JP2003067749A - Device and method for collating finger-print, and program for collating finger-print - Google Patents

Abstract

PROBLEM TO BE SOLVED: To enhance precision for finger-print collation, by specifying positions of a fingertip and an articulation to get out finger-print portions of a collating object. SOLUTION: Sobel conversion (S32), and Gauss conversion (S33) are applied to an input image, and a line power value is found in every sectionally sliced line thereof (S34), to prepare an input pattern power value distribution 103. The input pattern power value distribution 103 is observed in order from its head part, and a line position in a spot where the line power value LPj exceeds a prescribed value consecutively in two lines or more is determined as an assigned position (S35). Then, the input pattern power value distribution is DP-collated with a standard pattern power value distribution to judge a line position of the articulate (S36). Finger-print images are got out in spots of the fingertip position and the articulate line position (S39) to extract features thereof.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a fingerprint collation device for identifying a person, a fingerprint collation method, and a computer program for performing fingerprint collation.

[0002]

2. Description of the Related Art In recent years, interest in security technology for controlling access to information has increased due to the rapid progress of computerization of information and networking, and fingerprint collation is one of such security technologies. A variety of products for performing personal authentication have been introduced.

In order to authenticate a person by using a fingerprint, it is necessary to first input a fingerprint image with a fingerprint sensor and then compare the registered fingerprint image with the input fingerprint image. The fingerprint verification is performed by extracting the features of the input fingerprint pattern and comparing the features with the registered features of the fingerprint.

The input sensor is mainly of a flat type for placing a finger on the input and a sweep type for pulling the finger to input. When the sensor surface of the flat type fingerprint sensor is large or when inputting with the sweep type fingerprint sensor, fingerprint input is possible from the fingertip to the first joint and beyond. The range of fingerprints to be input varies greatly depending on the input method. For example, the input may be from the fingertip to the very limit of the first joint, or even near the second joint.

[0005]

However, in the conventional fingerprint collation device or the like, even though the input range of the fingerprint varies depending on the input method, the input fingerprint is used as it is for registration and collation. There was a problem that the accuracy of fingerprint authentication was poor due to rejection and incorrect input. In order to improve the accuracy of fingerprint input, it is desirable to cut out only the fingertip part as a fingerprint, but for that purpose it is necessary to correctly recognize the position of the joint, and conventional fingerprint collation devices, etc. It was difficult to recognize joints without fail regarding fingerprints. Therefore, there has been a problem that the authentication accuracy is lowered due to a joint recognition error.

The present invention has been made in order to solve the above-mentioned problems, and a fingerprint collation device, a fingerprint collation method, and a fingerprint collation method are provided by identifying the positions of fingertips and joints and cutting out a fingerprint portion to be collated. The purpose is to improve the matching accuracy of the fingerprint matching program.

[0007]

In order to achieve the above object, a fingerprint collation device according to a first aspect of the present invention is a fingerprint input device for inputting a fingerprint image, and a fingerprint image from the fingerprint image input by the fingerprint input device. Fingerprint area cutout means for cutting out a fingerprint area suitable for collation, feature quantity extraction means for calculating and extracting a feature quantity for the fingerprint area cut out by the fingerprint area cutout means, and the feature quantity extraction means And a discriminating means for discriminating the fingerprint based on the characteristic amount.

In the fingerprint collation device having this structure, the feature amount is not extracted for the entire input fingerprint image, but the fingerprint portion to be collated is cut out in advance and the feature is extracted from the cut fingerprint portion. It is possible to improve the fingerprint authentication accuracy because the feature amount to be generated is not easily influenced by the variation of the input habit of the individual or the state at the time of input.

According to a second aspect of the fingerprint collating apparatus, in addition to the configuration of the fingerprint collating apparatus according to the first aspect, the fingerprint area cutting means includes joint identifying means for identifying a joint position in the fingerprint image. It is characterized by having.

In the fingerprint collating apparatus having this structure, the fingertip portion can be taken out for fingerprint collation by specifying the joint position when the fingerprint is cut out, and the extracted feature amount is the habit of personal input. The fingerprint authentication accuracy can be improved because it is not easily affected by variations due to the input and input conditions.

In the fingerprint collation device according to a third aspect, in addition to the configuration of the fingerprint collation device according to the first or second aspect, the fingerprint area cutout means identifies a fingertip position in the fingerprint image. It is characterized by having means.

In the fingerprint collating apparatus having this structure, the tip position of the finger is identified in order to take out the fingertip portion. Therefore, since the finger is placed under the sensor with a flat sensor or the like, the fingerprint is input on the upper portion of the input image. Even if it has not been done, it is possible to appropriately cut out the fingertip portion that is the target of fingerprint collation.

According to a fourth aspect of the fingerprint collating apparatus, in addition to the configuration of the fingerprint collating apparatus according to the third aspect, the fingerprint area cutting means includes the joint position identified by the joint identifying means and the fingertip identification. The fingerprint area is cut out based on the fingertip position identified by the means.

In the fingerprint collating apparatus having this structure, the fingerprint area is cut out between the fingertip position and the joint position, so that the fingertip portion to be subjected to fingerprint collation can be cut out appropriately and the fingerprint authentication accuracy can be improved.

According to a fifth aspect of the fingerprint collation device, in addition to the configuration of the fingerprint collation device according to any one of the second to fourth aspects, the joint identification means is shape information for extracting the shape information of the fingerprint image. The joint position is characterized by including extraction means, and comparing the shape information extracted by the shape information extraction means with the shape information of a standard fingerprint including a preset joint.

