JP3386733B2 - Image data collation method, image data collation device, and storage medium storing image data collation program - Google Patents

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a technique for collating image data, and more particularly to a technique for making it possible to maintain the precision of image collation even if the image data is not sufficiently clear. It is suitable for collating data.

[0002]

2. Description of the Related Art Fingerprint image data of a subject (hereinafter referred to as "registered fingerprint") registered in a database or the like and data of a subject fingerprint image acquired by an image sensor or the like this time (hereinafter referred to as "matching"). There is a fingerprint collation device that collates with a "fingerprint") to personally confirm the subject.

An example of a fingerprint matching method used in a conventional fingerprint matching device will be described with reference to the matching fingerprint example shown in FIG. 6, and the procedure is as follows. The collation fingerprint is acquired as the data of the multi-tone image regarding the shade as shown in FIG. 6A (hereinafter referred to as “multi-tone collation fingerprint”). A multi-gradation collation fingerprint is simply binarized without performing any secondary processing of data such as interpolation, and data of a binarized image as shown in FIG. Get a "fingerprint"). From the binarized collation fingerprint, feature points such as wrinkle branch points and end points that constitute the fingerprint are extracted. These features are
It is called a characteristic figure or Manusha. The extracted characteristic figure and the characteristic figure of the registered fingerprint are compared to determine their identity.

[0004]

In the collation fingerprint shown in FIG. 6, the difference between the "wrinkles" and "valleys" of the wrinkles is clear, and the characteristic figure can be extracted well. Satisfactory fingerprint matching can be performed.

However, in the case of an unclear multi-gradation collation fingerprint as shown in FIG. 7 (a), the above-mentioned method has a problem that the precision of fingerprint collation remarkably deteriorates.

2 for the multi-gradation collation fingerprint of FIG.
The digitized collation fingerprint is shown in FIG. FIG. 6B and FIG.
With reference to (b), it should be easily understood that sufficient accuracy cannot be obtained even if fingerprint matching is performed by the method described above with reference to FIG. 7.

The reason why the fingerprint image becomes unclear as shown in FIG. 7 is considered as follows. 1. The wrinkle “valley” on the subject's fingerprint is shallow. 2. Since the subject's skin is soft, the "wounds" of the wrinkles on the fingerprint will collapse.

The above causes are often due to individual differences in skin. The percentage of people with such fingerprints is on the order of a few percent, but this value requires about 0.1 percent as a failure rate of fingerprint matching such as rejecting the correct person or misidentifying another person as the person. Considering the current situation, it is a large value that cannot be ignored. As another cause, it is possible that the fingerprint is damaged due to injury or the like.

In view of the above problems, it is a problem to be solved by the present invention to provide a method of collating image data that can maintain the precision of collation even if the image data is not sufficiently clear. .

[0010]

In order to solve the above-mentioned problems, the image data forming the first image and the image data forming the second image are collated to determine the degree of similarity between the two images. In the image data collation method of the present invention for determining, a plurality of regions are defined in the first image, and for each of the plurality of regions, the maximum correlation region having the maximum correlation with the region is defined as the second image. The degree of similarity is determined based on the difference between the positional relationship between the plurality of areas and the positional relationship between the maximum correlation areas.

FIG. 1 is a diagram showing the principle configuration of the present invention. In the figure, the first image data 1a and the second image data 1b are collated to determine the degree of similarity between the images formed by the two, and the result is output as the determination result 2.

The first image data 1a is, for example, image data registered in advance in a database or the like, and corresponds to, for example, a registered fingerprint when applied to the example of the fingerprint collation device described above.

The second image data 1b is, for example, image data acquired by using an image sensor or the like, and corresponds to, for example, a collation fingerprint when applied to the example of the fingerprint collation device described above.

The first and second image data (1
The relationship between a, 1b) and the registered fingerprint and the collation fingerprint may be interchanged.

The definition unit 3 defines a plurality of areas in the first image composed of the first image data 1a.
The definition of these regions is arbitrary, and if they are not the same, they may include, for example, overlapping portions.

For each of the plurality of areas defined by the definition section 3, the extraction section 4 defines a maximum correlation area, which is an area having the maximum correlation with the area, with the second image data 1b. It is extracted from the second image.

The determination unit 5 checks the positional relationship between the plurality of regions defined by the definition unit 3 and the positional relationship between the maximum correlation regions extracted by the extraction unit 4, respectively, and determines the difference in the positional relationship between the two. Based on this, the degree of similarity between the first image data 1a and the second image data 1b is determined and the determination result 2 is output.

