JPH02196380A - Fingerprint register device - Google Patents

Abstract

PURPOSE:To specify the feature of fingerprints with high accuracy by specifying a 1st window area with a center part including the decreased distortion of a fingerprint, i.e., a picture part including many minutiae. CONSTITUTION:A finger print picture is binarized by a binarizing means 2, and respective minutiae included in the center part of the binarized picture are extracted by an extracting means 3 as an extracted minutiae group. These minutiae are sorted by an initial sorting means 4 as an initial minutiae group in the order of positions closer to the center of the binarized picture. These sorted minutiae are sorted again by a resorting means 5 as a resorted minutiae group in the order of the larger number of minutiae. Then a deciding/registering means 6 decides a prescribed range of the picture center part as a 1st window area and other area is defined as a 2nd window area and they are registered. Thus the proper collation of fingerprints is always performed.

Description

DETAILED DESCRIPTION OF THE INVENTION (Industrial Application Field) The present invention relates to a fingerprint verification system for a keyless entry system, an entry control system for a computer room, etc. Concerning suitable fingerprint registration devices.

(Prior art) Conventionally, in this type of fingerprint matching system, prior to fingerprint matching, as shown in Japanese Patent Application Laid-open No. 6.3-78286, characteristic features are identified from the entire reference fingerprint image. There is a method in which a plurality of small regions are extracted and each is registered as a partial image.

(Problem to be Solved by the Invention) However, in such a configuration, the quality of each registered partial image deteriorates due to crushing, blurring, etc., and when minutiae, which is a characteristic of a fingerprint, is not clear, There is a problem that the fingerprint matching accuracy becomes low. Furthermore, if the number of minutiae for specifying each registered partial image is insufficient, there is also the problem that it is difficult to ensure appropriate matching accuracy.

SUMMARY OF THE INVENTION In order to cope with this problem, the present invention provides a fingerprint registration device that creates and registers a fingerprint registration image that allows fingerprint verification to be performed properly at all times.

(Means for solving the problem) In solving the problem, the structural features of the present invention are as follows:
As illustrated in FIG. 1, an optical reading means 1 optically reads a human fingerprint as a fingerprint image, a binarizing means 2 binarizes the fingerprint image into a binary image, and an extraction means 3 for extracting each minutia in the image central area of the digitized image to form an extracted minutia group; and an extraction means 3 for classifying each minutia in the extracted minutia group in order of proximity to the center of the binarized image to form an initial classified minutia group. The initial classification means 4 checks the number of minutiae within each predetermined range in the central area of the image centered on each minutiae of the initial classification minutiae group, and replays each minutiae of the initial classification minutiae group in descending order of the number. a reclassifying means 5 for classifying into a reclassified minutiae group, determining a predetermined range of the central area of the image centered on one minutiae of the reclassified minutiae group as a first window area, and overlapping with the first window area; and determination and registration means 6 for determining another predetermined range of the image central area as at least a second window area so as to reduce mutual overlap, and registering these first and second window areas. It's because I did it.

(Function) By configuring the present invention as described above, when a human fingerprint is optically read as a fingerprint image by the optical reading means 1, this fingerprint image is converted into a binary image by the binarizing means 2. Each minutia in the image central area of this binarized image is extracted as an extraction minutia group by the extraction means 3, and each minutia of this extraction minutia group is passed to the initial classification means in order of proximity to the center of the binarized image. 4, the number of minutiae within each predetermined range in the central area of the image centered on each minutiae of this initial classification minutiae group is checked by the reclassification means 5, and the number of minutiae is The minutiae in the initial classification minutiae group are reclassified by the reclassification means 5 as a reclassification minutiae group in descending order. After that, by the decision registration means 6,
A predetermined range of the image central area centered on one minutiae of the reclassified minutiae group is determined as a first window area, and another predetermined range of the image central area is determined as the first window area.
At least the second window area is determined so as to reduce overlap with the window area and with each other, and these first and second window areas are registered.

(Effect) Therefore, the first window area is specified in the central part of the fingerprint where distortion is less and includes more minutiae, and the second window area is specified in the image part where the fingerprint is less distorted and includes more minutiae.
After the window area, each image portion containing the largest number of minutiae is identified, which are substantially different from each other and also different from the first window area. As a result, even if the fingerprint image is of low quality, the first and second registration window areas become registered fingerprint information that can accurately identify the characteristics of the fingerprint. Further, by using such registered fingerprint information, fingerprint matching can always be performed with high accuracy even if the quality of the fingerprint image at the time of matching is low.

