JP3009542B2 - Fingerprint collation processor - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention executes a fingerprint collation process by binarizing a grayscale image read by a fingerprint sensor and detecting characteristic points of the fingerprint from the binarized image. More particularly, the present invention relates to a fingerprint matching processing apparatus that enables a long island that causes a pseudo feature point to be quickly and reliably removed from a grayscale image.

With the spread of computers throughout society,
How to ensure safety is becoming an important issue. On the other hand, in the personal verification system using an ID card or personal identification number, which is widely used as a means for personal identification, many questions have been raised from the aspect of ensuring security. Against this background, in recent years, a personal verification system using fingerprints having two major characteristics, "everyone is unidentified" and "lifelong unchanged" has been proposed. In order to put this personal verification system using fingerprints into practical use, it is necessary to build a configuration that can realize accurate recognition.

[0003]

2. Description of the Related Art A fingerprint collation processing apparatus for realizing a personal collation system first operates in a registration mode to binarize a gray-scale image of a fingerprint to be registered read by a fingerprint sensor, and to convert the binarized image into a binary image. Of the fingerprint, such as branch points and end points, is extracted from the personal ID.
The information is registered in the dictionary in association with the information. When the feature point information of all fingerprints to be recognized is registered according to the registration mode, the operation is performed in the collation mode, and the user first inputs his / her own ID. The feature point information read from the dictionary is read from the dictionary. Further, the grayscale image of the fingerprint to be recognized read by the fingerprint sensor is binarized, and the binarized image is compared with the feature point information read from the dictionary. Processing will be performed so as to perform identity verification.

At the time of executing such a collation process, as shown in FIG. 9, an image density substantially equal to that of a peripheral ridge is formed in a valley of a fingerprint as shown in FIG. A portion called an island may exist, and when this island exists, a portion called a bridge connecting between the island and the ridge appears in the binarized image, as shown in FIG. From now on, if this island exists, a problem that the matching ability of the personal ID information is deteriorated due to the detection of the pseudo feature point will occur.

Conventionally, in order to remove the island from the grayscale image, a method of locally setting a threshold for binarization has been adopted. However, since the image density of the island is almost the same as that of the surrounding ridges, such a conventional technique cannot reliably remove the island, which results in a problem that the accuracy of the individual verification is not sufficient. There was a point.

[0006]

In order to solve such a problem, the present applicant has previously filed Japanese Patent Application No. 4-113.
No. 98 (Title of Invention: Fingerprint collation device) disclosed an invention for removing an island present in a grayscale image read by a fingerprint sensor. In the present invention, a configuration for detecting contour lines of a fingerprint grayscale image is adopted, and when the detected contour lines form a closed curve, a portion surrounded by the closed curve is specified as an island, and the island of the specified island is specified. By changing the image density of the part in the density direction of the valley line, processing is performed to remove the islands existing in the grayscale image.

Certainly, according to the present invention, small islands can be reliably removed from the grayscale image, but long islands may have contour lines that form closed curves, but tracking them takes time. There was a problem.

The present invention has been made in view of such circumstances, and binarizes a grayscale image read by a fingerprint sensor and detects feature points of the fingerprint from the binarized image. It is an object of the present invention to provide a new fingerprint collation processing device which can remove a long island which causes a pseudo feature point from a grayscale image at high speed and surely when adopting a configuration for executing fingerprint collation processing. .

[0009]

FIG. 1 shows the principle configuration of the present invention. In the figure, reference numeral 1 denotes a fingerprint matching processing device provided with the present invention, and 2 denotes a fingerprint sensor provided in the fingerprint matching processing device 1.

The fingerprint collation processor 1 includes a grayscale image storage unit 10, a binarization unit 11, and a binarized image storage unit 1.
2, thinning means 13, thinning image storage means 14,
Feature point extracting means 15, dictionary means 16, collating means 17
Ridge and valley detection means 18, density change means 19, ridge and valley storage means 20, ridge and valley determination means 21, and density change means 22.