In the fingerprint collating apparatus having this structure, the fingerprint information extracted from a large number of people is averaged to form the shape information of the standard fingerprint, and the joint position is specified by comparing it with the shape information of the input fingerprint. The accuracy of joint identification is high, and thus the fingerprint authentication accuracy can be improved.

According to a sixth aspect of the fingerprint collating apparatus, in addition to the configuration of the fingerprint collating apparatus according to the fifth aspect, the joint identifying means uses the DP collating method to obtain the shape information and the shape information of the standard fingerprint. It is characterized by comparing.

In the fingerprint collation device having this structure, the joint position can be stably specified even with respect to the expansion and contraction of the input fingerprint in the vertical direction which occurs in the sweep type fingerprint sensor, and the fingerprint authentication accuracy is improved. Can be improved.

According to a seventh aspect of the fingerprint collating apparatus, in addition to the configuration of the fingerprint collating apparatus according to any one of the second to sixth aspects, the joint position identified by the joint identifying means is the first joint of the finger. It is a position.

In the fingerprint collating apparatus having this structure, the portion above the first joint is cut out as a fingertip portion by specifying the position of the first joint of the finger, and the collation is performed with this as a target. Can be improved.

The fingerprint collation method according to claim 8 includes a fingerprint input step of inputting a fingerprint image, and a fingerprint area cutting step of cutting out a fingerprint area suitable for fingerprint collation from the fingerprint image input in the fingerprint input step. A feature amount extraction step of calculating and extracting a feature amount of the fingerprint area cut out in the fingerprint area cutting step, and an identification step of identifying a fingerprint based on the feature amount extracted in the feature amount extraction step. It is characterized by having.

In the fingerprint collation method of this structure, the feature amount is not extracted for the entire input fingerprint image, but the fingerprint portion to be collated is cut out in advance and the feature is extracted from the cut out fingerprint portion. It is possible to improve the fingerprint authentication accuracy because the feature amount to be generated is not easily influenced by the variation of the input habit of the individual or the state at the time of input.

According to a ninth aspect of the fingerprint collation method, in addition to the configuration of the fingerprint collation method according to the eighth aspect, the fingerprint area cutting step includes a joint identification step of identifying a joint position in the fingerprint image. It is characterized by having.

In the fingerprint collation method of this configuration, by specifying the joint position when cutting out the fingerprint, the fingertip portion can be taken out and can be the target of fingerprint collation, and the extracted feature amount is the habit of personal input. The fingerprint authentication accuracy can be improved because it is not easily affected by variations due to the input and input conditions.

In the fingerprint collation method according to a tenth aspect, in addition to the configuration of the fingerprint collation method according to the eighth or ninth aspect, in the fingerprint area cutting step, a fingertip identification for identifying a fingertip position in the fingerprint image. It is characterized by having a process.

In the fingerprint collation method of this structure, since the tip position of the finger is identified in order to take out the fingertip portion, the fingerprint is input to the upper part of the input image because the finger is placed below the sensor with a flat sensor or the like. Even if it has not been done, it is possible to appropriately cut out the fingertip portion that is the target of fingerprint collation.

In the fingerprint collation method according to claim 11, in addition to the configuration of the fingerprint collation method according to claim 10, the fingerprint area cutting step includes the joint position identified by the joint identification step and the fingertip identification. The fingerprint area is cut out based on the fingertip position identified by the process.

In the fingerprint collation method of this configuration, by cutting out between the fingertip position and the joint position as a fingerprint area, it is possible to appropriately cut out the fingertip portion which is the target of fingerprint collation and improve the fingerprint authentication accuracy.

The fingerprint matching method according to claim 12 is the same as the fingerprint matching method according to any one of claims 9 to 11, and the joint identifying step includes shape information for extracting shape information of the fingerprint image. An extraction step is provided, and the joint position is identified by comparing the shape information extracted in the shape information extraction step with the shape information of a standard fingerprint including a preset joint.

In the fingerprint collation method of this structure, the fingerprint information extracted from a large number of people is averaged to form the shape information of the standard fingerprint, and the joint position is specified by comparing with the shape information of the input fingerprint. The accuracy of joint identification is high, and thus the fingerprint authentication accuracy can be improved.

A fingerprint collation method according to a thirteenth aspect is the same as the fingerprint collation method according to the twelfth aspect. In the joint identification step, the shape information and the standard fingerprint shape information are obtained by a DP collation method. It is characterized by comparing.

In the fingerprint collation method of this configuration, the joint position can be stably specified even with respect to the expansion and contraction of the input fingerprint in the vertical direction which occurs in the sweep type fingerprint sensor, and the fingerprint authentication accuracy is improved. Can be improved.

The fingerprint matching method according to a fourteenth aspect is the same as the fingerprint matching method according to any one of the ninth to thirteenth aspects, wherein the joint position identified in the joint identifying step is the first joint of the finger. It is a position.

In the fingerprint collation method of this configuration, by identifying the position of the first joint of the finger, the portion above the first joint is cut out as a fingertip portion and the collation is performed with this as the target, so that the fingerprint authentication accuracy is improved. Can be improved.

A fingerprint collation program according to a fifteenth aspect includes a fingerprint input step of inputting a fingerprint image, and a fingerprint area cutting step of cutting out a fingerprint area suitable for fingerprint collation from the fingerprint image input in the fingerprint input step. A feature amount extraction step of calculating and extracting a feature amount of the fingerprint area cut out in the fingerprint area cutting step, and an identification step of identifying a fingerprint based on the feature amount extracted in the feature amount extraction step. Let the computer run.

In the fingerprint collation program with this configuration, the feature amount is not extracted for the entire input fingerprint image, but the fingerprint portion to be collated is cut out in advance and the feature is extracted from the cut fingerprint portion. It is possible to improve the fingerprint authentication accuracy because the feature amount to be generated is not easily influenced by the variation of the input habit of the individual or the state at the time of input.

A fingerprint collation program according to a sixteenth aspect is the same as the fingerprint collation program according to the fifteenth aspect, and the fingerprint area cutting step includes a joint identification step for identifying a joint position in the fingerprint image. It is characterized by having.

In the fingerprint collation program with this configuration, by specifying the joint position when cutting out the fingerprint, the fingertip portion can be taken out and used as the target of fingerprint collation, and the extracted feature amount is the personal input habit. The fingerprint authentication accuracy can be improved because it is not easily affected by variations due to the input and input conditions.

A fingerprint collation program according to a seventeenth aspect is the same as the fingerprint collation program according to the fifteenth or sixteenth aspect, and in the fingerprint area cutting step, a fingertip identification for identifying a fingertip position in the fingerprint image. It is characterized by having a process.

In the fingerprint collation program with this configuration, the tip position of the finger is identified in order to take out the fingertip portion. Therefore, since the finger is placed under the sensor with a flat sensor or the like, the fingerprint is input at the upper part of the input image. Even if it seems not to be
It is possible to appropriately cut out the fingertip portion that is the target of fingerprint matching.

In the fingerprint collation program according to claim 18, in addition to the configuration of the fingerprint collation program according to claim 17, in the fingerprint area cutting step, the joint position identified by the joint identification step and the fingertip identification are performed. The fingerprint area is cut out based on the fingertip position identified by the process.

In the fingerprint collation program with this configuration, the area between the fingertip position and the joint position is cut out as a fingerprint area, so that the fingertip portion targeted for fingerprint collation can be cut out appropriately and the fingerprint authentication accuracy can be improved.

A fingerprint collation program according to a nineteenth aspect is the same as the fingerprint collation program according to any one of the sixteenth to eighteenth aspects, and the joint identification step includes shape information for extracting shape information of the fingerprint image. An extraction step is provided, and the joint position is identified by comparing the shape information extracted in the shape information extraction step with the shape information of a standard fingerprint including a preset joint.

In the fingerprint collation program having this configuration, the shape information of the standard fingerprint is created by averaging the fingerprint samples extracted from many people, and the joint position is specified by comparing with the shape information of the input fingerprint. The accuracy of joint identification is high, and thus the fingerprint authentication accuracy can be improved.

The fingerprint collation program according to claim 20 is the same as the fingerprint collation program according to claim 19, and in the joint identification step, the shape information and the shape information of the standard fingerprint are obtained by a DP collation method. It is characterized by comparing.

In the fingerprint collation program with this configuration, the joint position can be stably specified even with respect to the expansion and contraction of the input fingerprint in the vertical direction, which occurs in the sweep type fingerprint sensor, and the fingerprint authentication accuracy is improved. Can be improved.

A fingerprint collation program according to a twenty-first aspect is the same as the fingerprint collation program according to any one of the sixteenth to the twentieth aspects, and the joint position identified by the joint identification step is the first joint of the finger. It is a position.

In the fingerprint collation program of this configuration, the position above the first joint of the finger is specified, the portion above the first joint is cut out as the fingertip portion, and the collation is performed with this as the target. Can be improved.

[0049]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment in which the present invention is applied to an electronic lock equipped with a fingerprint collation device will be described below with reference to the drawings. The fingerprint collation device of the present invention is not limited to an electronic lock, but can be applied to various devices such as a PDA with a fingerprint lock function, a FAX, and a mobile phone that performs personal identification by fingerprint authentication. As shown in FIG. 1, the electronic lock 100 according to the present embodiment includes a fingerprint input device 7 including an input sensor 7a,
It has a fingerprint registration switch 8a, a fingerprint erasure switch 8b, and a liquid crystal display 10, which is attached to a door or the like and is locked by a plunger 9a which is moved by an electromagnetic solenoid 9 (see FIG. 2).

As shown in FIG. 2, the electronic lock 100 of this embodiment is composed of an electronic circuit centering on the CPU 1, and includes a RAM 2, a ROM 3, an EEP-ROM 4, an input sensor 7a, and a fingerprint registration switch 8a. The fingerprint erasing switch 8b, an input / output (I / O) port 5 that connects the electromagnetic solenoid 9 to the CPU 1, and a liquid crystal display controller (LCDC) 6 that controls the liquid crystal display 10.

As shown in FIG. 3, in the RAM 2, an input image memory 20 for temporarily storing a fingerprint image input from the fingerprint input device 7, a portion used for fingerprint collation is cut out from the input fingerprint image, and a fingerprint area is cut out. There is provided a fingerprint area memory 22 for storing as, and an input feature quantity memory 21 for storing the feature quantity extracted from the fingerprint area. Also, FIG.
As shown in, the EEP-ROM 4 is provided with a registered feature amount memory 41 for storing the feature amount of the input fingerprint when the fingerprint registration switch 8a is pressed. The registered feature amount memory only needs to be stored even if the power is turned off.
A flash memory may be used instead of the EEP-ROM. In addition, the ROM 3 stores various programs in addition to the fingerprint input program and the fingerprint collation program shown in the flowcharts of FIGS. 5 to 9, and controls the electronic lock 100.

