JPH0318980A - Fingerprint collating device - Google Patents

Abstract

PURPOSE:To obtain high collation accuracy irrespective of pressing force by providing a superposing means for obtaining the protrusion quantity in the positive and the negative directions, a pressing force sensor for detecting the pressing force at the time of input, and a fuzzy controller. CONSTITUTION:Image data inputted from an image input device 1 is inverted and superposed on reference image data held in a reference image memory 3, and also, the reference image data is inverted and superposed on the inputted image. The device is provided with a superposing means 5 for obtaining the protrusion quantity in the positive and the negative directions, respectively, a pressing force sensor 2 for detecting the pressing force at the time of inputting a fingerprint image, and a fuzzy controller 8 for executing a fuzzy inference so that the degree of coincidence of a collation becomes large when the protrusion quantity is small, and the degree of coincidence of a fingerprint is raised at the time of becoming the protrusion quantity in the positive direction and the protrusion quantity in the negative direction, when the pressing force becomes large and when the pressing force becomes small, respectively. In such a way, a correct collation result can be obtained irrespective of the pressing force.

Description

DETAILED DESCRIPTION OF THE INVENTION [Field of the Invention] The present invention relates to a fingerprint matching device that determines whether image data of a human-generated fingerprint matches image data of a predetermined fingerprint.

[Conventional technology]

Conventional verification devices for fingerprints, seals, etc.
As shown in No. 54, etc., by inverting the input image using pre-held reference image data and input image data and superimposing it on the reference image data, the entire image can be made to have the same color. Then, by inverting the reference image data and superimposing it on the input image data, it is determined by two comparisons whether the superimposed images have the same color, and the correct seal or stamp is determined. It is designed to detect patterns such as fingerprints.

[Problem to be solved by the invention]

However, when comparing fingerprints using such conventional matching devices, the input image becomes thicker or thinner due to the pressure applied when the fingerprint is collected, which has the disadvantage that it does not match the standard pattern. . Therefore, when comparing fingerprints, there is a drawback that even if a person inputs a fingerprint, it may not be determined to be the person's fingerprint.

The invention of claim 1 of the present application was made in view of the problems of the conventional fingerprint verification device, and the technical problem is to be able to obtain a correct verification result regardless of the pressing force. shall be.

Moreover, the technical object of the invention of claim 2 of the present application is to improve the security when this fingerprint verification device is used in an automatic teller machine or the like.

[Means to solve the problem]

The invention according to claim 1 of the present application is a fingerprint verification device for verifying fingerprints, comprising a reference image memory for holding image data of a reference fingerprint and an image input device for inputting the fingerprint image, and comprising:
The input image data is inverted and superimposed on the reference image data held in the reference image memory, and the reference image data is inverted and superimposed on the input image, and the amount of protrusion in the positive and negative directions is calculated respectively. and a pressing force sensor that detects the pressing force when inputting the fingerprint image.If the amount of protrusion is small, the matching degree of verification is increased, and if the pressing force is large, the amount of protrusion in the positive direction and the pressing force are Multiple fuzzy inference units that perform fuzzy inference to increase the degree of fingerprint matching when the amount of protrusion in the negative direction becomes smaller, and a determination system that obtains a non-fuzzy definitive value based on the parallel output of each fuzzy inference unit. It is characterized by having the following parts.

In addition, in the invention of claim 2 of the present application, the fingerprint verification device further includes a number of people determination device that detects the number of people in the vicinity of the device, and the fuzzy inference unit determines that the fingerprints do not match when a plurality of people are determined. It is characterized by reasoning.

[Effect]

According to the present invention having such characteristics, the input image data of the fingerprint reference image data held in the memory and the collected technical image data are inverted and superimposed, and the reference image data is inverted on the input image. The protrusion amounts in the positive and negative directions are obtained by superimposing them. If the pressing force is appropriate, the degree of matching will be high when the amount of protrusion is near zero, and if the pressing force is large, the pattern of the collected fingerprint will become thicker, so the degree of matching will increase when the amount of protrusion shifts in the positive direction. If the pressing force is small, the pattern of the sampled fingerprint becomes thinner, so fuzzy inference is performed so that the amount of protrusion increases the degree of matching in the negative direction.Then, based on the parallel outputs of each fuzzy inference section, the fingerprint pattern is I am trying to get a matching result.

In addition, in the invention of claim 2 of the present application, the number of people in the vicinity of the fingerprint verification device is determined by the number of people discriminating device and inputted, and when a plurality of people are detected, the fingerprint verification output is made to be a mismatch.