The gray image storage means 10 stores a gray image of a fingerprint detected by the fingerprint sensor 2. The binarizing unit 11 binarizes the grayscale image stored in the grayscale image storage unit 10. The binarized image storage unit 12 stores the binarized image output from the binarization unit 11. Thinning means 13
Thins the binarized image stored in the binarized image storage means 12. The thinned image storage means 14 is
Is stored.

The feature point extracting means 15 specifies feature points of the thinned image stored in the thinned image storage means 14,
A binary image around the specified feature point is extracted as feature point information. The dictionary unit 16 manages the feature point information of the fingerprint to be registered extracted by the feature point extracting unit 15 in association with the corresponding personal ID information. The matching means 17
The identity of the fingerprint to be verified is verified by comparing the binary image indicated by the fingerprint to be recognized stored in the binary image storage means 12 with the feature point information registered in the dictionary means 16. And output the result.

The ridge / valley line detecting means 18 detects the ridge / valley line portion of the fingerprint from the connection of the density of the gray image stored in the gray image storage means 10. The density changing means 19
If the length of the detected ridge portion which is developed by the ridge / valley line detecting means 18 does not reach the specified length, the density of the gray image stored in the gray image storage means 10 for the valley line is calculated. Change to the density direction you have. Ridge / valley line storage means 20
Stores the ridge / valley line portion information detected by the ridge / valley line detection means 18. The ridge / valley line determining means 21 determines whether or not a valley portion exists between two adjacent ridge portions stored in the ridge / valley line storage means 20. Density changing means 22
When the ridge-valley line determining means 21 determines the absence of a valley line portion, the density of the gray image of the non-existent portion stored in the gray image storage means 10 is changed to the density direction of the valley line.

[0014]

First, the fingerprint matching processing apparatus 1 of the present invention operates in the registration mode, binarizes the grayscale image of the fingerprint to be registered read by the fingerprint sensor 2 and thins it,
Feature point information of the registered fingerprint is extracted from the thinned image, and the extracted feature point information is registered in the dictionary means 16 in association with personal ID information. Then, when the feature point information of all fingerprints to be collated is registered in accordance with the registration mode, the operation in the collation mode is performed to binarize the grayscale image of the fingerprint to be collated read by the fingerprint sensor 2,
Valued image and user ID read from dictionary means 16
By comparing with the registration feature point information indicated by, the identity verification process is performed.

At the time of executing this collation processing, the ridge / valley line detecting means 18 detects the ridge / valley line portion of the fingerprint from the connection of the density of the gray image indicated by the fingerprint. At this time, the density changing means 19 provided in the ridge and valley line detecting means 18 regards the detected ridge portion whose length does not reach the specified length as a pseudo ridge portion, and The density of the gray image of the portion is changed in the density direction of the valley line. By this change processing, a long island having the same form as the pseudo ridge portion is removed.

When the ridge / valley detecting means 18 detects a ridge / valley portion of the fingerprint, the ridge / valley determining means 21
Determines whether or not a valley portion exists between two detected adjacent ridge portions. Based on this determination result, the density changing unit 22 determines whether the ridge / valley line determination unit 21 has a valley line. When determining the absence of a line portion, it is considered that a pseudo ridge portion exists in the valley line portion, and the density of the grayscale image of the non-existent portion is changed to the density direction of the valley line. By this change processing, a long island having the same form as the pseudo ridge portion is removed.

As described above, in the fingerprint collation processing apparatus 1 of the present invention, long islands causing pseudo feature points can be removed from the grayscale image at high speed and surely, so that the fingerprint collation processing is executed with high accuracy. You can do it.

[0018]

DESCRIPTION OF THE PREFERRED EMBODIMENTS The present invention will be described below in detail with reference to embodiments. FIG. 2 illustrates a fingerprint matching processing flow executed by the fingerprint matching processing apparatus 1 of the present invention.

