JP2976362B2 - Dictionary registration matching method for fingerprint recognition - Google Patents

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a dictionary registration / collation method for fingerprint recognition using a plurality of dictionaries. In a fingerprint recognition apparatus, an input fingerprint image is compared with a fingerprint image registered in a dictionary, and when a predetermined number of feature points or more match, matching is confirmed and the identity is confirmed.

[0002]

2. Description of the Related Art As shown in FIG. 3, for example, a fingerprint recognition apparatus includes a core unit 10 and a control unit 20. The control unit 20 transmits a command such as fingerprint collation to the core unit 10, and transmits the command to the core unit 10. Performs a process corresponding to this command, and returns a response to the control unit 20.

[0003] The core unit 10 includes a central processing unit (CPU) 11 for controlling each unit, and a sensor 12 for capturing an imprint fingerprint.
A frame memory 13 for storing a multivalued fingerprint image captured by the sensor 12 for one screen, an image processing circuit 14 for performing image processing such as a binarization process on the multivalued fingerprint image, and holding the binary fingerprint image. Extracted from binary memory 15 and binary fingerprint image
Dictionary memory 1 for storing extracted feature points as dictionary images
6, a matching memory 17 used at the time of matching, a bus 18 for connecting each unit, and a control interface 19 for performing transmission / reception processing with the control unit 20.

[0004] As the dictionary registration process using such a fingerprint recognition device and the collation process with an input fingerprint, for example, a process according to a flowchart shown in FIG. 4 is known. That is, (a) of the figure shows a registration process, and (b) shows a collation process.

[0005] The registration process (a) includes the steps (1) to (7), and each of the steps (1) to (7) is performed as follows. (1) A fingerprint image obtained by capturing the imprint fingerprint by the sensor 12 is defined as a first image. That is, the imprint fingerprint image captured by the sensor 12 is stored in the frame memory 13 and the image processing circuit 1
4 and stores the result in the binary memory 15. (2) A feature point is extracted from the binary fingerprint image stored in the binary memory 15 and temporarily registered in the matching memory 17 as a dictionary.

(3) Next, fingerprint printing is performed again, and the sensor 12
Is taken as a second image. That is, the captured fingerprint image is held in the frame memory 13, binarized by the image processing circuit 14, and stored in the binary memory 15.

(4) Second data stored in the binary memory 15
The image is collated with the first image provisionally registered in the collation memory 17. That is, it is determined whether or not the number of matches of each feature point is equal to or greater than a predetermined number.

(5) As a result of the collation, if the collation is successful due to the coincidence of a certain number of feature points or more, the feature points of the first image temporarily registered in the collation memory 17 are permanently registered in the dictionary memory 16. I do.

[0009] (6). The main registration of the dictionary for the first image is performed, and the registration is successful, and the processing is terminated. (7) If the verification fails, the registration is failed and the processing up to that point is cleared.

In the above-mentioned processing, the feature points of the first image are temporarily registered in the dictionary memory 16 and the collation memory 17, and when the collation is passed by the collation processing in step (4), the dictionary memory is checked. A dictionary containing the feature points of the 16 first images is
It is also possible to change from temporary registration to full registration.

The collation processing (b) for confirming the identity is as follows:
Steps (1) to (4) are shown. (1) An image is captured by the sensor 12 by fingerprint imprinting and a collation image is input. Also in this case, the captured fingerprint image is stored in the frame memory 13, and the binary fingerprint image binarized by the image processing circuit 14 is stored in the binary memory 15.

(2) In response to a command from the control unit 20,
The dictionary of the designated dictionary number is transferred from the dictionary memory 16 to the matching memory 17. In this state, the central processing unit 1
By executing the first collation processing program, collation between the input fingerprint image and the dictionary is performed.

(3) If the collation results in a pass,
The control unit 20 is notified via the control interface 19 that the verification is successful. (4) If the collation fails as a result of the collation, the controller 20 is notified via the control interface 19 to the effect that the collation has failed.

As described above, the conventional fingerprint registration dictionary registration / collation method collates the input first and second fingerprint images, registers the first image as a dictionary only when the fingerprint image passes, and at the time of collation, The collation is performed using this dictionary.

