JP2698453B2 - Fingerprint matching method - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION [Summary] A fingerprint collation method for collating a fingerprint is intended to perform fingerprint collation with high reliability. The first invention presses a finger against a fingerprint sensor. The fingerprint to be read out between the frames is read out as a plurality of frame images, and among the fingerprint feature information extracted for each of the frame images, the fingerprint feature information is set as the feature amount of the fingerprint in the order of the number of matching points over the plurality of frame images. While the fingerprint is read as a plurality of frame images while the finger to be collated is pressed against the fingerprint sensor stored in the reference data file in advance, the feature amount of the fingerprint read from the reference data file for each of the read frame images and Comparing, when at least one frame image has a coincidence point with the feature value equal to or more than a predetermined threshold value, or when the number of frame images exceeds a predetermined number, When the matching point with the feature amount is equal to or greater than a predetermined threshold value, it is determined that the fingerprint read from the reference data file matches the fingerprint of the finger to be collated. When creating and storing the file, the finger is pressed against the fingerprint sensor, the fingerprint is read as one frame image, and the characteristic information of the fingerprint extracted from the frame image is stored in advance in the reference data file as the characteristic amount of the fingerprint, While reading the fingerprint as a plurality of frame images while pressing the finger to be collated against the fingerprint sensor, each read frame image is compared with the feature amount of the fingerprint read from the reference data file for each of the read frame images.
If the number of coincidences with the feature in one frame image is equal to or greater than a predetermined threshold, or if the number of coincidences with the feature is equal to or greater than a predetermined threshold over a predetermined number or more of frame images, the reference data file Is configured to determine that the fingerprint read out from the fingerprint matches the fingerprint of the finger to be collated.

The present invention relates to a fingerprint matching method for matching fingerprints.

In recent years, with the introduction of computers into a wide range of social systems, stakeholders have been interested in system security. ID cards and passwords, which have been used as a means of entering a computer room and confirming the identity of a terminal, have raised many questions in terms of security. On the other hand, fingerprints are considered to be the most powerful means of identity verification because they have two major characteristics, that is, everyone's identity and lifelong invariance, and much research and development has been conducted on a simple personal identification system using fingerprints. I have.

[Prior Art] FIG. 4 is a conceptual diagram showing the configuration of a conventional fingerprint matching system. First, the operation at the time of registration will be described. A fingerprint pattern is detected by pressing a finger on the fingerprint sensor 1 and converted into digital data by an A / D converter (not shown) of the fingerprint sensor. The converted digital data is converted into binary data of 0 and 1 by the following binarization circuit 2 and stored in the frame memory 3.

After the image data stored in the frame memory 3 is sequentially read out, the image data enters a feature information extracting / collating circuit 4 where the feature information is extracted. Here, the characteristic information means, for example, a branch point as shown in FIG. 5 (a), an end point as shown in (b), and the like. A fingerprint can be specified based on where and how many such branch points and end points exist. There is no different person having exactly the same feature information. The extracted feature information is stored in a personal feature information data file (hereinafter simply referred to as a reference data file) 5. The above operation is
The processing is repeated for a plurality of individuals, and stored in the individual characteristic information and the reference data file 5.

When the registration of the characteristic information amount in the reference data file 5 is completed in this way, the operation of collating the fingerprint of the individual is started. In the case of collation, when a finger (a finger previously used for registration, for example, an index finger) is pressed against the fingerprint sensor 1, a fingerprint pattern is detected in the same manner as at the time of registration, and an A / D converter (not shown) in the fingerprint sensor is used. ) To convert to digital data. The converted digital data is converted by the following binarization circuit 2.
The data is converted into binary data of 0 and 1 and stored in the frame memory 3.

The characteristic information extraction / collation circuit 4 extracts characteristic information from the collation image data stored in the frame memory 3, reads out the characteristic information for each individual stored in the reference data file 5, and performs both collation. When the number of matches of the feature information is equal to or greater than a predetermined number, it is determined that the fingerprints match.

[Problem to be Solved by the Invention] As described above, the conventional fingerprint collation system generally has one
Personal characteristic information data (dictionary data) is created from the imprinted fingerprints. By the way, in the case of stamping, the reproducibility of the pattern itself is good, and only the rotation problem occurs. On the other hand, the degree of force applied when the finger is pressed against the fingerprint sensor 1, the position of the impression, and the like usually differ slightly each time the impression is made. For this reason, there is a problem that even the person himself may be rejected due to a difference in pressing between registration and comparison.

FIG. 6 is a diagram showing an example of a read fingerprint pattern.
(A) shows the same pattern at the time of registration, and (b) shows the same pattern at the time of collation. The pattern shown in (a) is relatively clearly read, whereas the pattern shown in (b) is thinner than that shown in (a), and ridges are cut off at some places. In such a case, the matching system may reject the two patterns without considering them as identical.

The present invention has been made in view of such a problem, and an object of the present invention is to provide a fingerprint collation method capable of performing fingerprint collation with high reliability.