As shown in this processing flow, the fingerprint collation processing apparatus 1 first operates in the registration mode, and inputs a gray-scale image of the fingerprint to be registered read by the fingerprint sensor 2 in step 1. Then, in Step 2, the density of this grayscale image is changed to remove long islands.
Then, the density-changed grayscale image is binarized, then, in step 4, the binarized image is thinned, and in step 5, feature point information is extracted from the thinned image. In step 6, this extracted feature point information is
The information is registered in the dictionary means 16 in association with the information.

Then, operating in the collation mode, in step 1 a gray-scale image of the fingerprint to be collated read by the fingerprint sensor 2 is input, and then in step 2 this is used to remove long islands. The density of the grayscale image is changed. Subsequently, in step 3, the grayscale image whose density has been changed is binarized. Subsequently, in step 6, the feature point information of the registered fingerprint pointed to by the input user ID is input to the dictionary means. 16 and comparing the read feature point information with the binarized image obtained in the process of step 3, the fingerprint to be verified and the user I
The identity verification is performed by checking whether there is any contradiction with D.

FIG. 3 shows an example of a grayscale image of a fingerprint received from the fingerprint sensor 2. Hereinafter, for convenience of explanation, it is assumed that this grayscale image has a high density at the ridge portion of the fingerprint and a low density at the valley line portion.

The density change processing of the gray image performed in step 2 of the processing flow of FIG. 2 is a characteristic processing of the present invention performed to remove long islands. 4 and 5 show an example of a detailed processing flow of the density change processing. Next, the long island detection / removal processing according to the present invention will be described in detail in accordance with these processing flows.

When entering the density change processing of the grayscale image, the fingerprint collation processing apparatus 1 first, in step 20, as shown in the processing flow of FIG. One unselected start point is selected from. Next, in step 21, it is determined whether or not all the start points have been selected. If all the start points have been selected, the process is terminated. On the other hand, if the selection of all the start points has not been completed, the process proceeds to step 22, where the selected start point is used as a starting point, and as shown in FIG. According to the density determination values TH1 and TH2 as shown in FIG. 6B, when the start point is at the valley line position, the length to the ridge is obtained, and when the start point is at the ridge position, the valley is determined. The length to the line is obtained, and the scanning direction and the length having the longest connection among the obtained ridge / valley line portions are specified.

Subsequently, in step 23, when the portion specified in step 22 is a ridge portion, it is determined whether or not the length of the ridge portion is equal to or longer than a specified length. It is determined whether the obtained ridge portion is a pseudo ridge portion having a specified length or less or an original ridge portion having a specified length or more. If it is determined in step 23 that the ridge portion is an original ridge portion, and if the portion specified in step 22 is a valley portion, the process proceeds to step 24, where information on the detected ridge / valley line portion is obtained. Is registered in the ridge and valley line storage means 20 and the process returns to step 20. Step 2
When judging the pseudo ridge portion in the judgment of 3,
Proceeding to step 25, the density of that portion is changed to a smaller value, and then the process returns to step 20. According to the density change processing in step 25, the long island detected as the pseudo ridge portion is removed by being changed to the valley line portion.

By executing the processing flow of FIG. 4 in this manner, as shown in FIG. 7, a scanning line (solid line in FIG. 7) for scanning over a ridge and a scanning line (solid line in FIG. 7) for scanning on a valley line. (Broken line in the middle) will be detected.

As shown in FIG. 7, it is not possible to completely remove the long island existing in the valley line portion only by executing the processing flow of FIG. Therefore, when the fingerprint matching processing device 1 detects the ridge / valley line portion by executing the processing flow of FIG. 4, the fingerprint matching processing device 1 then enters the execution of the processing flow of FIG. Enter the process to completely remove the island.

That is, as shown in the processing flow of FIG.
Select two adjacent ridges from zero. Next, in step 27, it is determined whether or not all the selections have been completed, and if all the selections have been completed, the process ends. On the other hand, if all selections have not been completed, step 2
Proceeding to 8, it is checked whether there is a valley portion between the two selected ridge portions by referring to the management data of the ridge / valley line storage means 20. That is, as shown in FIG. 8A, when a long island exists between two selected ridge portions, a valley line portion that should originally exist exists between the two ridge portions. Therefore, it is checked whether or not a valley line crossing a line orthogonal to the scanning direction of the selected ridge portion is registered in the ridge / valley line storage means 20. Check if there is a long island between the line parts.