In the configuration of FIG. 1, the area defined by the definition unit 3 and the maximum correlation area extracted by the extraction unit 4 corresponding to the area have a maximum correlation literally. If this is applied to the example of the fingerprint collation device described above, even if the collated fingerprints are unclear as shown in FIG. 7, the images included in both areas have the same characteristic figure. Can be estimated.

Therefore, the determination unit 5 grasps the magnitude of the difference between the positional relationship between the plurality of regions defined by the definition unit 3 and the positional relationship between the maximum correlation regions extracted by the extraction unit 4, It is possible to determine the degree of similarity between the first image data 1a and the second image data 1b, and it is possible to maintain the matching accuracy even for image data of an image that is not sufficiently clear.

Although the first and second image data can be determined by the binarized image data, the multi-tone image data is preferable because it can be used as a material for determining the magnitude of the correlation. is there.

In addition to the above-described method of the present invention and the apparatus for performing the method of the present invention, a computer-readable storage medium storing a program for causing a computer to perform the above-described method of the present invention is also included in the present invention. . Examples of the storage medium here include storage devices such as a ROM and a hard disk device built in the computer, these storage devices as an external device attached to the computer, and other computers network-connected to the computer. The aforementioned storage device attached, or a flexible disk, MO (magneto-optical disk),
There are portable storage media such as CD-ROM and DVD-ROM.

Further, in FIG. 1, one reference region is defined in the first image before the defining unit 3 defines a plurality of regions in the first image, and the extracting unit 4 determines the reference region and the maximum region. The maximum correlation reference area having the correlation is extracted from the second image, and the definition of the plurality of areas by the definition unit 3 is based on the position of the reference area, and the extraction unit 4 defines the maximum correlation reference area. The extraction may be performed from the image in the range based on the maximum correlation reference area in the second image.

In this case, the reference area and the maximum correlation reference area literally have the maximum correlation. Therefore, with respect to the maximum correlation reference region extracted corresponding to the plurality of regions defined with the position of the reference region as the reference, the range of the second image to be extracted by the extraction unit 4 is based on the maximum correlation reference region. It can be limited to the range. Since the range of the image to be extracted is reduced, as a result, it is possible to reduce the processing load for the extraction of the maximum correlation region performed by the extraction unit 4.

In FIG. 1, the determination of the degree of similarity between the two images by the determination unit 5 is performed by the extraction unit 4 and the mutual distance between the reference points of each of the plurality of regions defined by the definition unit 3. Alternatively, the determination may be made based on the difference between the reference distance of each maximum correlation region and the mutual distance.

As described above, by grasping the difference in the positional relationship between the regions as the difference in the mutual distance between the reference points in the regions, the processing load for the determination performed by the determination unit 5 can be reduced.

Further, in FIG. 1, the extraction of the maximum correlation area by the extraction unit 4 is carried out for each of the plurality of areas defined by the definition unit 3 and the maximum area which is composed of pixels selected from the defined area. You may make it extract the area | region which has correlation.

The area formed by the pixels selected from the area defined by the definition unit 3 can be considered as an area formed by thinning out some of the pixels included in the defined area. Even if the extraction unit 4 extracts the maximum correlation region corresponding to such a region configured by thinning out some of the pixels, the determination unit 5 maintains the required determination accuracy. If it is possible, the processing load for extracting the maximum correlation area performed by the extraction unit 4 can be reduced, which is advantageous.

[0028]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings.

Here, a mode in which the present invention is implemented by the above-mentioned fingerprint collation device will be described.

FIG. 2 is a diagram showing the configuration of a fingerprint collation device for implementing the present invention. This device includes a CPU 11, a storage device 12, a RAM 13, an image reading device 14, a display unit 15, and an input unit 16, which are connected to each other via a bus 17.

The CPU 11 uses a RAM 13 as a work area in accordance with a control program stored in the storage device 12, and controls the entire operation of this fingerprint collation device (hereinafter referred to as "this device") by a central processing unit. Is. By executing the control program, the CPU 11 performs a fingerprint collation process, which will be described later, which is a process corresponding to the definition unit 3, the extraction unit 4, and the determination unit 5 illustrated in FIG.

The storage device 12 is composed of a ROM, a hard disk device, etc., and stores in advance the above-mentioned control program read by the CPU 11 immediately after the power supply of this device is turned on. In addition, the first image shown in FIG. The registered fingerprint of the subject corresponding to the data 1a is stored in advance.

The RAM 13 is a work memory used by the CPU 11 when executing the above-mentioned control program.

