JPH06176134A - Fingerprint recognition device - Google Patents

Abstract

PURPOSE:To remove noise on account of the distortion of the image of a fingerprint and to improve a recognition rate by moving or rotating a transparent body perpendicularly to the pressure-installation direction of the transparent body and to a two-dimensional direction. CONSTITUTION:A fingerprint pattern is obtained from the difference of a reflection condition when light is projected from a light source (LED) 330 to a prism 320 and a finger 301 is pressed. A lens 312 imageforms the fingerprint pattern on a CCD camera 310, and the CCD camera 310 converts the pattern into a video signal and transmits it to a recognition device 10. It frequently occurs that distortion is generated in the fingerprint pattern by the inputting method of force into the finger 301 and an attitude, and therefore erroneous recognition is liable to occur. Thus, erroneous recognition is prevented by driving stages 422, 424 and 426 and giving vibration. The recognition device 101 collates all the generated fingerprint patterns with a previously registered pattern so as to recognize them on a real time basis. When the coincident pattern exists, it is recognized to be the same or the previously registered one.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an optical information processing technique using spatial Fourier transform in pattern recognition and the like.

[0002]

2. Description of the Related Art Since pattern recognition can be used to identify an individual, a device for identifying an individual by using a fingerprint pattern as personal information, for example, has been proposed (Japanese Patent Laid-Open No. 3-3003). 204625).

Various methods for performing pattern correlation at high speed by optical calculation have been proposed, and one of them has been called the joint transform correlation (JTC) method ("Fourier in optical information processing"). conversion",
Optics 21,6,392-399). This method is more accurate than the method using an optical correlator of Vander-Lugt (“Pattern recognition by an optical correlation system phase filter”, Applied Physics 58,6 (1989)), which is more accurate than a hologram dry plate on which a pattern is written. The advantage is that no alignment is required.

FIG. 6 shows an example of the joint transform correlation method. It is configured by using a high resolution spatial light modulator (SLM). The input pattern and the reference pattern are simultaneously written in the SLM, these patterns are read again by the SLM, and the optical Fourier transform is performed to obtain the correlation signals of these patterns.

FIG. 7 shows a fingerprint pattern recognizing device configured by the device of FIG. 6, and an attempt is made to simultaneously draw an input pattern (input image) and a reference pattern (reference image) on a CRT to detect a match. To do. The input image (left) and the reference image (right) as shown in FIG.
The Fourier pattern shown in (b) is obtained. Since the input image and the reference image match, two high-luminance points appear on the left and right corresponding to the positions of the input image and the reference image, as shown in FIG. As a result, a match between the input pattern and the reference pattern is detected.

[0006]

When a fingerprint pattern is taken, the image is often distorted depending on how the finger is pressed against the prism or how it is placed. When the input pattern and the reference pattern are recognized by correlation as in the apparatus shown in FIG. 7, such image distortion causes noise and is not regarded as a match, which causes erroneous recognition. Due to such distortion, in an actual fingerprint recognition device, even a person who is registered in advance is determined not to be in agreement, so it is determined that the person is not the person and is excluded.

In order to suppress this distortion, a finger guide system, a feature extraction system, etc. have been considered. However, even when the guide system is used, distortion due to friction between the finger and the prism occurs, resulting in erroneous recognition.
In addition, it is known that minutia features can achieve a high recognition rate as feature extraction, but there is a drawback that a large-scale computer is required for that (“fingerprint automatic identification technology”, measurement And Control 25,8,701 (1986)
JP-A-59-32064). Further, there is a computer that performs image processing to correct the distortion, which also has the same drawback (Japanese Patent Laid-Open No. 2-270076).

In view of the above problems, it is an object of the present invention to provide a high-speed and inexpensive fingerprint recognition device having a high recognition rate by adding a mechanism for removing noise due to distortion of a fingerprint image.

[0009]

In order to solve the above problems, the fingerprint recognition device of the present invention registers in advance a fingerprint image obtained by reflecting light from a light source on a transparent body on which a finger is pressed. A fingerprint recognition device for performing a joint Fourier transform with a pattern to be referenced and recognizing a fingerprint image from the correlation value thereof, wherein the transparent body is moved in a two-dimensional direction perpendicular to the pressing direction of the transparent body or First means for driving to rotate (for example, either or both of the XY stage and the rotating stage) and control of driving of the first means when recognizing a fingerprint image (for example, a predetermined length). Second reciprocating motion, vibration, or random motion).