[0015]

According to the conventional fingerprint registration dictionary registration / collation method, at the time of registration, the result of collation between the temporary registration dictionary based on the first input fingerprint image and the fingerprint image input immediately after the registration is passed. In this case, it is registered as a dictionary.

However, the fingerprint image picked up by the sensor 12 differs depending on the printing conditions such as the printing position and the printing pressure. For this reason, there has been a problem that when the registered dictionary is compared with the input fingerprint image, the possibility of rejection is increased. An object of the present invention is to register a plurality of dictionaries at the time of registration of a dictionary for fingerprint recognition and to collate with the plurality of dictionaries at the time of collation to prevent a reduction in the pass rate of collation.

[0017]

SUMMARY OF THE INVENTION A fingerprint registration dictionary registration / collation method according to the present invention is a method for binarizing an input fingerprint image and holding the binarized image.
A value memory 1, a feature point extracting means 2 for extracting feature points from the binary fingerprint image held in the binary memory 1, and a dictionary including the feature points extracted by the feature point extracting means 2 are registered. A dictionary memory 3, a matching memory 4 for holding feature points at the time of matching, and matching means 5 for matching the contents of the matching memory 4 with respect to an input fingerprint image at the time of matching.
A plurality of fingerprint images for the same finger are input and collated with each other, a plurality of dictionaries including feature points of the fingerprint images that have passed the collation are registered in the dictionary memory 3, and the plurality of dictionaries for the input fingerprint image are compared with the plurality of dictionaries. When the collation with any one of the plurality of dictionaries is successfully performed by the collation processing, it is determined that the collation is successful.

[0018]

[Function] A multi-valued fingerprint image obtained by imaging the first imprinted fingerprint is converted into a two-valued fingerprint image.
The binary image is stored in a binary memory 1, and feature points such as branch points and end points are extracted by a feature point extracting means 2, temporarily registered in a matching memory 4, and a second imprint fingerprint is imaged. The obtained multi-valued fingerprint image is binarized, and the binary image is stored in a binary memory 1.
, And the feature points are extracted by the feature point extracting means 2,
The collation is performed by the collation means 5, and if the collation is successful, a dictionary including feature points obtained by the first and second fingerprint imprints is stored in the dictionary memory 3. In this case, the first and second dictionaries are registered, but they can be registered as arbitrary plural dictionaries. Then, at the time of matching, if any one of the plurality of dictionaries passes the matching, it is determined that the matching is successful.

[0019]

FIG. 1 shows the basic configuration of the present invention. When corresponding to the configuration shown in FIG.
5, the feature point extracting means 2 corresponds to the image processing circuit 14 and the central processing unit 11 executing the feature point extracting program, the dictionary memory 3 corresponds to the dictionary memory 16, and the matching memory 4 corresponds to the matching memory. Corresponding to No. 17, the collation means 5 corresponds to the central processing unit 11 which executes the collation processing program.

FIG. 2 shows an embodiment of the present invention, in which (a)
Indicates a registration process consisting of steps, and (b) shows a matching process consisting of steps. First, the registration process (b)
Will be described.

[0021] The first image is input by the first fingerprint imprint. That is, the imprint fingerprint image is captured by the sensor, and the binarized binary fingerprint image is stored in the binary memory 1. . Feature points are extracted from the binary fingerprint image stored in the binary memory 1 by the feature point extracting means 2 and temporarily registered in the matching memory 4 as a dictionary of the first image.

[0022] A second image is input by a second fingerprint imprint. The binary fingerprint image in this case is stored in the binary memory 1. . The second image is collated with the dictionary of the first image temporarily registered in the collation memory 4.

[0023] If the collation result between the first image and the second image indicates that the collation is successful, the dictionary based on the first image is registered as the dictionary 1 in the dictionary memory 3. . The dictionary based on the second image is also used as the dictionary 2 in the dictionary memory 3.
Register for official registration. . The registration is successful by registering the dictionary 1 and the dictionary 2, and the process is terminated. . If the verification result fails, the registration is unsuccessful.
Clear the previous process. As described above, the present invention is not limited to the first and second fingerprint imprints,
A plurality of dictionaries can be registered in the dictionary memory 3 by repeating fingerprint imprinting three times or more and repeating collation.