The image reading device 14 is a device for obtaining a collation fingerprint of the subject corresponding to the second image data 1b shown in FIG. 1, and is, for example, an image scanner device or a CCD (charge coupled device). It is an image sensor.

The display unit 15 is a display device such as a CRT or a liquid crystal display which displays the judgment result of collation corresponding to the judgment result 2 shown in FIG.

The input unit 16 is an input device such as a keyboard device for the user of the apparatus to instruct the CPU 11 to acquire a collated fingerprint and to start the fingerprint collation processing described later.

The configuration of this device is not so different from that of a standard computer, and therefore, it is possible to realize the function of this device by such a computer. To that end, for example, from the above-mentioned storage medium that stores a fingerprint collation processing program for causing a computer to perform processing corresponding to the fingerprint collation processing described later, a reading device included in the computer corresponding to the storage medium. It may be read by, for example, temporarily stored in the main memory, and then executed by the central processing unit provided in the computer.

Next, the control performed by the CPU 11 will be described.

FIG. 3 is a flow chart showing the processing contents of the fingerprint collation processing particularly related to the present invention among the control processing of the entire apparatus realized by the CPU 11 executing the above-mentioned control program. In the figure, the processing contents of steps S101 and S104 correspond to the processing performed by the definition unit 3 of FIG.
The processing contents of 02 to S103 and step S106 correspond to the extraction unit 4, and the processing of steps S108 to S109 correspond to the processing performed by the determination unit 5, respectively.

FIG. 4 shows an example of a fingerprint image used for explaining the fingerprint collation process. The figure (a) shows the subject's registered fingerprint stored in advance in the storage device 12, and the figure (b) shows the subject's collation fingerprint acquired by the image reading device 14. Both images in the figure are multi-tone images.

FIG. 5 is a memory map relating to the fingerprint collation process, showing the usage of the RAM 13 when the CPU 11 performs the fingerprint collation process.

The fingerprint collation process performed by the CPU 11 will be described below along with FIG. 3 and with reference to FIGS. 4 and 5.

When the instruction to start the fingerprint collation process is given to the input unit 16, first, as shown in FIG. 4A, a rectangular area K is defined in the image of the registered fingerprint (step S101). The size of the rectangular area K may be arbitrary as long as the required collation accuracy of fingerprint collation can be obtained.

Next, as shown in FIG. 4B, a rectangular area having the same shape and size as the rectangular area K is set on the image of the collation fingerprint, and the rectangular area is two-dimensionally pixel by pixel. It is moved, and the correlation coefficient between the rectangular area and the rectangular area K is calculated for each movement. As a result, the rectangular area at the position where the correlation coefficient becomes maximum is set as the rectangular area KR (step S
102). The calculation of the correlation coefficient will be described later.

Here, the coordinates serving as a reference for specifying the positions of the rectangular region K and the rectangular region KR on the image, for example, the coordinates of the vertex at the upper left end of both rectangles are acquired, and K (X0, Y) are respectively obtained.
0) and KR (XR0, YR0) shown in FIG.
13 are stored in predetermined areas (step S1)
03).

Next, as shown in FIG. 4A, rectangular areas P1 to P4 are defined in the image of the registered fingerprint so that each vertex of the rectangular area K is the center (step S104). The arrangement of the rectangular areas defined here is arbitrary, and the number of the rectangular areas is not limited to four, and may be arbitrary as long as the required collation accuracy of fingerprint collation can be obtained. Similarly, the size of the rectangular area defined here may be arbitrary.

Then, similarly to the extraction of the rectangular area KR, as shown in FIG. 4B, one of the rectangular areas P1 to P4 (hereinafter referred to as Pi (i = 1, 2, 3, 4) is representative. And show)
In, a rectangular area having the same size as the rectangular area Pi is set on the image of the collation fingerprint, and the rectangular area is set to 2 in pixel units.
It is moved dimensionally, and the correlation coefficient between the rectangular area and the rectangular area Pi is calculated each time it is moved. As a result, the rectangular area at the position where the maximum correlation coefficient is obtained is set to the rectangular area PR.
Let i be the coordinates that serve as a reference for identifying the positions of the rectangular region Pi and the rectangular region PRi on the image, for example, the coordinates of the vertex at the upper left end of both rectangles, and obtain Pi (Xi, Yi) and P, respectively.
Ri (XRi, YRi) are respectively stored in predetermined areas of the RAM 13 shown in FIG. 5 (steps S105 to S105).
S107).

Thereafter, the difference Δi between the relative distance between K and Pi and the relative distance between KR and PRi is calculated for all i (= 1, 2, 3, 4) based on the following equation ( Step S108).