The second means may be characterized in that the drive of the first means is controlled so that the drive of the transparent body becomes smaller as the correlation value increases.

The fingerprint image may be recognized by performing a joint Fourier transform of the image obtained by picking up the fingerprint image and the pattern to be referenced by the spatial light modulator. Further, the feature may be that the matching between the fingerprint image and the pattern to be referred to is detected within a predetermined time.

[0012]

In the fingerprint recognition device of the present invention, when recognizing a fingerprint image, the transparent body moves or rotates in a two-dimensional direction perpendicular to the pressing direction, and the fingerprint image moves the transparent body. Or the shape changes by rotation. If the person who owns the fingerprint image has a pre-registered fingerprint, even if the fingerprint image is initially distorted, the change causes a fingerprint image that matches the pattern to be referred (the distortion is corrected). Become. Matching of fingerprint images is detected in real time from the correlation value of the joint Fourier transform, and it is detected that the owner of the fingerprint image is a person who has a fingerprint registered in advance. on the other hand,
This drive is performed by the first means according to the control of the second means, and if the person who does not have the fingerprint registered in advance does not match the fingerprint images no matter how the person drives, recognition is performed for a certain period of time. If the correlation value becomes large, the driving of the transparent body is made small so that the recognition can be performed with higher accuracy.

[0013]

Embodiments of the present invention will be described with reference to the drawings.
FIG. 1 is a conceptual diagram of an embodiment of a fingerprint recognition device of the present invention, in which an X linear stage 422 and a Y linear stage 424 are used as a mechanism for vibrating a prism 320 as a transparent body, and a rotary stage 426 is used as a mechanism for rotating. It was realized. These mechanisms are drive-controlled by the controller unit 410. In addition, the total reflection input method is used to obtain a fingerprint pattern (fingerprint image), and light is emitted from a light source (LED) 330 to a prism 320 and a fingerprint pattern is obtained from the difference in reflection condition when a finger 301 is pressed. ing. The lens 312 forms an image of this fingerprint pattern on the CCD camera 310, and CC
The D camera 310 converts it into a video signal and sends it to the recognition device 101.

At this time, there are many cases in which the fingerprint pattern is distorted due to the way of applying force to the finger 301 and the posture thereof, which easily causes erroneous recognition.
Misrecognition is prevented by driving 424 and 426 to give vibrations. The fingerprint pattern is further distorted by applying vibration or the like to equivalently generate a plurality of fingerprint patterns. The recognition device 101 performs real-time recognition by collating all the generated fingerprint patterns with a pre-registered pattern (hereinafter referred to as a reference image), and if there is a match, the same person or a pre-registered one. Certify as.

FIG. 2 shows the configuration more specifically.

The prism 320, the LED 330, the lens 312,
The CCD camera 310 constitutes a fingerprint input section and obtains a fingerprint pattern as described above. Prism 320
Are attached to the stages 422, 424, and 426 together with the LED 330 so as to move in the x, y and rotation directions. That is, two-dimensional movement is possible on a plane perpendicular to the direction in which the finger 301 is pushed. The recognition unit 101
The fingerprint pattern is used as an input image, and recognition is performed by optical correlation calculation with a reference image. The CRT 150, lenses 121, 122, 123,
124, SLMs 110 and 112, a filter 114, and the like.

The CRT 150 displays the input image and the reference image on the same screen, and the lens 121 displays the input image and the reference image on the SLM 11.
0 is an image to be formed. Lens 123 is SLM110
The input image and the reference image read from are subjected to Fourier transform to form a Fourier pattern on the SLM 112. Lens 124
Re-transforms the Fourier pattern formed on the SLM 112, and forms a pattern corresponding to the correlation value on the detector (photodiode, PD) 140. Detector 140
Is placed at a point where the brightness becomes high when they match, and this is detected and output as a correlation signal.

SLM 110 placed on the focal plane of lens 121
Is for reading out an image written by writing light by reading light and reproducing the image (Opttronics (1985) No. 4). The written input image and reference image are reproduced by laser light. This includes so-called FLC-SL using amorphous silicon and liquid crystal.
M (ferroelectric liquid crystal spatial light modulator), LCLV (liquid crystal light valve), TN-SLM (twisted nematic liquid crystal-SLM), or a combination of a photocathode and an optical crystal (MSLM) can be used.