[Means for Solving the Problems] FIG. 1 is a flowchart showing the principle of the first invention method, and FIG. 2 is a flowchart showing the principle of the second invention method.

According to a first aspect of the present invention, a fingerprint to be read out while a finger is pressed against a fingerprint sensor is read out as a plurality of frame images, and coincidence points among a plurality of frame images are extracted from the fingerprint feature information extracted for each of the frame images. The characteristic information of the fingerprint is stored in advance in the reference data file as the characteristic amount of the fingerprint in the descending order (step 1), and the fingerprint is read as a plurality of frame images while the finger to be collated is pressed against the fingerprint sensor. Comparing each of the read frame images with the feature amount of the fingerprint read from the reference data file (step 2), and determining whether or not the coincidence point with the feature amount in at least one frame image is equal to or greater than a predetermined threshold, or If the coincidence point with the feature amount is equal to or greater than the predetermined threshold value over a plurality of frame images, the fingerprint read out from the reference data file is used as the corresponding point. Determines that the fingerprint was collated target finger match is characterized in that the (Step 3) As.

In the second invention, when creating and storing the reference data file,
The finger is pressed against the fingerprint sensor, the fingerprint is read as one frame image, and the characteristic information of the fingerprint extracted from the frame image is stored in advance in the reference data file as the characteristic amount of the fingerprint (step 1). While reading a plurality of frame images of the fingerprint while pressing the finger on the fingerprint sensor, each of the read frame images is compared with the feature amount of the fingerprint read from the reference data file (step 2). If the matching point with the feature value is equal to or greater than a predetermined threshold value in one frame image, or if the matching value with the feature value is equal to or greater than a predetermined threshold value over a predetermined number or more of frame images, the reference data file is used. It is characterized in that it is determined that the read fingerprint matches the fingerprint of the finger to be verified (step 3).

[Operation] The feature information for reference data or the feature information for collation is extracted from a plurality of frame images, and, for example, ten pieces having the highest number of matching points are used as the reference data or collation data. By using such a method, the reliability of the feature information is improved, and fingerprint collation can be performed with high reliability.

[Example] Hereinafter, an example of the present invention will be described in detail with reference to the drawings.

FIG. 3 is a block diagram showing the configuration of an embodiment of a fingerprint collation system for implementing the method of the present invention. 4 are denoted by the same reference numerals. In the figure, reference numeral 1 denotes a fingerprint sensor which presses a finger on the surface to read fingerprint pattern information and converts it into digital data. The fingerprint sensor 1 reads the image information and converts it into an electric signal.
It comprises a photoelectric conversion element such as a CCD, an amplifier for amplifying the output of the photoelectric conversion element, and an A / D converter for converting the output of the amplifier into digital data.

2 indicates that the output of the fingerprint sensor 1 is 0 or 1 using a predetermined threshold.
This is a binarization circuit for converting into a binary. 10 is a binarization circuit 2
And a plurality of # 1 to #n frame memories for storing binarized data for each frame. Normal CRT
In the case of a screen, since one screen changes every 1/30 second, the value of n may be set to a value of 30 or less. Reference numeral 4 denotes a feature information extraction / matching circuit for sequentially reading out the binary data stored in the frame memory 10 to extract feature information and performing a comparison operation between a reference pattern and a matching pattern at the time of matching. Reference numeral 5 denotes a personal characteristic information data file (referred to as a reference data file) that stores reference characteristic information output from the characteristic information extraction / collation circuit 4. The method of the present invention will be described in detail using the circuit configured as described above.

(1) At the time of registration When the reference data is registered as a dictionary in the reference data file 5, it is as follows. First, the registrant's finger is pressed against the fingerprint sensor 1. The pressure applied to the fingerprint sensor 1 at the time of the pressing contact is weak at first, gradually increases, and finally becomes a constant pressure. In such a series of operations, the fingerprint sensor
Scan fingerprint images at 30 second intervals. The read image information is converted into digital data by an internal A / D converter. The converted digital data enters the subsequent binarization circuit 2 and
The data is converted into coded data and stored in the # 1 frame memory 10.

Next, the image of the next frame is read with the state of the finger as it is, and similarly stored in the # 2 frame memory 10 in the same manner. In this way, the image data is read at 1/30 second intervals while the finger is pressed against the fingerprint sensor 1, binarized, and sequentially stored in the frame memory 10.

In this case, the binarized data is stored in the frame memory 10 for each frame memory while the finger is pressed against the fingerprint sensor 1, but the finger is not moved.
It does not mean that it is touching, but sweating or pushing force changes. Therefore, although the positions where the pattern data appear are almost the same, the same data cannot be stored in the frame memory 10.

The characteristic information extraction / comparison circuit 4 reads the binarized data stored in the frame memory 10 in order from # 1, checks the position of the characteristic information as shown in FIG. Store in memory. Next, the second # 2 frame memory 10 is read and the feature information is stored in the memory. In this way, the feature information on all the frame memories 10 is stored in the memories.