[0049]

[Equation 1]

Then, it is determined whether or not all of the calculated Δi are within a predetermined value, and if all are within the predetermined value, it is determined that the registered fingerprint and the verification fingerprint match. On the other hand, if not, the registered fingerprint is determined. It is determined that it does not match the collated fingerprint, and the determination result is displayed on the display unit 15 (step S109). The predetermined value used here is, for example, a value at which Δi is actually calculated from fingerprint image data obtained from a plurality of people and a desired matching accuracy is obtained based on the distribution of the calculation results.

The above-mentioned processing is the fingerprint collation processing.
In the fingerprint matching process described above, K, KR, P
Although each of the areas 1 to P4 and PR1 to PR4 is a rectangular area, the shape of these areas is not limited to a rectangular shape and may be any shape. Also, K and KR, and P1 to P4 and PR1 to PR4
Although it is desirable that the sizes and shapes of the fingerprints be the same, even if there are some differences, they are acceptable as long as the required matching accuracy of fingerprint matching can be obtained.

In the collation determination, in addition to the method of evaluating Δi in the above-described processing, for example, the difference in shape or area between the figure formed with Pi as the vertex and the figure formed with PRi as the vertex. It is also possible to adopt various methods such as determination based on.

Next, the calculation of the correlation coefficient used in steps S102 and S106 of the fingerprint matching process shown in FIG. 3 will be described. Note that here, the calculation of the correlation coefficient between the rectangular area A and the rectangular area B will be described.

First, the pixels included in each of the rectangular area A and the rectangular area B are A (i, j) and B (m,
n). However, the total number of pixels included in each of the rectangular area A and the rectangular area B is equal. Further, signal intensities, which are multi-gradation values indicating light and shade for these pixels, are defined as X _ij and Y _mn , respectively.

When these signal intensities are generalized and expressed as Z _pq , the following equation is defined.

[0056]

[Equation 2]

In the above equation, N represents the total number of pixels included in the rectangular area. Further, in the above equation, Σ indicates that it is the sum total of all the pixels included in the rectangular area. That is, the above equation represents the average value of the signal intensities of the pixels included in the rectangular area.

Next, the following equation will be further defined.

[0059]

[Equation 3]

The above equation shows the root mean square value of the signal intensity for the pixels included in the rectangular area.

Here, the correlation coefficient C _AB between the rectangular area A and the rectangular area B can be calculated by the following equation expressed using the definition of the above equation.

[0062]

[Equation 4]

The correlation coefficient between regions is calculated using the above equation.

In the calculation of the correlation coefficient according to the above equation, instead of using the signal intensities of all the pixels in the rectangular area, for example, pixels arranged in a line in an arbitrary line in the rectangular area are calculated. Calculation using only pixels, using only pixels included in a part of the rectangular area, or using only pixels selected by arbitrarily thinning out the rectangular area. However, there is no problem as long as the required matching accuracy of fingerprint matching can be obtained. The use of such a calculation method is advantageous because the number of pixels for which the correlation coefficient is calculated is reduced and the calculation amount is reduced. Further, another method of calculating the correlation coefficient may be adopted for the above-mentioned fingerprint matching process.

[0065]

Since the present invention is configured as described above in detail, it provides an image data collation method that can maintain the precision of collation even for image data that is not sufficiently clear. There is an effect that can be done.

[Brief description of drawings]

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

FIG. 2 is a diagram showing a configuration of a fingerprint collation device embodying the present invention.

FIG. 3 is a flowchart showing processing contents of fingerprint collation processing.

FIG. 4 is a diagram showing an example of a fingerprint image used for explaining a fingerprint matching device.

FIG. 5 is a diagram showing a memory map of a RAM relating to fingerprint matching processing.

FIG. 6 is a diagram showing an example of a fingerprint image used in the description of the related art.

FIG. 7 is a diagram showing an example of an image of a fingerprint used for explaining a problem of the conventional technique.

[Explanation of symbols]

1a First image data 1b Second image data 2 Judgment result 3 definition section 4 extractor 5 Judgment section

─────────────────────────────────────────────────── --Continued from the front page (58) Fields surveyed (Int.Cl. ⁷ , DB name) G06T ^7/ 00-7/60

Claims (12)

(57) [Claims] 1. A picture data matching determining the degree of similarity of the multi-gradation image data and the multi-tone image data and the by collating both images constituting the second fingerprint image constituting the first fingerprint image A method comprising: defining a reference region in the first fingerprint image, the maximum correlation reference region having a maximum correlation with the reference region.
Area is extracted from the second fingerprint image, and in the first fingerprint image, the one reference area defined.
Defining a plurality of regions arranged so as to surround the region, and regarding each of the plurality of regions, the maximum correlation region having the maximum correlation with the region is defined in the second fingerprint image .
Extracting from an image of a range based on the maximum correlation reference region, and determining the degree of similarity based on a difference between the positional relationship between the plurality of regions and the positional relationship between the maximum correlation regions. Image data matching method. 2. The determination of the degree of similarity is based on a relative distance between a reference point of each of the plurality of regions and a reference point of the reference region, and a reference point of each of the maximum correlation regions and the maximum correlation. The image data matching method according to claim 1, wherein the determination is performed based on a difference between a relative distance between the reference area and a reference point. 3. The extraction of the maximum correlation region includes extracting, for each of the defined regions, a region having a maximum correlation with a region composed of pixels selected from the defined region. The image data collating method according to claim 1 or 2, which is characterized. 4. The determination of the degree of similarity is performed by a figure formed with a vertex serving as a reference point of each of the plurality of regions,
The image data matching method according to claim 1, wherein the determination is made based on a difference in area of a figure formed with a reference point of each of the maximum correlation regions as an apex. 5. An image collation device that collates multi-tone image data forming a first fingerprint image with multi-tone image data forming a second fingerprint image to determine the degree of similarity between the two. And a reference area defining one reference area in the first fingerprint image.
Area defining means and the reference area defined by the reference area defining means and the maximum phase
The maximum correlation reference area having a relationship with the second fingerprint image
Maximum correlation reference area extracting means for extracting from the first fingerprint image, and the reference area defining means defines the first fingerprint image.
Defining means for defining a plurality of areas arranged so as to surround one reference area ; and, for each of the plurality of areas, a maximum correlation area having a maximum correlation with the area, the second fingerprint in the image
Extraction means for extracting from the image of the range based on the maximum correlation reference region, based on the difference between the positional relationship between the plurality of regions and the positional relationship between the maximum correlation regions, the determination of the degree of similarity An image data collation device comprising: 6. The determination of the degree of similarity is performed by determining a relative distance between a reference point of each of the plurality of areas and a reference point of the reference area, a reference point of each of the maximum correlation areas, and the maximum correlation. The image data collation apparatus according to claim 5 , wherein the determination is performed based on a difference between a relative distance between the reference area and a reference point. 7. The extraction of the maximum correlation region includes extracting, for each of the defined regions, a region having a maximum correlation with a region composed of pixels selected from the defined region. The image data collation device according to claim 5 or 6 , characterized in that. 8. The similarity determining means includes a figure formed with a vertex serving as a reference point of each of the plurality of areas, and a figure formed with a vertex serving as a reference point of each of the maximum correlation areas. The image data collating apparatus according to claim 5 , wherein the determination is performed based on a difference in area of the image data. 9. When executed by a computer,
An image data collation program for making the computer collate the multi-tone image data forming the first fingerprint image and the multi-tone image data forming the second fingerprint image to determine the degree of similarity between the two is stored. A computer-readable storage medium, wherein one reference area is defined in the first fingerprint image, and a maximum correlation reference area having a maximum correlation with the reference area.
Area is extracted from the second image, and the first fingerprint image has one reference area defined as above.
Defining a plurality of regions with a predetermined arrangement so as to surround the region, and regarding each of the plurality of regions, the maximum correlation region having the maximum correlation with the region in the second fingerprint image.
Image data matching, which is extracted from an image in a range based on the maximum correlation reference region, and is made to determine the degree of similarity based on the difference between the positional relationship between the plurality of regions and the positional relationship between the maximum correlation regions. A storage medium that stores a program. 10. The determination of the degree of similarity is performed by determining a relative distance between a reference point of each of the plurality of areas and a reference point of the reference area, a reference point of each of the maximum correlation areas, and the maximum correlation. The storage medium storing the image data collation program according to claim 9 , wherein the determination is made based on a difference between a relative distance between the reference area and a reference point. 11. The extraction of the maximum correlation region is performed by extracting, for each of the defined regions, a region having a maximum correlation with a region composed of pixels selected from the defined region. Claim 9 or 1 characterized
A storage medium storing the image data matching program described in 0 . 12. The determination of the degree of similarity is made between a figure formed with a point serving as a reference of each of the plurality of areas as a vertex and a figure formed with a point serving as a reference of each of the maximum correlation areas as a vertex. The storage medium storing the image data matching program according to claim 9 , wherein the determination is performed based on a difference in area.