The polarization beam splitters 130 and 132, the half mirror 133 and the mirror 134 form an optical system for reading light to the SLMs 110 and 112. The detector 140 is
If they are placed on the focal plane of the lens 124 and they match each other, FIG.
It is placed at a position corresponding to the high brightness point in (c).

The input image (input pattern) and the reference image (reference pattern) are formed on the SLM 110 by the lens 121. This image is read out from the SLM110 by the laser beam and Fourier-transformed to obtain a Fourier pattern SLM112.
Written in. At this time, the S / N may be increased by blocking the 0-shaped light that becomes noise by the filter 114. And the Fourier pattern is SLM11
It is read from 2 and Fourier transformed. Then, a correlation signal is output from the pattern according to the correlation value.

The controller unit 410 drives and controls the stages 422, 424 and 426 according to the pattern matching, that is, the correlation signal, and also functions as a man-machine interface, and also functions as a controller 412, a recognition result display unit 414, and a reference image holding unit. 416 and a personal identification number input device 418. The personal identification number input device 418 is a part for inputting a personal identification number,
The controller 412 reads the reference image from the reference image holding unit 416 holding the reference image corresponding to the personal identification number and outputs it to the CRT 150. The controller 412 detects the degree of coincidence from the strength of the correlation signal, and reduces the drive range of the stages 422, 424, 426 if the correlation signal is large. Further, whether or not they match is output to the recognition result display unit 414, and the recognition result display unit 414 displays this. If they do not match for a certain period of time, the fact that they are not registered is displayed. When the operation starts, the power is turned on, and when the recognition result is obtained, the power is turned off to save the power.

Next, the operation of this apparatus will be described step by step.

First, when a personal identification number is input from the personal identification number input device 418, the reference pattern corresponding to the personal identification number out of the registered reference patterns (fingerprint patterns) is recognized by the recognition unit 10.
It is sent as a reference image to 1. LED330 and CCD at the same time
The power supplies of the cameras 310, SLMs 110 and 112, and these laser light sources are turned on. Finger placed on prism 320
The fingerprint pattern of 301 is recognized by the CCD camera through the recognition unit 101.
Sent as an input image.

In the recognition unit 101, the fingerprint pattern and the reference pattern are displayed on the screen of the CRT 150 with a distance d, and are scratched by the SLM 110 (FIG. 3A, input image left, reference image right). The image on the SLM110 is read by the laser beam, and the pattern obtained by performing the spatial Fourier transform is S
It is written in the LM112 (FIG. 3 (b)). This Fourier pattern is read out from the SLM 112 by laser light, and the correlation value, that is, the result of the joint Fourier transform appears as a bright spot on the detector 140 (FIG. 3C). The central portion of the figure is due to the 0th-order light, and the bright spot detected as a correlation signal is in the target inland. This is detected by the detector 140 and output to the controller 412 as a correlation signal.

If the correlation signal is larger than a predetermined threshold value, it is determined that the fingerprint pattern and the reference pattern match, and the person is identified as the person. And
The power sources of the LED 330, the CCD camera 310, the SLMs 110 and 112 and the power sources of these laser light sources are turned off.

If the correlation signal is less than the threshold,
The stage 422, 424, 426 is driven and moved to give a proper distortion to the fingerprint pattern and input as a new input image,
Recognize again. Then, this is repeated until a correlation signal equal to or more than a threshold value is obtained, and if no match is detected even after a lapse of a certain period of time, it is displayed on the recognition result display unit 414 that it is excluded as an unregistered one.

The distortion of the fingerprint pattern is roughly classified into image distortion due to the difference in how the finger is pressed, rotation of the image depending on how the finger is placed, and parallel movement of the image depending on the way the finger is placed.
With respect to these distortions, an image in which the distortions are canceled is obtained while moving the stages 422, 424, and 426, and the coincidence is detected by this image.