When the feature information of all the frame memories 10 is stored in the memory, the feature information extracting / collating circuit 4 searches the contents of the memory to find out how many pieces of the same feature information exist at the same position. Then, they are arranged in descending order of the number. After arranging, for example, 10 pieces (features may be arbitrarily set) are read out from the feature information having a large number of matching points and stored in the reference data file 5.

(2) At the time of collation At the time of collation, the procedure from reading the image information by the fingerprint sensor 1 to storing it as binary data in the frame memory 10 is the same. Next, feature information extraction /
The matching circuit 4 sequentially reads out from the # 1 frame memory 10 in the same manner as at the time of registration, extracts feature information, reads out feature information from the reference data file 5, and reads out the read reference feature information and the matching feature. The information is compared sequentially. The result of the comparison is stored in a built-in memory. Next, the collation data is read from the # 2 frame memory 10, and the characteristic information is extracted by the characteristic information extraction / collation circuit 4. As the extracted characteristic information for collation, the characteristic information is read from the reference data file 5 and the read characteristic information of the reference and the characteristic information for collation are sequentially compared. The result of the comparison is stored in a built-in memory.

In this way, when the point of coincidence with the feature value is equal to or more than the predetermined threshold value in at least one frame image, or when the coincidence point with the feature value is equal to or more than the predetermined threshold value over a predetermined number or more of frame images. If it is determined that the fingerprints match. According to the present invention, since the feature information is extracted from the plurality of frame memories and compared with each other for both the reference data and the collation data, the difference in the pressing force when the finger is pressed against the fingerprint sensor is obtained. As a result, the possibility of making a wrong decision is greatly reduced, and the reliability at the time of fingerprint collation is improved.

In the above description, the case where characteristic information is extracted from a plurality of frame images for both the reference data and the matching data has been described. This method is optimal in terms of reliability, but the procedure is too complicated.

Therefore, when registering the reference data, fingerprint feature information is extracted from one frame image by a careful procedure and registered in the reference data file, and when reading a fingerprint for collation, the feature information is extracted from a plurality of frame images. May be extracted and compared with the reference data. Third system
The configuration shown in the figure can be used as it is. By doing so, it is not necessary to pay much attention to the pressing of the finger on the fingerprint sensor at the time of collation, and the reliability of fingerprint collation does not decrease so much.

[Effects of the Invention] As described above in detail, according to the present invention, a plurality of fingerprint images are stored in the frame memory at the time of reference data registration or collation, and the binary data read from the plurality of frame memories is stored. By extracting the feature information from the digitized data, fingerprint collation can be performed with high reliability, and the practical effect is great.

[Brief description of the drawings]

FIG. 1 is a flowchart showing the principle of the first invention method, FIG. 2 is a flowchart showing the principle of the second invention method, and FIG. 3 is a configuration showing one embodiment of a fingerprint collation system for implementing the method of the present invention. FIG. 4 is a block diagram, FIG. 4 is a diagram showing an example of characteristic information of a fingerprint, and FIG. 6 is a diagram showing an example of a read fingerprint pattern. In FIG. 3, 1 is a fingerprint sensor, 2 is a binarization circuit, 4 is a feature information extraction / collation circuit, 5 is a reference data file, and 10 is a frame memory.

Claims (2)

(57) [Claims] 1. A fingerprint read out while a finger is pressed against a fingerprint sensor is read out as a plurality of frame images, and a coincidence point is detected over a plurality of frame images among the characteristic information of the fingerprint extracted for each of the frame images. The feature information of the fingerprint is stored in the reference data file in advance as the feature amount of the fingerprint in descending order (step 1), and the fingerprint is read as a plurality of frame images while the finger to be collated is pressed against the fingerprint sensor. For each of the read frame images, a comparison is made with the feature amount of the fingerprint read from the reference data file (step 2). If the point of coincidence with the feature value is equal to or greater than a predetermined threshold value over the above frame image, the fingerprint read from the reference data file is used as the corresponding point. Fingerprint verification method characterized by fingerprints was collated target finger is the the match determines (Step 3) As. 2. When a reference data file is created and stored, a finger is pressed against a fingerprint sensor, a fingerprint is read as one frame image, and feature information of the fingerprint extracted from the frame image is used as a feature amount of the fingerprint in the reference data file. (Step 1). While reading the fingerprint as a plurality of frame images while pressing the finger to be verified against the fingerprint sensor, the fingerprint of the fingerprint read from the reference data file is read for each of the read frame images. Comparing with the feature amount (step 2), and when the coincidence point with the feature amount in at least one frame image is equal to or more than a predetermined threshold value, or when the number of frame images exceeds a predetermined number, the coincidence point with the feature amount is obtained. Is greater than or equal to the predetermined threshold, it is determined that the fingerprint read from the reference data file matches the fingerprint of the finger to be compared. Step 3) as the fingerprint matching method is characterized in that the.