JP2006072758A - Fingerprint collating device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To solve the problem that although it is requisite that a fingerprint image is stably imaged like an oval gold coin, and that the finger is floating in a joint section in a finger collating device, the joint section is also adhered to a sensor, and the extraction of the joint is impossible in the case of such a fingerprint image that wrinkles are imaged, such a fingerprint image that the horizontal width of the fingerprint section is narrow and such a fingerprint image that the finger is obliquely placed. <P>SOLUTION: This fingerprint collating device is provided with an image inputting means for inputting a fingerprint image, a ridge direction extracting means for dividing the fingerprint image into predetermined regions, and for extracting the direction of the ridge and a joint extracting means for extracting the location of the joint from the extracted direction and way of arrangement of the ridge. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to a fingerprint collation apparatus that determines collation with known fingerprint information based on an input fingerprint image.

  The fingerprint feature extraction method installed in the system based on the assumption that a fingerprint image is acquired by a two-dimensional sensor is for the entire image, but if this is applied to a sweep type sensor as it is, nothing is captured on the fingertip. The unexplained region, the first joint of the finger, and the region below the target region are also subject to feature amount extraction, which has been a factor in reducing the matching rate of the same finger. In particular, since there are many unstable patterns such as wrinkles and joint patterns in the vicinity of the joint, it is essential to check only the region above the joint in order to improve accuracy.

  As a method of extracting the joint position with a conventional fingerprint verification device, after performing Sobel transformation on the input image, create an image with Gauss transformation, and calculate the square sum of the lateral density of the finger for each line. There is a method in which the position of the joint is obtained by creating the obtained power distribution graph, obtaining a fingertip from the obtained value, and further DP matching this with a standard power distribution value (see, for example, Patent Document 1).

Japanese Patent Laying-Open No. 2003-67749 (page 8-14, FIG. 9)

  In this conventional method, it is assumed that the fingerprint image is stably captured in an oval shape and the finger is floating at the joint portion, but cannot be applied to other fingerprint images. For example, in the case of a fingerprint image in which the joint part is in close contact with the sensor and wrinkles are imaged, a fingerprint image in which the width of the fingerprint part is narrow, or a fingerprint image in which the finger is placed diagonally, there is a problem that the joint cannot be extracted It was.

  The present invention divides a fingerprint image input from a sensor into small regions (blocks), obtains the direction angle of the fingerprint ridge for each block, and extracts the joint position from the direction of the direction angle. The purpose is to correctly extract joints for various fingerprint images such as fingerprint images with wrinkles imaged in close contact, fingerprint images with narrow width of fingerprint part, and fingerprint images with fingers placed diagonally It is.

  The fingerprint collation device according to the present invention includes an image input means for inputting a fingerprint image, a ridge direction extracting means for extracting the direction of the ridge by dividing the fingerprint image into predetermined regions, and the direction of the extracted ridge And joint extraction means for extracting the position of the joint from the arrangement.

  The fingerprint collation device according to the present invention includes an image input means for inputting a fingerprint image, a ridge direction extracting means for extracting the direction of the ridge by dividing the fingerprint image into predetermined regions, and the direction of the extracted ridge And a joint extracting means for extracting the position of the joint based on the way of arrangement, the joint can be accurately extracted.

Embodiment 1 FIG.
FIG. 1 is a diagram for explaining the first embodiment, and specifically shows a block configuration of a fingerprint collation apparatus. The configuration will be described below. In the figure, a line connecting the blocks means that information is exchanged. The image input device 1 that acquires a fingerprint image, specifically a fingerprint sensor, is a device that reconstructs a sweep type line sensor and outputs a two-dimensional image. The ridge direction extraction means 2 for extracting the direction of the ridge from the fingerprint image divides the entire image into small regions such as 16 × 16 pixels as a specific extraction method, and a filter having directionality in the region is used. Select the most number of directions by calling. The ridge direction extracting means 2 divides the fingerprint image into predetermined regions and extracts the ridge direction. Specifically, the background extraction means 3 for extracting the background from the fingerprint image determines that the block is the background when the directionality is difficult to determine uniquely from the result of the ridge direction extraction means 2 or when the directionality is weak. Can do. In addition, there is a method of determining the background when the variance value of the image density in the block is small.