FIG. 4 shows a simple recognition result (relationship between trial Event and correlation value Correration). 3 is a diagram showing a change in intensity of a correlation signal when the prism is arbitrarily (randomly) moved in one direction in the apparatus of FIG. The input image and the reference image use the same fingerprint. Although the input image is distorted, moving the prism changes the correlation value, that is, the strength of the correlation signal, and the strength of the correlation signal reaches a peak when the distortion is canceled. This result shows that by moving the prism arbitrarily, a match can be obtained.

In particular, in this embodiment, a recognition unit 101 is used as a device capable of high-speed correlation calculation for recognition. The optical operation of this device is very fast, so the CRT
150 1 frame or 1 field time (1/3
Recognition is completed in less than 0, 1/60 second), and the fingerprint pattern is distorted. It is possible to recognize by changing the fingerprint pattern 30 or 60 times or more per second. for that reason,
It can be fast and have a high recognition rate.

A stage as a mechanism for moving the prism
422, 424, and 426 are used to make the x-direction, y-direction, and rotation the simplest, and when the correlation signal increases, the vibration speed distance is reduced to reduce the change in the strain to be added next. This enables more reliable recognition.

The present invention is not limited to the above-described embodiment, but various modifications can be made.

Although the mechanism for moving the prism is the simplest, a pressure sensor is arranged below the prism to detect the direction in which the finger is pressed, predict the direction of strain, and move in that direction on the stage. If you do, it will be a more reliable recognition. As the stage, for example, a stage having a moving speed of 100 mm / s and a moving amount of 20 mm is commercially available. In addition to this, a piezo actuator can be used, and according to this, it is possible to move several tens of μm at several tens of kHz.

The fingerprint input may be holographic as shown in FIG. 5, instead of prism. This sensor is a transparent glass plate (light guide plate)
A hologram 324 is formed on a part of 322, and laser light is emitted from the opposite side of the other end.
At 0, a fingerprint pattern is detected.

[0034]

As described above, according to the fingerprint recognition device of the present invention, the shape of the fingerprint image changes due to the movement or rotation of the transparent body when the transparent body is driven. Since a corrected product can be produced, the structure is simple, and it can be made resistant to distortion of the fingerprint image at the time of pressing, which can contribute to the improvement of the recognition rate.

[Brief description of drawings]

FIG. 1 is a conceptual diagram of an embodiment of the present invention.

FIG. 2 is a configuration diagram of an embodiment of the present invention.

FIG. 3 is a diagram showing how a Fourier transform is performed.

FIG. 4 is a diagram showing a relationship between a trial event and a correlation value Correration.

FIG. 5 is a diagram showing another modification of fingerprint input.

FIG. 6 is a diagram showing an example of a spatial Fourier transform device based on a joint transform correlation method.

7 is a diagram showing an example in which a fingerprint pattern recognition device is configured by the device of FIG.

[Explanation of symbols]

101 ... Recognition device, 110, 112 ... SLM, 121,
122, 123, 124, 312 ... Lens, 140 ... Photo diode, 310 ... CCD camera, 320 ... Prism, 330 ... Light source, 410 ... Controller, 422 ...
X linear stage, 424 ... Y linear stage 424, 4
26 ... Rotary stage.

Claims (5)

[Claims] 1. A fingerprint image obtained by reflecting light from a light source on a transparent body on which a finger is pressed is subjected to a joint Fourier transform with a pattern to be referenced in advance, and the correlation value of the fingerprint image of the fingerprint image is obtained. A fingerprint recognition device for recognizing, comprising: first means for driving to move or rotate the transparent body in a two-dimensional direction perpendicular to the pressing direction of the transparent body; A fingerprint recognition device comprising: a second means for controlling the drive of the first means for recognition. 2. The first means has one or both of an XY stage for driving the transparent body and a rotary stage for rotationally driving the transparent body in a direction perpendicular to the pressing direction of the transparent body. The fingerprint recognition device according to claim 1. 3. The first means is configured to reduce the drive of the transparent body when the correlation value increases.
The fingerprint recognition device according to claim 1, wherein the driving of the means is controlled. 4. The recognition of the fingerprint image is performed by performing a congruent Fourier transform of an image obtained by picking up the fingerprint image and the pattern to be referenced by a spatial light modulator. The fingerprint recognition device according to 1. 5. The fingerprint recognition device according to claim 1, wherein the fingerprint image is recognized by detecting a match between the fingerprint image and the pattern to be referred within a predetermined time.