The fingerprint input device 7 is composed of a strip-shaped input sensor 7a for inputting by pulling a finger and an A / D converter (not shown). The user inputs the sensor 7a.
When the finger is slid up, the temperature change of the finger moving on the sensor is output in the form of an input partial image and stored in the input image memory 20 in the RAM 2. The input sensor in the fingerprint input device 7 is not limited to such a sweep type sequential input sensor, and a flat sensor may be used. In this case, a minute change in capacitance formed by the pattern of a finger placed on the flat sensor is output in the form of an image.

Next, the operation of the electronic lock 100 of this embodiment will be described with reference to the flowcharts of FIGS. 5 to 9. For each step of the flowchart,
Abbreviated as S. The electronic lock 100 according to the present embodiment is configured such that the fingerprint registration switch 8a is pressed and then the input sensor 7a of the fingerprint input device 7 is pressed.
The fingerprint is registered by sliding the finger up, and the plunger 9a is ejected and locked. When a finger is slid on the input sensor 7a of the fingerprint input device 7 without pressing the fingerprint registration switch 8a, fingerprint collation is performed, and if the fingerprint is the same as the registered fingerprint, the plunger 9a is retracted and unlocked. Further, when the finger is slid on the input sensor 7a of the fingerprint input device 7 after pressing the fingerprint erasing switch 8b, the confirmation display is displayed on the liquid crystal display 10, and then the fingerprint erasing switch 8b is pressed again.
Delete the registered same fingerprint.

FIG. 5 is a flowchart showing the entire processing of the electronic lock 100. When the power is turned on, it is reset and the memory and the like in the RAM 2 are initialized (S10).
First, the fingerprint registration switch 8a and the fingerprint deletion switch 8b
If any of the buttons is pressed, this is detected (S1
1). If the fingerprint registration switch 8a is pressed (S1
2: YES), waiting for a fingerprint to be input from the input sensor 7a, cutting out a fingerprint portion to be collated from the input image (S13), and if enough fingerprints have been collated (S14: YES), the input fingerprint Extract image features and
It is stored in the input feature amount memory 21 provided in AM2 (S15). Then, it is further stored and registered in the registered feature amount memory 41 in the EEP-ROM 4 (S16). The fingerprint input process and the feature extraction process are described in the flowcharts of FIGS. 6 and 9, respectively, and a detailed description thereof will be given later.

Not the fingerprint registration switch 8a (S12: N
O), if the fingerprint erase switch 8b is pressed (S1
7: YES), fingerprint input processing is performed after waiting for a fingerprint to be input from the input sensor 7a, a fingerprint portion to be collated is cut out from the input image (S18), and if enough fingerprints have been collated (S19). : YES), the features of the cut out fingerprint image are extracted, and the extracted features and EEP-R
A fingerprint matching process is performed to match the fingerprint feature amount registered in the registered feature amount memory 41 in the OM 4 (S20). The fingerprint input process and the fingerprint collation process are described in the flowcharts of FIGS. 6 and 8, respectively, and the detailed description thereof will be described later. If they match as a result of collation (S21: YE
S), the feature amount of the corresponding fingerprint is deleted from the registered feature amount memory 41 of the EEP-ROM 4 and erase processing is performed (S2).
2). If they do not match (S21: NO), the process returns to S11 to detect the switch input.

If there is no switch input (S17: N
O), waiting for a fingerprint to be input from the input sensor 7a, performing fingerprint input processing, cutting out a fingerprint portion to be collated from the input image (S23), and if fingerprints sufficient for collation have been input (S24: YES). ), The characteristic of the cut out fingerprint image is extracted, and the fingerprint collation processing is performed to collate the extracted characteristic amount with the characteristic amount of the fingerprint registered in the registered characteristic amount memory 41 in the EEP-ROM 4 (S25). The fingerprint input process and the fingerprint collation process are described in the flowcharts of FIGS. 6 and 8, respectively, and the detailed description thereof will be described later. As a result of the collation, when it is determined that the input fingerprint matches the registered fingerprint (S26: YE
S) and pull the plunger 9a to unlock (S27). If no match is determined (S26: NO), S1
Returning to 1, the switch input is detected.

FIG. 6 is a flowchart of S1 of the flowchart shown in FIG.
3 is a flowchart showing details of fingerprint input processing performed in S3, S18, and S23. When the user slides his / her finger on the input sensor 7a to input a fingerprint, a fingerprint image as shown in the sensor fingerprint image 100 of FIG. 10 is acquired (S3).
1). This is an input image memory 2 provided in the RAM 2.
Store in 1. Next, for this input image Sobel
The pattern is applied by applying the conversion (S32). Sob
The image after el conversion is the Sobel conversion fingerprint image 1 in FIG.
It looks like 01. It should be noted that the Sobel transform is performed in order to emphasize the edge portion of the image. Depending on the nature of the fingerprint image, an image differentiating process may be applied instead of the Sobel transform to apply a filter for edge extraction. Good.

Next, the Gauss transform is applied to the image to crush and widen the pattern portion (S33). The Gauss-converted image becomes the Gauss-converted fingerprint image 102 in FIG. A filter for smoothing the image may be applied instead of the Gauss transform. Next, Ga
Line 102 obtained by slicing the uss-converted fingerprint image 102
The line power value LPj for each a is obtained (S34), and this is created in the input pattern power value distribution 103 as shown in FIG. The line power value LPj is obtained by summing the squares of the pixel values of the width of the input sensor 7a (280 dots in this embodiment) for each line, as shown in the following Expression 1. Details of the calculation method of the line power value LPj will be described later.