  Core extracting means 4 for extracting the position of the core called the core of the flow of the fingerprint ridge, joint extracting means 5 for extracting the joint position from the arrangement of the ridge directions obtained by the ridge direction extracting means 2, and the fingerprint There is an image cutout means 6 for cutting out a region from the fingertip to the joint.

  The feature quantity extraction means 7 for extracting feature quantities from the image produced by the image cutout means 6 specifically extracts direction angle data and feature points at branch points and end points called ridge minutia. There is a technique. As a minutiae extraction method, a method of binarizing an image, thinning a ridge portion, and tracking is known. In addition, there are other features such as a method using the ridge image itself as a feature amount and a feature amount based on frequency analysis, but any method may be used.

  There is a feature quantity registration means 8 for registering the feature quantity obtained by the feature quantity extraction means 7 together with user information. The feature quantity storage means 9 for storing the feature quantity extracted by the feature quantity registration means 8, specifically, a memory or a hard disk is stored. There is a feature amount collating unit 10 that collates the feature amount obtained by the feature amount extracting unit 7 with the feature amount stored in the feature amount storing unit 9 in advance. There is a collation result determination unit 11 that determines the person / others based on the result of the feature amount collation unit 10.

  There are variations in the shape of fingerprints such as spirals and bows, but when you look at the ridge shape from the fingertip toward the first joint, the fingerprint has a gentle arch shape close to the horizontal. Since there are vortex and loop cores near the belly of the finger, there are many complicated ridges such as vertical and diagonal directions, and the shape becomes closer to the horizontal direction again as it approaches the joint. Therefore, in the present invention, the direction angle for each small region of the fingerprint is obtained, and the position of the joint is obtained by extracting the fingertip, the vicinity of the core, and the vicinity of the joint by looking at how the direction angles are arranged.

  FIG. 2 shows an example of the input image. FIG. 3 shows the result of direction angle extraction of the image of FIG. FIG. 4 shows an image cut out from the result of FIG. 3 according to the present invention.

  A specific method of joint extraction processing will be described below. Check the alignment of direction angles from the upper direction (fingertip) to the lower direction (finger base). First, the presence of the fingertip area is confirmed. Focusing on the top (fingertip) block of the finger, the average value, variance value, maximum value, and minimum value of the direction angle for one line in the horizontal direction are obtained. An example is shown in FIG. When the variance value and the maximum value / minimum value are smaller than the threshold value and the average value is within the horizontal direction ± threshold value, it is determined that the fingertip region has been extracted. At this time, the average value of the direction angles is recorded as the finger tilt angle.

  Further, when it is determined that the original image is tilted, it is possible to repeat the process of correcting the original image based on the tilt angle and obtaining the direction angle. Specifically, by providing fingertip extraction means for extracting the fingertip area, the inclination of the fingerprint image can be corrected with the average direction of the ridges of the fingertip as the inclination of the fingerprint image.

  Next, the core vicinity region is extracted. Find the number, width, and height of the area below the fingertip in the direction angle that is not horizontal. When the number, width, and height are each greater than the threshold value, it is determined that the core vicinity has been extracted. Thereby, the vicinity of the center of the fingerprint can be extracted from the direction and arrangement of the ridges. An example is shown in FIG. The intermediate color block is a non-horizontal block, and the white background block is a block close to the horizontal. Here, the horizontal means a direction perpendicular to the direction from the fingertip toward the base of the finger. Further, the core extracting means 4 can extract the vicinity of the center (core) of the fingerprint from the direction and arrangement of the ridges only in a region continuous to the fingertip region, and can efficiently extract the core. .

  Next, a joint vicinity region is extracted. The number of blocks whose direction angle is close to the inclination of the finger obtained when checking the presence of the fingertip in the block below the core and the width / height of the area are obtained. An example is shown in FIG. A block (blank) in which no line is drawn is a background block. When the width, height, and number of the areas exceed the threshold values, the area is determined as the joint vicinity area. A location where blocks close to the horizontal are arranged is determined as the joint vicinity region. Note that, depending on the image, the joint portion may not be included. Therefore, if the joint portion is not extracted, the long image may be cut out at the end of the input image.