(Example) Hereinafter, one example of the present invention will be described with reference to the drawings.
The figure shows an example in which the present invention is applied to a fingerprint verification system applied to a door lock mechanism 10 provided on the door of a special room such as a computer room that requires entry control. This fingerprint verification system includes a registration button switch 20, which is operated during fingerprint registration to generate a registration operation signal. Verification button switch 3
0 is operated during fingerprint verification to generate a verification operation signal. The numeric keypad 40 is operated to generate a designation code signal when the operator designates his or her own registered fingerprint image information from among a plurality of registered fingerprint image information registered in the microcomputer 60, as will be described later. Note that the designation code that is the content of the designation code signal corresponds to the operator's own fingerprint, which is known only to the operator.

The image input device 50 optically reads the fingerprint of the operator's finger when the operator's finger is pressed against the reading screen of the built-in image sensor, and converts the reading result into a fingerprint representing fingerprint image information. Generated as an image signal. The microcomputer 60 is constantly supplied with power from the DC power supply B, and executes a computer program according to the chart 70 shown in FIG. During this execution, arithmetic processing necessary for registration and verification of the fingerprint of the operator's finger and drive control of the drive circuit 70 for the door lock mechanism 10 is performed.

Note that the above-mentioned computer program is stored in the ROM of the microcomputer 60 in advance.

In the present embodiment configured as described above, the microcomputer 60 executes a process in which both the registration button switch 20 and the verification button switch 30 are set in step 81 of the computer program started in step 80a according to the chart 70 in FIG. Based on the operation, the process of determining a non-registration request and a non-verification request is repeated. In such a situation,
- The fingerprint of the operator's finger is read by the image input device 50 while the operator presses the finger against the reading screen of the built-in image sensor, and the image input device 50 corresponds to the read fingerprint image. Fingerprint image information is generated as a fingerprint image signal.

When a registration operation signal is generated from the registration button switch 20 at this stage, the microcomputer 60
However, in step 81, it is determined that it is a registration request, and step 8
At step 2, a fingerprint image signal is received from the image input device 50, and at step 83, the contents of the signal are binarized to form a binarized fingerprint image.

In such a case, each point (i.e., "0") of each pixel of this binary fingerprint image corresponds to each ridge of the operator's fingerprint, while each (i.e., "1") pixel corresponds to each ridge of the operator's fingerprint. A pixel corresponds to each valley line of the operator's fingerprint. However, the valley line corresponds to a valley between a pair of adjacent ridge lines of the fingerprint.

Next, in step 84, the microcomputer 60 smoothes the binarized fingerprint image by removing noise, etc. to obtain a smoothed fingerprint image, and in step 85, the smoothed fingerprint image is thinned and subjected to minutiae extraction processing. At the same time, pseudo minutiae removal processing is performed. In such cases, pseudo-minutiae refers to broken ridges, hides,! ! It refers to the apparent minutia that occurs due to wear, etc. Thereafter, in step 86, the microcomputer 60 converts the central image area of the image after removing the pseudo minutiae (see FIG. 4), excluding the peripheral image area, into a fingerprint area A.
, and each minutiae existing in this fingerprint area A (hereinafter, each minutiae M (i=0.1.
...+n)). However, the number of pixels for specifying the fingerprint area A is, for example, 512×512.

In this way, when the process in step 86 is performed, the microcomputer 60 classifies each minutiae M+i in step 87a according to the distance from the center pixel of the smoothed fingerprint image, and classifies each minutiae M2i in step 87a.
Find (i = 0, 1,...,!). However, if the pixels of the fingerprint area A are represented by coordinates (x, y), the central pixel is specified by (x, y)=(255, 255). Next, the microcomputer 60 performs step 8.
71), the number of minutiae belonging to each range (Δx×Δy=128X128) centered on each minutiae M2i is checked, and each minutiae M
2 Reclassify i to create each minutiae M3i (i20, 1
．． ..., []) is found. In such a case, both steps 8
As a result of the classification processing in 7a and 87b, each minutiae M3i is composed of minutiae in the order closer to the center of the fingerprint area A. However, if the included minutiae are the same, the one closer to the center is given priority.

When the minutiae classification process is completed as described above, the microcomputer 60 assumes that each minutiae M3i corresponds to the first window area A1, and determines whether the image of the first window area A1 is blurred or not. , when there are not many pixel areas (1) centered on a single minutia M3k in each minutiae Mji
28Xi28) is determined as the first window area A1 and is registered in its RAM. However, in the minutiae M3k, the symbol indicates i in the range of o<k<n, and specifies the center of the first window area A.