〔Effect of the invention〕

Therefore, according to the present invention, it is possible to obtain a high verification accuracy regardless of the pressing force of the person inputting the fingerprint. Furthermore, in the invention of claim 2 of the present application, when there are a plurality of people in the vicinity of the fingerprint verification device, it is possible to prevent an abnormal state when used in an automatic teller machine etc. by detecting an abnormality. You can get the effect that you can.

[Explanation of Examples]

FIG. 2 is a block diagram showing the overall configuration of a fingerprint matching device according to an embodiment of the present invention. In this figure, this fingerprint verification device is provided with an image input device 1 for inputting image data of a fingerprint, and a pressing force sensor 2 for detecting the pressing force at that time. The output of the image input device 1 and the output of the reference image memory 3 in which reference fingerprint information is held are respectively applied to a switching circuit 4. The switching circuit 4 switches the signals at these two input terminals and provides them to two output terminals, one output terminal 4a is connected to a superimposing circuit 85, and the other output terminal 4b is connected to a black and white signal. The superposition circuit 5 is connected to the superposition circuit 5 through the inversion circuit 6. The superposition circuit 5 first inverts the human image and then the reference image data and superimposes them on the other input image, thereby removing any protruding images. The area of the portion is output as positive and negative signals, respectively. Here, the image input device 12 image memory 3 to black-and-white inversion circuit 6 constitute a superimposition means for inverting and superimposing either the reference image or the input image, and the first input of the fuzzy controller 8 given to the department. Further, this fingerprint verification device is provided with a number of people determination device 7 that determines the number of people in the vicinity of the device, and both the determination output and the output of the pressing force sensor 2 are sent to a fuzzy controller 8.
given to. The fuzzy controller 8 determines whether the fingerprints match based on these human inputs, and the determination result is displayed on the display 9.

Next, the structure of the main parts including the fuzzy controller 1 and the roller 8 will be explained with reference to FIG. In the present invention, fingerprints are compared using the amount of protrusion resulting from superposition with the reference image of the superimposition circuit 5, the number of people, and the output of the pressing force sensor 2. The fuzzy controller 8 is made up of a plurality of fuzzy inference sections and a determining section. Each input shown in Figure 2 has multiple inputs,
In this embodiment, the information is transmitted to eight fuzzy inference units 11-1-118. As shown in the figure, each fuzzy inference section includes the output of the superimposition circuit 5 serving as input, the output of the one-person discrimination device 7, a membership function generation circuit (MFC) that generates a membership function corresponding to the pressing force, and a fingerprint to be output. A membership function generator (MFG) is provided that generates a membership function corresponding to the matching degree of matching. In this embodiment, as a membership function, the output of the superposition circuit 5 is divided into seven states as shown in FIG. The pressing force is divided into four states. Here, NL is a large negative value, NM is a medium negative value, NS is a small negative value, ZR is almost zero, PS is a small positive value, PM is a medium positive value, and PL is a positive medium value. Each represents a large value. In this case, as the pressing force increases, the pattern of the obtained fingerprint also becomes thicker, so the output of the superimposition circuit 5 shifts in the positive direction.If the pressing force of the pressed fingerprint is small, the pattern of the obtained fingerprint becomes thinner, so the output of the superposition circuit 5 shifts in the negative direction. Furthermore, the matching degree of verification is also classified into four states of ZR-PL as shown in FIG. 3 Fdl. The fuzzy determining unit I2 calculates the maximum value of each output unit based on the parallel outputs obtained from each fuzzy inference unit. Then, fuzzy definite values are obtained based on the centroid values of these outputs. This output is then transmitted to the display 9 to display the collated results.

Next, the fuzzy inference units 11-1 to 11-8 will be explained. Each fuzzy inference unit has two or three MFC21
-1 to 21-8. 22-1 to 22-8. 23-2~
It has 23-8. The fuzzy inference unit 11-1 is an inference unit that performs the inference of rule 1 as described later, and is
21-1 is a generation circuit (MFC) in which the input Sl from the superposition circuit 5 generates a membership function of ZR, M
The FC 22-1 is an MFC in which the output S2 from the number of people determining device 7 generates a PM membership function, and each output is given to the MIN circuit 24-1. Also MFG25-
1 is a membership function generator (MFG) that generates a parallel membership function ZR with a degree of coincidence shown in FIG.
and its parallel output is M! N circuit 26-1. The MIN circuit 26-1 generates a smaller parallel fuzzy signal by comparing the outputs of the MIN circuits 24-1, and its output is given to the MAX array circuit 27.