  As described above, since the joint position is extracted based on the direction of each small area of the fingerprint ridge, the joint position can be correctly extracted even in a fingerprint image with a narrow finger width or a fingerprint image placed obliquely. Therefore, the region from the fingertip to the first joint including information effective for collation can be extracted, so that collation accuracy can be improved.

  Image input means for inputting a fingerprint image, ridge direction extraction means for extracting the direction of the ridge by dividing the fingerprint image into predetermined regions, and the position of the joint is extracted from the direction and arrangement of the extracted ridges Since the joint extraction means is provided, the joint can be cut out from the fingerprint images in various states by extracting the joint position.

  Further, since the entire image is divided into small areas and one method angle is obtained for each small area, the amount of data to be handled can be reduced and the calculation time can be shortened.

  Moreover, since the core extraction means for extracting the vicinity of the center of the fingerprint from the direction and arrangement of the ridges is provided, the core can be reliably extracted.

  Furthermore, since the joint extraction means extracts only the side opposite to the fingertip with respect to the vicinity of the center of the fingerprint, it extracts the joint region only on the lower side of the core region (the base of the finger). The joint can be determined accurately and at high speed.

  In addition, it includes a fingertip extraction means for extracting a fingertip region, and corrects the inclination of the fingerprint image with the average direction of the ridges of the fingertip as the inclination of the fingerprint image, so that the image of the finger can be detected and corrected. It is not necessary to perform alignment in consideration of rotation at the time of collation, thereby speeding up the process and improving collation accuracy.

  Further, since the core extraction means performs extraction only in a region continuous with the fingertip region, the core can be efficiently extracted.

  In addition, since fingerprint collation is performed on the region from the fingertip to the first joint, fingerprint collation can be performed efficiently.

It is a block block diagram of the fingerprint collation device by Embodiment 1 of this invention. It is explanatory drawing which shows the example of the input image by Embodiment 1 of this invention. It is explanatory drawing which shows the result of having extracted the direction angle of the image of FIG. 2 by Embodiment 1 of this invention. It is explanatory drawing which shows the image cut out by Embodiment 1 of this invention. It is explanatory drawing of the joint extraction process by Embodiment 1 of this invention. It is explanatory drawing of core vicinity area | region extraction by Embodiment 1 of this invention. It is explanatory drawing of the joint vicinity area | region extraction by Embodiment 1 of this invention.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Image input device, 2 Ridge direction extraction means, 3 Background extraction means, 4 Core extraction means, 5 Joint extraction means, 6 Image clipping means, 7 Feature quantity extraction means, 8 Feature quantity registration means, 9 Feature quantity storage means, 10 feature amount collating means, 11 collation result judging means.

Claims (6)

Image input means for inputting a fingerprint image, ridge direction extraction means for extracting the direction of the ridge by dividing the fingerprint image into predetermined regions, and the position of the joint from the direction and arrangement of the extracted ridges A fingerprint collation device comprising joint extraction means for extraction. 2. The fingerprint collation apparatus according to claim 1, further comprising core extraction means for extracting the vicinity of the center of the fingerprint from the direction and arrangement of the ridges. 3. The fingerprint collation device according to claim 2, wherein the joint extraction means extracts only the side opposite to the fingertip with respect to the vicinity of the center of the fingerprint. The fingerprint according to any one of claims 1 to 3, further comprising fingertip extraction means for extracting a fingertip region, wherein the inclination of the fingerprint image is corrected with the average direction of the ridges of the fingertip as the inclination of the fingerprint image. Verification device. 5. The fingerprint collation apparatus according to claim 1, wherein the core extraction means extracts only a region continuous with a fingertip region. The fingerprint collation apparatus according to any one of claims 1 to 5, wherein fingerprint collation is performed on an area from a fingertip to a first joint.

Images