Then, the microcomputer 60 performs step 88b.
, each minutiae M3 in the range k+1≦i≦n
Assume that i corresponds to the second window area A2,
When the image in the second window area A2 is not very blurred or blurred, the overlapping portion with the first window area A is reduced, so that a single minutiae M3 in each minutiae M3i! A pixel area centered at (for example, 32
X64) is determined as the second window area A2 and is registered in its RAM. However, in minutiae M3p, the code p
is k+1. Indicate i in the range < l < n and specify the center of the second window area A2 (see FIG. 4). Furthermore, the above-mentioned condition of reducing the overlapping portion with the first window area A1 means that the pixel area 64×64 centered on the minutiae M3+[ in the first window area A1 does not overlap with the second window area A2. corresponds to the case. In addition, the pixel area 64×64 corresponds to the range of the silk interlocking at the time of verification.

After that, the microcomputer 60 performs step 88c.
Assuming that each minutiae M and i in the range of j7+1≦j≦n corresponds to the third window area A3 (see Fig. 4), the image of the third window area A3 may be blurred or When the number of eyebrows is not large, each eyebrow 1-sha M is
The pixel area centered on the single minutiae M, m in 31 is determined as the third window area A3, and is registered in its RAM. However, in minutiae M and Bn, the code 1n is 1
．． Indicate i in the range p+1<m<n, and specify the center of the third window area A3. Further, the above-described condition of reducing the overlapping portion with the second window area A2 corresponds to the case where the pixel area 32X64 centered on the minutiae M3n in the second window area A2 and the third window area A3 do not overlap. do. Note that the first to third window areas are determined and registered for the remaining operator's fingerprints in substantially the same manner as described above.

After that, when the operator enters the special room,
If the same operator presses his finger on the reading screen of the built-in image sensor of the image input device 50 and operates the collation button switch 30 and the numeric keypad 40, the image input device 50
0 generates a fingerprint image signal as described above, the verification button switch 30 generates a verification operation signal, and the numeric keypad 40
generates a specified code signal. Then, based on the collation operation signal, the microcomputer 60 selects the Stellar 781
, it is determined that it is a verification request, and the computer program proceeds to a fingerprint verification routine 89. However, in this fingerprint verification routine 89, the microcomputer 6
Based on the specified code signal, the first to third window areas corresponding to the contents of this signal are read out from the RAM and compared with the binary image of the fingerprint image signal. If the verification results match, the door lock mechanism 10 is connected to the drive circuit 70.
is driven to unlock the door.

As explained above, the first to third window areas A + -A 2 and A 3 are determined and registered as described above in the calculation process through steps 86 to 88c, so that the first window area A1 is , the second and third window areas A2 and A3 are identified as the central part of the fingerprint area where the fingerprint is less distorted and include more minutiae, and the second and third window areas A2 and A3 have the minutiae next to the first window area A1. It is specified by image portions that are substantially different from each other and different from the first window area A.

Therefore, even if the binarized image is of low quality, each window area A, to A3 can identify the characteristic portion of a fingerprint with high accuracy. In addition, such a window area A, □~
By using A3, highly accurate fingerprint matching is always possible even if the quality of the binarized image at the time of matching is low.

In addition, in this example, the number of pixels in the first window area A is set to 128 x 128 because if the number of pixels is too large, distortion will increase and the matching recognition rate of the person will decrease; This is because if it is too high, the rate of misrecognition with other people will increase. In addition, the reason why each window area A1~Hetsu is made into a square shape is as follows.
This is because it is convenient for hardware and software.

[Brief explanation of the drawing]

FIG. 1 is a diagram corresponding to the claims, FIG. 2 is a block diagram showing an embodiment of the present invention, and FIG. 3 is a flow chart showing the operation of the microcomputer shown in FIG.
and FIG. 4 are diagrams illustrating binarization of a fingerprint image. Description of symbols 20... Registration button switch, 50... Image input device, 60... Microcomputer. 1st

Claims (1)

[Claims] an optical reading means for optically reading a human fingerprint as a fingerprint image; a binarizing means for binarizing the fingerprint image to produce a binarized image; and each minutiae in the central area of the binarized image. an extraction means for classifying each minutiae of the extracted minutiae group in order of proximity to the center of the binarized image to form an initial classification minutiae group; Reclassifying means for checking the number of minutiae within each predetermined range in the central area of the image centered on each minutiae, and reclassifying each minutiae of the initial classification minutiae group in descending order of the number to form a reclassified minutiae group; , a predetermined range of the image central area centered on one minutiae of the reclassified minutiae group is determined as a first window area, and the image central area is arranged so as to reduce overlap with the first window area and mutual overlap. 2. A fingerprint registration device, further comprising determining and registering means for determining another predetermined range as at least a second window area and registering the first and second window areas.