If it is determined in step 28 that a valley line exists, this means that there is no long island, and the process returns to step 26 without executing any processing. On the other hand, if it is determined in step 28 that there is no valley line portion, the process proceeds to step 29, where the density of the portion is changed to a small value, and
Return to According to the density change processing in step 26, as shown in FIG. 8B, the long island existing between the two ridge portions is removed by being changed to the valley line portion.

As described above, the fingerprint collation processing device 1 of the present invention executes the processing flows of FIGS.
Long islands appearing in the grayscale image of the fingerprint are detected and removed.

Although the illustrated embodiment has been described, the present invention is not limited to this. For example, in the embodiment,
Although the configuration in which the identification process is performed by evaluating the similarity between the fingerprint image to be recognized and the registered fingerprint image indicated by the user ID is disclosed, the present invention is not limited to this. When evaluating the similarity between the fingerprint image to be recognized and all the registered fingerprint images, a configuration is performed in which the recognition process of the user ID indicated by the fingerprint image to be recognized is performed. Can be applied as is.

[0031]

As described above, according to the present invention,
The grayscale image read by the fingerprint sensor is binarized and
By detecting the characteristic points of the fingerprint from the binarized image,
When performing fingerprint matching processing, long islands that cause pseudo feature points can be quickly and reliably removed from the grayscale image, so that fingerprint matching processing can be performed quickly and with high accuracy. become able to.

[Brief description of the drawings]

FIG. 1 is a principle configuration diagram of the present invention.

FIG. 2 is a flowchart of a fingerprint collation process.

FIG. 3 is an example of a grayscale image of a fingerprint.

FIG. 4 is an example of a process flow of a density change process.

FIG. 5 is an example of a process flow of a density change process.

FIG. 6 is an explanatory diagram of a process according to the embodiment.

FIG. 7 is an explanatory diagram of a process according to the embodiment.

FIG. 8 is an explanatory diagram of a process according to the embodiment.

FIG. 9 is an explanatory diagram of an island appearing in a grayscale image.

FIG. 10 is an explanatory diagram of a bridge appearing in a binarized image.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Fingerprint collation processing apparatus 2 Fingerprint sensor 10 Grayscale image storage means 11 Binarization means 12 Binary image storage means 13 Thinning means 14 Thinning image storage means 15 Feature point extraction means 16 Dictionary means 17 Matching means 18 Ridge valley Line detecting means 19 Density changing means 20 Ridge valley line storage means 21 Ridge valley line determining means 22 Density changing means

Claims (2)

(57) [Claims] 1. A fingerprint matching process that performs a fingerprint matching process by binarizing a grayscale image read by a fingerprint sensor and detecting characteristic points of the fingerprint from the binary image. In the apparatus, detecting means (18) for detecting a ridge portion and a valley portion of the fingerprint from the connection of the density of the grayscale image, and a length within the ridge portion detected by the detecting means (18) is defined. The image processing apparatus further comprises a density changing unit (19) that regards an image that does not reach the length as a pseudo ridge portion and changes the density of the gray image of the portion in the density direction of the valley line. Fingerprint matching device. 2. A fingerprint matching process which performs a fingerprint matching process by binarizing a grayscale image read by a fingerprint sensor and detecting feature points of the fingerprint from the binary image. in the apparatus, to detect a ridge portion and a valley portion having a fingerprint from connection of density grayscale images Rutotomoni, of the detected ridge portion thereof
If the length does not reach the specified length, a pseudo ridge
Is considered to be a part, and the density of the grayscale
And detecting means for changing the concentration direction with the line (18), said detecting means (18) by determining means for determining whether valley portion exists between the detected two adjacent horns ridge portion (21) and a density changing means (22) for changing the density of the grayscale image of the non-existing part in the density direction of the valley line when the determining means (21) determines the absence of the valley line A fingerprint matching processing device comprising: