JP2002334331A - Device and method for image collation, program, and recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To improve collation precision. SOLUTION: A converting means 10 repeatedly perform at least one of a translation process and a rotating process for an inspection image by a fixed quantity within a specific range. A collating means 30 compares the inspection image and a registered image within a plurality of collation areas each time the converting means 10 performs one process to find the consistency 403 in each collation area. A maximum consistency extracting means 40 finds the maximum consistency 404 by the collation areas from consistency 403. An arithmetic means 50 finds the mean maximum consistency 405 as the mean value of maximum consistency 404 by the collation areas. A decision means 60 decides that the inspection image is identical with the registered image when the mean maximum consistency 405 is larger than a previously set threshold.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention evaluates the degree of similarity between a pre-registered image and a newly input image in an image having periodicity such as a fingerprint, a nose print, an iris, and a texture pattern. The present invention relates to an image matching device, an image matching method, a program, and a recording medium for determining whether a newly input image is the same as a previously registered image.

[0002]

2. Description of the Related Art Conventionally, various image collating apparatuses have been known. For example, the document “Kobayashi,“ Fingerprint Matching Using Thinned Image Pattern Matching ”, IEICE Transactions (D-II), vo1.J79-D-II, no.3, pp.330-340, March1996.
In the fingerprint matching device as an example of the image matching device described in “1”, the images themselves are subjected to pattern matching to determine whether the two images are the same fingerprint image or different fingerprint images. FIG. 23 is a block diagram showing the configuration of a fingerprint matching device using such pattern matching. This fingerprint matching device includes an image input device 101, an image database 201, and a processing device 309.

An image input apparatus 101 detects the unevenness of a fingerprint of a finger placed on a sensor of the image input apparatus 101 by a sensor, and performs analog / digital conversion and 2 conversion on a signal output from the sensor.
Perform image processing such as binarization. The output of the image input device 101 is a binary image in which the convex portion of the fingerprint is represented by a pixel having black luminance (black pixel), and the concave portion of the fingerprint is represented by a pixel having white luminance (white pixel). The convex portion of the fingerprint may be a white pixel, and the concave portion of the fingerprint may be a black pixel. Here, the image output by the image input device 101 is called an inspection image.

[0004] The image database 201 stores fingerprint images obtained in advance as registration data. Here, the image stored in the image database 201 is called a registered image. The processing device 309 compares the inspection image output from the image input device 101 with the registered image output from the image database 201, and determines whether the two images are the same fingerprint image or different fingerprint images. Is determined. In order to improve the determination accuracy (collation accuracy), the processing device 309 includes the conversion unit 19, the collation unit 39, the maximum matching rate extraction unit 49, and the determination unit 69.

[0005] The conversion means 19 outputs an inspection image in which each pixel of the input inspection image is translated and rotated by a certain amount of change. The matching unit 39 compares the inspection image output from the conversion unit 19 with the image database 201.
In the registered image output from, the luminance value is compared for each pixel at the same position, the number of pixels having the same luminance value is totaled in a preset collation area, and the total number of matching pixels and the registered image are compared. The degree of similarity (matching rate) between the inspection image and the registered image is determined from the number of black pixels. Further, the matching means 39 sends the translation amount 418 to the conversion means 19 in order to repeat the translation and rotation by the conversion means 19 and the comparison and comparison by itself until the translation amount falls outside the preset range. Is output.

[0006] The maximum matching rate extracting means 49 includes a collating means 39.
Is the maximum value (maximum match rate) from the match rates 419 output by
420 is obtained and output. The judging means 69 compares the maximum matching rate 420 with a preset threshold value,
If the maximum matching rate 419 is equal to or larger than the threshold value, it is determined that the fingerprint images are the same, and if the maximum matching rate 419 is smaller than the threshold value, it is determined that the fingerprint images are different.

FIG. 24 is a flowchart showing the collating operation of the fingerprint collating apparatus shown in FIG. First, the image input device 1
01 detects a fingerprint of a finger placed on the sensor and generates an inspection image (step S51). Processing unit 309
When the inspection image is input from the image input device 101 (step S52), and the registered image is input from the image database 201 (step S53), the conversion unit 19 translates and rotates the inspection image (step S52).
54), the inspection image output from the conversion means 19 and the registered image output from the image database 201 are compared and collated by the collation means 39 to obtain a coincidence ratio 419 (step S).
55).

[0008] Then, the processing unit 309 determines the coincidence rate 419
Is determined by the maximum matching rate extracting means 49 (steps S56 and S57). Processing unit 3
In step 09, the parallel movement and the comparison and collation are repeated until the amount of parallel movement is out of the preset range (NO in step S58). Finally, the determination means 69 of the processing device 309 determines that the fingerprint image is the same when the maximum coincidence rate 420 is equal to or higher than the threshold value (YES in step S59), and when the maximum coincidence rate 420 is smaller than the threshold value. Is determined to be a different fingerprint image. Note that the processing of the conversion unit 19 may be executed for a registered image instead of the inspection image.

[0009]

FIG. 25 shows how the registered image and the inspection image are compared and collated by the fingerprint collating device shown in FIG. FIG. 25 shows the result of collation (identity verification) after correcting positional deviation between fingerprint images acquired from the same finger. As can be seen from FIG. 25, the registered image and the inspection image match overall, but a mismatch occurs in part. The reason why such partial inconsistency occurs despite personal verification is that the shape of the fingerprint is locally distorted when a fingerprint, which is a soft skin pattern, is detected by a contact-type sensor. In the fingerprint matching device of FIG. 23, the maximum matching rate used as a judgment index is 10 in the case of personal identification.
Ideally, it should be 0%. However, the actual maximum matching rate is significantly lower than 100% due to partial mismatch between the registered image and the inspection image due to local distortion of the fingerprint. For the above reasons, the difference in the maximum matching rate between the personal verification and the others verification becomes small. For this reason, in the conventional fingerprint matching device, it is difficult to set a threshold value for determining whether the fingerprint images are the same fingerprint image or different fingerprint images.
There is a problem that the matching accuracy is reduced. This problem may also occur when an image collation device other than the fingerprint collation device is used, particularly when the inspection target on which the inspection image is based is soft. The present invention has been made to solve the above-described problem, and has an image matching device, an image matching method, a program, and a recording medium that can improve matching accuracy even when local distortion exists in an inspection image. The purpose is to provide.

[0010]

An image collating apparatus according to the present invention repeatedly executes at least one of a translation process and a rotation process on a first image at predetermined intervals within a preset range. A first conversion means (10) for outputting a processed first image; and a first image output from the first conversion means for each processing of the first conversion means. The second image and the first image or the second image
The first matching means (30) for comparing and matching the plurality of images in the plurality of divided matching areas to obtain the matching rate (403) in each matching area, and the matching rate output from the first matching means And a maximum matching rate extracting means (40) for obtaining a maximum matching rate (404) for each of the matching areas, and a maximum matching rate which is an average value of the maximum matching rates for each matching area output from the maximum matching rate extracting means. Calculation means (5) for calculating the rate average (405)
0), and when the maximum matching rate average output from the calculating means is equal to or greater than a predetermined threshold value, the determining means (6) for determining that the first image and the second image are the same.
0). Further, the image matching device of the present invention includes a first converting means (10) and a first matching means (3
0), the maximum matching rate extracting means (40), and the minimum matching rate extracting means (4) for obtaining the minimum matching rate (406) for each matching area from the matching rates output from the first matching means.
1) and a difference between the maximum matching rate output from the maximum matching rate extracting means and the minimum matching rate output from the minimum matching rate extracting means is obtained for each of the matching regions, and a match which is an average value of the differences is obtained. Calculation means (5) for calculating the average rate difference (407)
1) and a judging means (6) for judging that the first image and the second image are the same when the average of the coincidence rate differences output from the arithmetic means is equal to or greater than a preset threshold value.
1). Further, the image matching device of the present invention includes a first converting means (10) and a first matching means (3
0), the maximum matching rate extracting means (40), the minimum matching rate extracting means (41), and the maximum matching rate output from the maximum matching rate extracting means to the minimum matching rate output from the minimum matching rate extracting means. The quotient divided by the ratio is obtained for each of the comparison areas, and the calculating means (5
(1) determining means for determining that the first image and the second image are the same when the quotient of coincidence rates output from the calculating means is equal to or greater than a preset threshold value;
Is provided. Further, the image matching device of the present invention includes a first converting means (10) and a first matching means (3
0), the maximum matching rate extracting means (40), the minimum matching rate extracting means (41), and the maximum matching rate average, which is the average value of the maximum matching rates for each collation region output from the maximum matching rate extracting means. (408) and a calculating means (52) for obtaining a minimum matching rate average (409) which is an average value of the minimum matching rates for each collation area output from the minimum matching rate extracting means, and an output from the calculating means. The maximum match rate average is equal to or greater than a preset first threshold value, and the minimum match rate average output from the arithmetic means is set to a preset second threshold value (first threshold value).
When the threshold value is smaller than the second threshold value, the determination unit (62) determines that the first image and the second image are the same. Further, the image matching device of the present invention outputs the first conversion means (10), the first matching means (30), the minimum matching rate extracting means (41), and the output from the minimum matching rate extracting means. A calculating means (53) for obtaining a minimum matching rate average (409) which is an average value of the minimum matching rates for each collation area; and a case where the minimum matching rate average output from the calculating means is smaller than a preset threshold value And determining means (63) for determining that the first image and the second image are the same.