Next, the inference rules determined based on these inputs are shown below.

(Rule 1) If the overhang l (SL) is ZR and the number of people (S2) is PM
If so, let the matching degree (e) be ZR.

This rule 1 is simplified and expressed as follows.

If 51=ZRand S2=PM then ZR
Such a rule states that if the pressing force is normal, the amount of protrusion is highest near zero, as the pressing force increases, the amount of protrusion increases in the positive direction, and as the pressing force decreases, the amount of protrusion increases in the negative direction. When the number of people matches, the match is high, and vice versa, the match is low, and when there are a plurality of people, the match is determined to be a mismatch. Such rules are shown in the table below.

In this way, eight inference rules are determined.

The fuzzy inference units 11-1 to 11-8 are inference units that execute respective fuzzy inferences, and the parallel fuzzy outputs that are the inference results of the respective rules are sent to the fuzzy inference units 11-1-11-8.
is applied to the MAX array circuit 27. The MAX array circuit 27 calculates the maximum value for each corresponding parallel line, and its parallel output is given to the defassifier circuit 28. The defassifier circuit 28 obtains a non-fuzzy output by calculating the center of gravity of its output, and the output is given to the display 9.

Next, the operation of this embodiment will be explained. First, the output St of the superposition circuit 5 is a predetermined value S1°, and N5=0.7
, NM=0.3, the output of the number of people determining device 7 is 1, that is, S2 is ps, the pressing force 33 is a predetermined value S3°, and PM=0.6 and P S =0.4 as shown. In this case, Rule 2.6 of the rule table mentioned above is applied, and the output of the degree of matching e is as follows.

PM=0.4, PM=0.3 Therefore, fuzzy parallel outputs are obtained from one of the eight fuzzy inference units, and these outputs are subjected to MAX calculation by the MAX array circuit 27 of the determining unit 12. be exposed. As a result, the degree of coincidence e1 is calculated by calculating the center of gravity as shown in FIG. 3(d). This political change will be displayed on the display 9.

Although this embodiment describes a fingerprint matching device that performs fuzzy inference according to the MIN-MAX calculation rule, the present invention can also be realized by fuzzy inference according to other calculation rules.

Furthermore, fuzzy processing is performed using a dedicated device for fuzzy inference (for example, Nikkei Electronics, July 28, 1987,
(Pages 148 to 152, page 5, Nikkei McGraw-Hill), but also a binary-type computer processor programmed to perform fuzzy inference.

Further, the membership function is not limited to the triangular shape shown in FIG. 3, but may be of any shape. Further, the membership functions and inference rules can be changed or modified as appropriate depending on the matching accuracy and the like.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing the main parts of a fingerprint verification device according to an embodiment of the present invention, FIG. 2 is a block diagram showing its overall configuration, and FIG. (bl is the number of people, Figure 3 (C) is the pressing force, Figure 3 (d)
l is a diagram showing a membership function of degree of coincidence. 1--Image input device 2-Press force sensor 3--1-Reference image memory 4--Partial replacement circuit 5--
--Superposition circuit 6----Black and white inversion circuit 8-Fuzzy controller 9-1---Display device 11-1
~11-8----Fuzzy inference section 12-・Determining section 21, 22.23----MFC24゜26----
M I N circuit 25---MFG 27M
AX array circuit 28---- differential asifier circuit

Claims (2)

[Claims] (1) A fingerprint matching device for comparing fingerprints, which has a reference image memory that holds image data of a reference fingerprint and an image input device that inputs the fingerprint image, the reference image being held in the reference image memory. a superposition means for inverting and superimposing input image data on the data, further inverting and superimposing reference image data on the input image, and obtaining protrusion amounts in positive and negative directions, respectively; and a fingerprint image. A pressing force sensor detects the pressing force when inputting , and if the protruding amount is small, the degree of matching is increased, if the pressing force is large, the protruding amount is in the positive direction, and if the pressing force is small, the protruding amount is in the negative direction. a plurality of fuzzy inference units that perform fuzzy inference to increase the degree of fingerprint matching when A unique fingerprint verification device. (2) The fingerprint verification device further includes a number of people determination device that detects the number of people in the vicinity of the device, and the fuzzy inference unit infers that the fingerprints do not match when a plurality of people are determined. The fingerprint verification device according to claim 1, characterized in that: