method of extracting minutiae having large individual differences, as a pattern matching method. The fingerprint collation technology related to such a feature extraction method is described in detail in the document "PRU88-83, pp.65-72, 1988" and the like.

Further, in the above embodiment, the pattern matching method is applied to one newly input fingerprint image. However, it is also preferable to apply a method of matching a plurality of newly input fingerprint images, a so-called optical multiple matched filter method. Regarding the fingerprint collation technique relating to such an optical multiple matched filter method, Japanese Patent Application Laid-Open No. 4-2254
No. 06 ”and the like.

[0121]

As described above in detail, in the person collation method and apparatus according to the present invention, a part of the pattern forming the human body is imaged in advance, and images of a plurality of patterns are respectively registered as the basic registration image and the image. The auxiliary registration image is divided and stored. Here, the input measurement image for newly capturing and holding the pattern is first judged to be matched with the basic registration image. At this time, if it is determined that the input measurement image does not match the basic registration image, the input measurement image is subsequently determined to be collated with the auxiliary registration image. At this time, when the coincidence of the input measurement image with the auxiliary registration image is determined, based on the matching in all the combinations selected from the auxiliary registration image and the input measurement image, the ones having the smallest correlation value are excluded. The registration image and the input measurement image are updated and stored as auxiliary registration images.

That is, in the process of collating the input measurement images, the basic registration image is stored as a reference image group that cannot be rewritten, that is, cannot be updated. On the other hand, in the process of collating the input measurement image, the auxiliary registration image is stored as a reference group that can be rewritten at any time, that is, can be updated. As a result, the auxiliary registration image is updated to include the input measurement image that is captured and held at any time, and thus when capturing a part of the pattern that constitutes the human body, the personal registration that appears in the installation state of that part of the human body is updated. It converges so that it has a tendency that reflects its unique habits. On the other hand, since the basic registration image is not changed by the input measurement image newly captured and held, various tendencies that can be considered as the installation state of a part of the human body when capturing a part of the pattern forming the human body. To hold.

Therefore, the reference group consisting of the basic registration image and the auxiliary registration image is the installation state that generally occurs as the installation state of a part of the human body when capturing a part of the pattern that constitutes the human body. It corresponds to both the installation state that occurs personally. That is, when the individual who provided the basic registration image and the auxiliary registration image and the individual who provided the input measurement image are the same person, the input measurement image is higher than at least one of the basic registration image and the auxiliary registration image. It has reproducibility. Therefore, the matching accuracy with respect to a part of the pattern forming the human body is improved, so that the individual can be identified more reliably than before.

[Brief description of drawings]

FIG. 1 is a block diagram showing the overall configuration of an embodiment of a person matching device of the present invention.

2A is a top view showing a detailed configuration of the input means of FIG. 1, and FIG. 2B is a side view showing a detailed configuration of the input means of FIG.

FIG. 3 is an optical path diagram showing a detailed configuration of a matching unit in FIG.

4A is a diagram showing two input measurement images captured by the input means of FIG. 1, and FIG. 4B is a diagram showing Fourier transform images for the two input measurement images of FIG. Yes,
(C) is a figure which shows the correlation image with respect to the Fourier-transformed image of (b).

5A is a diagram showing a method of storing a fingerprint measurement image in the frame memory of FIG. 1, and FIG. 5B is a diagram showing a method of displaying a fingerprint measurement image on the liquid crystal display of FIG.

6A is a diagram showing a method of storing a fingerprint measurement image in the frame memory of FIG. 1, and FIG. 6B is a diagram showing a method of displaying a fingerprint measurement image on the liquid crystal display of FIG.

7 (a) is a diagram showing a method of storing a fingerprint measurement image in the frame memory of FIG. 1, and FIG. 7 (b) is a diagram showing a method of displaying a fingerprint measurement image on the liquid crystal display of FIG.

8 is a flowchart showing the overall operation of the person verification apparatus of FIG.

9 is a flowchart showing a process of newly creating a basic registration image in the person matching device of FIG.

10 is a flowchart showing a process of updating a basic registration image in the person matching device of FIG.

11 is a flowchart showing a process of collating a measurement image of a fingerprint based on a reference of a basic registration image in the person collation device of FIG.

FIG. 12 is a flowchart showing a process of newly creating an auxiliary registration image in the person verification device of FIG.

13 is a flowchart showing a process of collating a measurement image of a fingerprint on the basis of a reference of an auxiliary registration image in the person collation device of FIG.