Next, the collation processing (b) will be described. . An image for collation is input by fingerprinting. That is, the imprint fingerprint is imaged by the sensor, binarized, and stored in the binary memory 1. . Then, it is first checked against the dictionary 1. That is, the dictionary 1 read from the dictionary memory 3 is set in the matching memory 4, and the matching unit 5 performs matching with the matching image.

[0025] When the collation fails as a result of the collation, the collation with the dictionary 2 is performed. That is, the dictionary 2 read from the dictionary memory 3 is set in the matching memory 4, and the matching unit 5 performs matching with the matching image.

[0026] If the collation with the dictionary 1 or the collation with the dictionary 2 is successful, the collation is successful, and the control unit 2 as a higher-level device.
0 (see FIG. 3). . If the verification fails for any of the dictionaries 1 and 2, the control unit 20 is notified that the verification has failed.

The dictionary is obtained by extracting end points, branch points, and the like of a fingerprint as feature points. If a match of a predetermined number or more of feature points with an input fingerprint image is detected, the dictionary is determined to pass the matching. However, even with the same finger of the same person, the fingerprint image differs due to a slight displacement of the position on the impression surface, a difference in the pressing force at the time of impression, and the like. That is, the dictionary 1 based on the first image and the dictionary 2 based on the second image do not always have the same dictionary contents.

However, since the first image and the second image pass each other, the dictionary 1 and the dictionary 2 function as a dictionary for the same finger of the same person within a certain allowable range. become. At the time of fingerprint collation, the dictionary 1
If any one of the dictionary and the dictionary 2 passes the verification, the verification is successful. As described above, the dictionary 1 and the dictionary 2
However, the present invention is not limited to these two types, and it is also possible to register as a larger number of dictionaries.

[0029]

As described above, according to the present invention, a plurality of dictionaries are registered on the basis of fingerprint images that have passed verification, and at the time of fingerprint verification, any one of the plurality of dictionaries passes verification. In this case, the matching is successful, and the plurality of dictionaries indicate fingerprint images within an allowable range, so that the matching rate of fingerprint matching can be improved.

[Brief description of the drawings]

FIG. 1 is a diagram showing a basic configuration of the present invention.

FIG. 2 is a diagram showing one embodiment of the present invention.

FIG. 3 is a diagram showing a configuration of a conventional fingerprint recognition apparatus to which the present invention is applied.

FIG. 4 is a diagram showing a flowchart of conventional registration processing and collation processing.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Binary memory 2 Feature point extraction means 3 Dictionary memory 4 Matching memory 5 Matching means

──────────────────────────────────────────────────続 き Continuing on the front page (72) Takaichi Narasaki 1-17-3 Sakado, Takatsu-ku, Kawasaki City, Kanagawa Prefecture Inside Fuji Denso Co., Ltd. (72) Takayuki Soga 1-1-17 Sakado, Takatsu-ku, Kawasaki City, Kanagawa Prefecture No. 3 Inside Fuji Denso Co., Ltd. (56) References JP-A-63-149777 (JP, A) JP-A-63-147285 (JP, A) JP-A-4-123276 (JP, A) JP-A-63 -149776 (JP, A) JP-A-2-178780 (JP, A) (58) Fields investigated (Int. Cl. ⁶ , DB name) G06T 7/00 G06T 7/60

Claims (1)

(57) [Claims] 1. A binary memory (1) for binarizing an input fingerprint image and storing the same, and a feature point extracting means for extracting a feature point from the binary fingerprint image stored in the binary memory (1). 2)
And the feature points extracted by the feature point extracting means (2).
A dictionary memory (3) for registering as a dictionary image; a matching memory (4) for storing feature points at the time of matching; and a matching means for matching the contents of the matching memory (4) with respect to the input fingerprint image at the time of matching. (5), a plurality of fingerprint images for the same finger are input and collated with each other, and a plurality of dictionaries including feature points of the fingerprint images that have passed the collation are registered in the dictionary memory (3); A method of comparing an input fingerprint image with any one of the plurality of dictionaries, the method comprising: determining that the matching is successful when the matching of the input fingerprint image with any one of the plurality of dictionaries is successful. .