[Equation 1]

Next, looking at the input pattern power value distribution 103 from the beginning, the line position at the point where the line power value LPj exceeds the predetermined value for two or more lines in succession is determined as the fingertip position Ls (S35). When the finger is placed on the lower side of the flat sensor, the pixel value is small because it is not a fingerprint for a while from the beginning of the input image, and thus the line power value LPj is low. By utilizing this, it is determined that the part where the line power value LPj does not reach a certain level is the part that is not input. In the present embodiment, the fingertip position is determined when two or more lines continuously exceed the predetermined value so that the fingertip can be surely determined even when there is variation, but it is determined that even one line exceeds the predetermined value. You may do it. In this case, the processing can be further speeded up. Then, the input pattern power value distribution 103 and the RO
The line position of the joint is determined by DP collation with the standard pattern power value distribution 104 as shown in FIG. 14, which is preset in M3 (S36). Here, in the standard pattern power value distribution 104, the pattern power value distribution is created by applying the above S31 to S34 to a large number of fingerprints (for 200 persons in this embodiment), and the vertical length of these patterns is calculated. The values are normalized and then added to each other and averaged to be created in advance and stored in the ROM 3.

DP matching (dynamic programming) is DP comparison, DP
It is also called matching or rubber matching, and is widely used because it has the characteristic of being able to perform smooth matching even for slightly varying data strings. The DP collation is an algorithm that draws two series of reference data and test data on the xy plane and takes into account the distance difference of the actual data and selects the correspondence of the data series with the shortest cumulative distance. Here, the standard pattern power value distribution 104 is used as reference data, and the input pattern power value distribution 103 is used as test data to obtain mutual matching levels. Although the input data changes and changes in the reference data and the test data are different from each other, DP matching is used to search for the vertical correspondence that finally brings the reference data and test data closest to each other. The degree of matching under vertical correspondence is output as a distance value. Furthermore, in the DP collation here, as shown in FIG.
The collation route diagram 105 is provided with a start point free section 105a and an end point free section 105b. Then, the position that has actually become the end point in the end point free section 105b is determined as the joint line position Le. Although the fingertip position is determined by the value of the line power value LPj in S35, the position that is the actual start point in the start point free section 105a in DP matching may be determined as the fingertip position.

Next, the fingertip position Ls and the joint line position Le
The fingerprint length is calculated by calculating the difference (S37), and if it is less than the predetermined value (300 in this embodiment), the input fingerprint is invalidated (S38: NO). As a result, it is possible to improve collation accuracy by excluding a case in which the fingerprint has not been input up to the joint and ending re-inputting. Then, the fingerprint image is cut out at the points of the fingertip position Ls and the joint line position Le (S39), and FIG.
The cut-out fingerprint image 106 as shown in (1) is stored in the fingerprint area memory 22 in the RAM 2 and used for the feature extraction processing.

FIG. 7 is a flow chart showing the details of the calculation processing of the line power value LPj performed in S34 of FIG. First, the value of the number of lines j is initialized (S51). Next, the value of the width position i is initialized (S52). further,
The value of the line power value LPj is initialized (S53). Next, the square of the pixel value Pji at the current position is added to the current line power value LPj (S54). In this embodiment, 2
56 gradations are adopted, and the pixel value is a value between 0 and 255. Then, the value of the width position i is incremented by 1 (S55). Next, it is checked whether or not the value of the width position i has reached the maximum value of the width position i (S56), and if it has not reached (S56: NO), the pixel value of the adjacent dot is squared to S5.
4 and S55 are repeated. In this embodiment, the width is 280 dots, and the maximum value of i is 280. When i reaches the maximum value (S56: YES), the value of the line number j is incremented by 1 (S57). Then, it is checked whether the value of the line number j has reached the maximum value of the line number j (S5
8) If not reached (S58: NO), move to the next line and calculate the line power value (S54 to S).
Repeat up to 57). In this embodiment, the number of lines is three.
00 to 700 lines. When j reaches the maximum value (S58: YES), the calculation of the line power value LPj is completed for all the lines, and the process is ended.

FIG. 8 is a flow chart showing details of the fingerprint collation processing executed in S20 and S25 of FIG. First, the extracted fingerprint image 106 cut out at the fingertip position Ls and the joint line position Le stored in the fingerprint area memory 22 in the RAM 2 is extracted (S9).
1), it is stored in the input feature amount memory 21 provided in the RAM 2 (S92). Details of the feature extraction processing will be described later. Next, DP matching (dynamic programming) is performed between the feature amount stored in the input feature amount memory 21 and the feature amount of the registered fingerprint stored in the registered feature amount memory 41 of the EEP-ROM 4 (S93). ).

In the DP collation performed here, the LPC cepstrum (details will be described later) obtained by the feature extraction processing is regarded as spectrum data for each fingerprint line, and the LPC cepstrum stored in the input feature amount memory 21 is used as the vertical direction of the fingerprint image. The degree of mutual matching is obtained by using the test data arranged in the direction and the LPC cepstrum stored in the registered feature amount memory 41 of the EEP-ROM 4 arranged in the vertical direction of the fingerprint image as the reference data. The reference data and test data have different input speed changes and changes,
The DP matching is used to search for the vertical correspondence that finally brings the reference data and test data closest to each other, and the matching degree under the closest vertical correspondence is output as a distance value. . A smaller distance value indicates a smaller difference between both LPC cepstrums, and a larger distance value indicates a larger difference between both LPC cepstrum.

Next, the distance value obtained by the DP collation is compared with a preset threshold value, and if the distance value is smaller than the threshold value, it is determined that they match, and if the distance value is greater than the threshold value, it is determined that they do not match. Yes (S94). The fingerprint collation processing is not limited to the above-described form, and can be performed using a known fingerprint collation technique such as minutiae method or pattern matching method.

FIG. 9 is a flow chart showing details of the feature extraction processing performed in S91 of FIG. 8 and S15 of FIG. As the feature extraction processing, a feature point extraction matching method (minutiae method), an image matching method (pattern matching method), etc. are known as well-known techniques, but in the present embodiment, feature points (minutiae) are extracted from a fingerprint. Instead, the ridge information itself is used for analysis, and processing is performed as the feature amount. That is, cut out the fingerprint image in the horizontal direction,
By taking the fingerprint position in the horizontal direction and the unevenness (shading) of the fingerprint in the vertical direction, one line of the cut-out fingerprint can be regarded as a waveform signal. Then, a power spectrum can be obtained by performing frequency analysis on this waveform signal, which is generally performed in audio data processing.

Specifically, each line of the fingerprint image subjected to the background separation processing is set as one frame, and the extraction processing is performed for each frame. First, as a pre-process, a Hamming window is applied (S100) to reduce the influence of the frame cutout on the end portion. Next, the frame data corrected by the Hamming windowing is received, and its autocorrelation function is obtained (S101). Further, based on the obtained autocorrelation function, a linear prediction method (LPC: Liner Predictive Cordin
The LPC coefficient according to g) is calculated and obtained (S102).
The linear prediction method is a well-known technique used for voice compression in mobile phones, etc. When this is used, voices such as vowels can be estimated with an all-pole transfer function, and the peak frequency can be estimated from the spectrum. Etc. can be estimated. Then, the LPC coefficients obtained in S102 are linearly combined to obtain the LPC cepstrum (S103). In the present embodiment, the LPC obtained in the feature amount extraction process in this way
The cepstrum is stored in the input feature amount memory 21 as a feature amount (S92).

As described above, according to the fingerprint collating apparatus of the present embodiment, the line power value LPj is calculated, the fingertip position Ls and the joint line position Le are calculated, and the cut-out fingerprint image 106 is extracted by cutting between them. The feature extraction processing is performed for this. Therefore, only the fingerprint feature amount from the fingertip to the first joint can be registered and used for matching, and the accuracy of fingerprint matching can be improved.

Although the line power value LPj is used as the parameter for obtaining the fingerprint pattern distribution in the present embodiment, the run length may be used instead. In this case, a step of binarizing is inserted after the Gauss transform (S33) to create a binarized image of the fingerprint, and the pixel value of 1 (fingerprint is The maximum length in which a certain area) appears continuously can be calculated and used as a parameter.

[0070]

As is apparent from the above description, claim 1
According to the fingerprint collation device described in (1), instead of extracting the feature amount for the entire input fingerprint image, the fingerprint portion to be collated is cut out in advance, and the feature is extracted from the cut out fingerprint portion. It is possible to improve the fingerprint authentication accuracy because the feature amount is not easily influenced by the individual's habit of inputting and the variation due to the state at the time of inputting.

According to the fingerprint collation device of the second aspect,
In addition to the effect of the fingerprint collation device according to claim 1, by specifying a joint position when cutting out a fingerprint, a fingertip portion can be taken out to be a target of fingerprint collation, and the extracted feature amount is an individual feature. The accuracy of fingerprint authentication can be improved because it is less affected by variations in input habits and input conditions.

According to the fingerprint collation device of the third aspect,
In addition to the effect of the fingerprint collation device according to claim 1 or 2, since the tip position of the finger is identified in order to take out the fingertip portion,
Even when the fingerprint is not input to the upper portion of the input image because the finger is placed below the sensor with a flat sensor or the like, the fingertip portion that is the target of fingerprint matching can be appropriately cut out.

According to the fingerprint collation device of the fourth aspect,
In addition to the effect of the fingerprint collation device according to claim 3, by cutting out between the fingertip position and the joint position as a fingerprint region,
It is possible to improve the fingerprint authentication accuracy by appropriately cutting out the fingertip portion that is the target of fingerprint collation.

According to the fingerprint collation device of the fifth aspect,
In addition to the effect of the fingerprint collation device according to any one of claims 2 to 4, shape information of a standard fingerprint is created by averaging fingerprint samples extracted from a large number of people and compared with shape information of an input fingerprint. Since the joint position is specified by the joint, the joint identification accuracy is high, and the fingerprint authentication accuracy can be improved.

According to the fingerprint collation device of the sixth aspect,
In addition to the effect of the fingerprint collation device according to claim 5, the joint position can be stably specified even with respect to expansion and contraction in the vertical direction of the input fingerprint that occurs in the sweep type fingerprint sensor. The authentication accuracy can be improved.

According to the fingerprint collation device of claim 7,
In addition to the effect of the fingerprint collation device according to any one of claims 2 to 6, by specifying the position of the first joint of the finger, a portion above the first joint is cut out as a fingertip portion and collated with this as a target. Therefore, the fingerprint authentication accuracy can be improved.

According to the fingerprint collation method described in claim 8,
The feature amount is not extracted for the entire input fingerprint image, but the fingerprint portion to be matched is cut out in advance and the feature is extracted from the cut fingerprint portion. It is unlikely to be affected by variations due to input conditions, and fingerprint authentication accuracy can be improved.

According to the fingerprint collation method of claim 9,
In addition to the effect of the fingerprint collation method according to claim 8, by specifying the joint position when cutting out the fingerprint, the fingertip portion can be taken out and can be the target of fingerprint collation, and the extracted feature amount is the The accuracy of fingerprint authentication can be improved because it is less affected by variations in input habits and input conditions.

According to the fingerprint collation method of the tenth aspect, in addition to the effect of the fingerprint collation method of the eighth or ninth aspect, since the tip position of the finger is identified to take out the fingertip portion, a flat sensor or the like is used. Even when the fingerprint is not input to the upper part of the input image because the finger is placed under the sensor at, the fingertip portion which is the target of fingerprint collation can be appropriately cut out.

According to the fingerprint collation method of the eleventh aspect, in addition to the effect of the fingerprint collation method of the tenth aspect, it is possible to appropriately perform the fingerprint collation by cutting out a portion between the fingertip position and the joint position as a fingerprint area. It is possible to improve the fingerprint authentication accuracy by cutting out the target fingertip portion.

According to the fingerprint collation method of the twelfth aspect, in addition to the effect of the fingerprint collation method of any of the ninth to eleventh aspects, a fingerprint sample extracted from a large number of persons is averaged to obtain a standard. Since the shape information of the fingerprint is created and the joint position is specified by comparing with the shape information of the input fingerprint, the accuracy of the joint identification is high and the fingerprint authentication accuracy can be improved.

According to the fingerprint collation method of the thirteenth aspect, in addition to the effect of the fingerprint collation method of the twelfth aspect, with respect to the expansion and contraction of the input fingerprint in the vertical direction which occurs in the sweep type fingerprint sensor. However, the joint position can be stably specified, and the fingerprint authentication accuracy can be improved.

According to the fingerprint collation method of the fourteenth aspect, in addition to the effect of the fingerprint collation method of any one of the ninth to thirteenth aspects, the position of the first joint of the finger is specified. Cut out the part above the joint as the fingertip part,
Since the verification is performed with this as a target, the fingerprint authentication accuracy can be improved.

According to the fingerprint collation program of the fifteenth aspect, the feature amount is not extracted for the entire input fingerprint image, but the fingerprint portion to be collated is cut out in advance and the feature is extracted from the cut out fingerprint portion. Since it is extracted, the extracted feature amount is unlikely to be affected by variations in individual input habits and input states, and fingerprint authentication accuracy can be improved.

According to the fingerprint collation program of the sixteenth aspect, in addition to the effect of the fingerprint collation program of the fifteenth aspect, by specifying the joint position when cutting out the fingerprint, the fingertip portion is taken out and the fingerprint collation is performed. It is possible to improve the fingerprint authentication accuracy because the extracted feature amount is unlikely to be affected by variations in the input habits of the individual and the state at the time of input.

According to the fingerprint collation program of the seventeenth aspect, in addition to the effect of the fingerprint collation program of the fifteenth or sixteenth aspect, since the tip position of the finger is identified in order to take out the fingertip portion, a flat sensor or the like is used. Even when the fingerprint is not input to the upper part of the input image because the finger is placed under the sensor at, the fingertip portion which is the target of fingerprint collation can be appropriately cut out.

According to the fingerprint collation program of the eighteenth aspect, in addition to the effect of the fingerprint collation program of the seventeenth aspect, by cutting out a space between the fingertip position and the joint position as a fingerprint area, the fingerprint collation program can be properly performed. It is possible to improve the fingerprint authentication accuracy by cutting out the target fingertip portion.

According to the fingerprint collation program described in claim 19, in addition to the effect of the fingerprint collation program according to any one of claims 16 to 18, the fingerprint samples extracted from a large number of people are averaged to obtain a standard. Since the shape information of the fingerprint is created and the joint position is specified by comparing with the shape information of the input fingerprint, the accuracy of the joint identification is high and the fingerprint authentication accuracy can be improved.

According to the fingerprint collation program of the twentieth aspect, in addition to the effect of the fingerprint collation program of the nineteenth aspect, with respect to the expansion and contraction of the input fingerprint in the vertical direction which occurs in the sweep type fingerprint sensor. However, the joint position can be stably specified, and the fingerprint authentication accuracy can be improved.

According to the fingerprint collation program of the twenty-first aspect, in addition to the effect of the fingerprint collation program of any one of the sixteenth to twentieth aspects, the position of the first joint of the finger is specified. Since the portion above the joint is cut out as a fingertip portion and collation is performed with this as a target, the fingerprint authentication accuracy can be improved.

[Brief description of drawings]

FIG. 1 is an external view of an electronic lock 100 according to this embodiment.

FIG. 2 is a block diagram showing a configuration of an electronic lock 100 of the present embodiment.

FIG. 3 is a memory configuration diagram of a RAM 2.

FIG. 4 is a memory configuration diagram of an EEP-ROM 4.

5 is a flowchart showing an outline of a processing flow of the electronic lock 100. FIG.

FIG. 6 is a flowchart showing details of fingerprint input processing.

FIG. 7 is a flowchart showing details of line power value calculation processing.

FIG. 8 is a flowchart showing details of fingerprint collation processing.

FIG. 9 is a flowchart showing details of feature extraction processing.

FIG. 10 is an example of an input image of a fingerprint.

FIG. 11 is an example of a fingerprint image after Sobel conversion.

FIG. 12 is an example of a fingerprint image after Gauss conversion.

FIG. 13 is a pattern power value distribution diagram of an input fingerprint.

FIG. 14 is a pattern power value distribution diagram of a standard fingerprint.

FIG. 15 is a DP verification route diagram.

FIG. 16 is an example of a cut out fingerprint image.

[Explanation of symbols]

1 CPU 2 RAM 3 ROM 4 EEP-ROM 7 Fingerprint input device 7a Fingerprint sensor 8a Fingerprint registration switch 8b Fingerprint erase switch 20 Input image memory 21 Input feature memory 22 Fingerprint area memory 41 Registered feature memory 100 electronic locks

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Taizo Umezaki             176-118, Akasaka-cho, Tajimi-shi, Gifu (72) Inventor Kenshi Miyoshino             4-13, Ogashirabashi, Nakagawa-ku, Nagoya             Ceremony company DDS F-term (reference) 5B043 AA04 AA09 BA02 EA02 EA05                       EA15 FA03 GA02

Claims (21)

[Claims] 1. A fingerprint input means for inputting a fingerprint image, a fingerprint area cutout means for cutting out a fingerprint area suitable for fingerprint collation from the fingerprint image input by the fingerprint input means, and a fingerprint area cutout means for cutting out the fingerprint area. A fingerprint collation device, comprising: a feature amount extraction means for calculating and extracting a feature amount for the fingerprint region thus extracted, and an identification means for identifying a fingerprint based on the feature amount extracted by the feature amount extraction means. . 2. The fingerprint collation apparatus according to claim 1, wherein the fingerprint area cutting means includes joint identification means for identifying a joint position in the fingerprint image. 3. The fingerprint collation device according to claim 1, wherein the fingerprint area cutting unit includes a fingertip identifying unit that identifies a fingertip position in the fingerprint image. 4. The fingerprint area cutout unit cuts out the fingerprint area based on a joint position identified by the joint identification unit and a fingertip position identified by the fingertip identification unit. 3. The fingerprint collation device according to item 3. 5. The joint identification means includes shape information extraction means for extracting shape information of the fingerprint image, and the shape information extracted by the shape information extraction means and a standard fingerprint including a preset joint are included. The fingerprint collation device according to claim 2, wherein the joint position is identified by comparing the joint position with shape information. 6. The fingerprint collation apparatus according to claim 5, wherein the joint identification unit compares the shape information with the shape information of the standard fingerprint by a DP collation method. 7. The fingerprint collation apparatus according to claim 2, wherein the joint position identified by the joint identification means is the position of the first joint of the finger. 8. A fingerprint input step of inputting a fingerprint image, a fingerprint area cutout step of cutting out a fingerprint area suitable for fingerprint collation from the fingerprint image input in the fingerprint input step, and a fingerprint area cutout step of cutting out the fingerprint area. A fingerprint collation method comprising: a feature amount extraction step of calculating and extracting a feature amount of the fingerprint region that has been extracted; and an identification step of identifying a fingerprint based on the feature amount extracted in the feature amount extraction step. . 9. The fingerprint collation method according to claim 8, wherein the fingerprint region cutting step includes a joint identification step of identifying a joint position in the fingerprint image. 10. The fingerprint collation method according to claim 8, wherein the fingerprint region cutting step includes a fingertip identification step of identifying a fingertip position in the fingerprint image. 11. The fingerprint area cutout step cuts out the fingerprint area based on a joint position identified by the joint identification step and a fingertip position identified by the fingertip identification step. 10. The fingerprint matching method described in 10. 12. The joint identification step includes a shape information extraction step of extracting shape information of the fingerprint image, and the shape information extracted in the shape information extraction step,
12. The fingerprint matching method according to claim 9, wherein the joint position is identified by comparing the shape information of a standard fingerprint including a preset joint. 13. The fingerprint matching method according to claim 12, wherein in the joint identifying step, the shape information and the shape information of the standard fingerprint are compared by a DP matching method. 14. The fingerprint collation method according to claim 9, wherein the joint position identified by the joint identification step is the position of the first joint of the finger. 15. A fingerprint input step of inputting a fingerprint image, a fingerprint area cutout step of cutting out a fingerprint area suitable for fingerprint collation from the fingerprint image input in the fingerprint input step, and a fingerprint area cutout step of cutting out the fingerprint area. A fingerprint collation program that causes a computer to execute a characteristic amount extraction step of calculating and extracting a characteristic amount of a fingerprint area that has been extracted, and an identification step of identifying a fingerprint based on the characteristic amount extracted in the characteristic amount extraction step. 16. The fingerprint collation program according to claim 15, wherein the fingerprint area cutting step includes a joint identification step of identifying a joint position in the fingerprint image. 17. The fingerprint collation program according to claim 15, wherein the fingerprint area cutting step includes a fingertip identification step of identifying a fingertip position in the fingerprint image. 18. The fingerprint area cutout step cuts out the fingerprint area based on a joint position identified by the joint identification step and a fingertip position identified by the fingertip identification step. 17. The fingerprint matching program described in 17. 19. The joint identification step includes a shape information extraction step of extracting shape information of the fingerprint image, and the shape information extracted in the shape information extraction step,
19. The fingerprint collation program according to claim 16, wherein the joint position is identified by comparing shape information of a standard fingerprint including a preset joint. 20. The fingerprint collation program according to claim 19, wherein the joint identification step compares the shape information with the shape information of the standard fingerprint by a DP collation method. 21. The fingerprint collation program according to claim 16, wherein the joint position identified by the joint identification step is the position of the first joint of the finger.