FIG. 14 is a flowchart showing a process of updating an auxiliary registration image in the person verification device of FIG.

15A to 15D are diagrams showing a storage state of a basic registration image and an auxiliary registration image in the hard disk of FIG.

16A to 16D are diagrams showing a storage state of a basic registration image and an auxiliary registration image in the hard disk of FIG.

17A to 17C are diagrams showing correlation values calculated among a plurality of measurement images in the dedicated hardware of FIG. 1.

18A and 18B are diagrams showing correlation values calculated among a plurality of measurement images in the dedicated hardware of FIG. 1.

[Explanation of symbols]

10 ... Person collation device, 20 ... Input means, 30 ... Control means, 33 ... First storage unit, 34 ... Second storage unit, 36 ...
Collation determination unit, 40 ... Collation means.

 ─────────────────────────────────────────────────── ─── Continuation of the front page (72) Inventor Shigehiro Yoshida 1 1126 Ichinomachi, Hamamatsu City, Shizuoka Prefecture 1126 Hamamatsu Photonics Co., Ltd. (72) Inventor Tsutomu Hara 1 1126 1 Nomachi, Hamamatsu City, Shizuoka Prefecture Hamamatsu Photonics Within the corporation

Claims (8)

[Claims] 1. A person collation method for identifying an individual based on collation of some patterns constituting a human body, wherein the pattern is imaged a plurality of times in advance, and images of the plurality of patterns are respectively a basic registration image and auxiliary registration. A first step of dividing and storing as one of the images; a second step of newly capturing the pattern and holding the image of the pattern as an input measurement image; the basic registration image and the input measurement A third step of determining whether or not the input measurement image matches the basic registration image based on a correlation value calculated with the image, and the input measurement image for the basic registration image in the third step. Is determined, the matching of the input measurement image with the auxiliary registration image is determined based on the correlation value calculated between the auxiliary registration image and the input measurement image. Is a fourth step of determining non-coincidence, and in the case where the coincidence of the input measurement image with the auxiliary registration image is determined in this fourth step, all combinations selected from the auxiliary registration image and the input measurement image are selected. A fifth step of updating and storing, as the auxiliary registration image, the auxiliary registration image and the input measurement image excluding those having the smallest correlation value based on the calculated correlation value. How to match people. 2. In the first step, after the pattern is captured in advance and the image of the pattern is stored as a basic registration image, the pattern is newly captured and the image of the pattern is held as an input measurement image. Then, based on the correlation value calculated between the basic registration image and the input measurement image, when the match of the input measurement image to the basic registration image is determined,
Storing the input measurement image as the auxiliary registration image separately from the auxiliary registration image until the existing number of the auxiliary registration images matches a preset number. The person matching method according to claim 1. 3. The person collation method according to claim 1, wherein the auxiliary registration image is an image of the plurality of patterns captured individually. 4. The person matching method according to claim 1, wherein the pattern is a fingerprint formed on a surface of a human finger. 5. A person collation apparatus for identifying an individual based on collation with respect to a part of a pattern forming a human body, wherein an input unit for capturing the pattern, and an image of the plurality of patterns captured in advance by the input unit. A control including a first storage unit that stores a basic registration image and an auxiliary registration image, respectively, and a second storage unit that holds an image of the pattern newly captured by the input unit as an input measurement image. Means for performing a correlation operation on the basic registration image or the auxiliary registration image and the input measurement image input from the control means, and between the basic registration image or the auxiliary registration image and the input measurement image. Collating means for calculating a correlation value of, the control means, between the basic registration image or the auxiliary registration image and the input measurement image input from the collating means. Based on the correlation value issued, the input measurement image for the basic registration image is further configured to further include a matching determination unit for determining whether the input measurement image matches or does not match the basic registration image or the auxiliary registration image. After determining the disagreement, if the match of the input measurement image to the auxiliary registration image is determined, the correlation calculated in all combinations selected from the auxiliary registration image and the input measurement image input from the collating means. A person collation apparatus for updating and storing the auxiliary registration image and the input measurement image excluding those having the smallest correlation value as the auxiliary registration image based on the value. 6. The person collation apparatus according to claim 5, wherein the auxiliary registration image is an image of the plurality of patterns captured individually. 7. The person verification apparatus according to claim 5, wherein the pattern is a fingerprint formed on the surface of a human finger. 8. The matching means includes an optical lens that performs two Fourier transforms on the basic registration image or the auxiliary registration image displayed in parallel arrangement and the input measurement image. The person matching device according to claim 5, wherein