In one embodiment of the image matching apparatus according to the present invention, a predetermined number of maximum matching rates are extracted in descending order from the maximum matching rates for each matching area output from the maximum matching rate extracting means. Selecting means (70) for determining the average of the maximum matching rates by using only the maximum matching rates extracted by the selecting means. One configuration example of the image matching device of the present invention extracts a predetermined number of maximum matching rates in descending order from the maximum matching rates for each matching area output from the maximum matching rate extracting means, The apparatus further comprises a selection means (70) for extracting a predetermined number of minimum matching rates in ascending order from the minimum matching rates for each matching area output from the minimum matching rate extracting means, and the arithmetic means includes: Using only the maximum match rate and the minimum match rate extracted by the selection means, the match rate difference average, the match rate quotient average, the maximum match rate average, and the minimum match rate average are obtained. Also, one configuration example of the image matching device of the present invention is a selection / selection method for extracting a predetermined number of minimum matching rates in ascending order from the minimum matching rates for each matching area output from the minimum matching rate extracting means. Means (70) are further provided, wherein the calculating means obtains the minimum matching rate average using only the minimum matching rate extracted by the selection means.

In one embodiment of the present invention, at least one of a translation process and a rotation process is performed on the first image at a first initial position in a predetermined range. A second conversion means (11) for repeatedly executing the processing every fixed amount within the range and outputting a processed first image;
For each processing of the second conversion means, the first image output from the second conversion means and the second image are compared and matched to obtain a matching rate by a second matching means (31 )When,
When the coincidence rate output from the second matching means is maximum, the first position from the first initial position to the current position is
Storage means (80) for storing the parallel movement amount, the rotation angle, or both the parallel movement amount and the rotation angle of the image, and wherein the first conversion means (12) stores the parallel movement amount stored in the storage means. A position obtained by adding the movement amount, the rotation angle or the parallel movement amount and the rotation angle to the first initial position is defined as a second initial position, and the first image is moved to the second initial position. At least one of a translation process and a rotation process is performed on the first image. In one configuration example of the image matching device of the present invention, a range preset in the first conversion unit (12) is smaller than a range preset in the second conversion unit (11). Things. In addition, one of the image matching apparatuses of the present invention
In the configuration example, the translation amount, the rotation angle, or the translation amount and the rotation angle per one time performed by the second conversion unit (11) on the first image may be determined by the first conversion unit ( 12) is to make the parallel movement amount, the rotation angle or the parallel movement amount and the rotation angle per one operation performed on the first image larger. In one configuration example of the image matching apparatus of the present invention, the area of the matching area for which the second matching means (31) obtains the coincidence ratio is determined by the first matching means (3).
2) is to make the total area smaller than the total area of each matching area for which the matching rate is to be obtained.

In one embodiment of the image collating apparatus according to the present invention, a reference point detecting means (8) for detecting respective reference points of the first image and the second image at a first initial position.
5) the translation amount of the first image required to match the position of each reference point of the first image and the second image;
Correction amount calculating means (90) for obtaining both the rotation angle or the parallel movement amount and the rotation angle, wherein the first conversion means (13) includes a parallel movement amount, a rotation amount, and a rotation amount obtained by the correction amount calculation means. A position obtained by adding the angle or the amount of translation and the rotation angle to the first initial position is defined as a second initial position, and the first image is moved to the second initial position.
At least one of a translation process and a rotation process is performed on the first image. Also,
One configuration example of the image matching device of the present invention includes: a reference point detection unit (85) that detects a reference point of each of the first image and the second image at a first initial position; Correction amount calculating means (90) for obtaining the translation amount, the rotation angle, or both the translation amount and the rotation angle of the first image necessary for matching the positions of the respective reference points of the image and the second image.
And the second conversion means (14) calculates a position obtained by adding the parallel movement amount, the rotation angle or the parallel movement amount and the rotation angle obtained by the correction amount calculating means to the first initial position. As a new first initial position, the first image is moved to the first initial position, and at least one of a translation process and a rotation process is performed on the first image.
It performs two processes.

According to the image collating method of the present invention, at least one of a translation process and a rotation process is repeatedly executed on a first image at a fixed amount within a preset range. A first conversion procedure for storing one image in a memory, and for each processing of the first conversion procedure, the first image and the second image after the processing are replaced with the first image or the second image. A first collation procedure in which the second image is compared and collated in a plurality of divided collation areas, a matching rate in each collation area is determined and stored in a memory; A maximum matching rate extraction procedure for obtaining a maximum matching rate in a memory and calculating the maximum matching rate average which is an average value of the maximum matching rates for each of the matching areas; If the average is greater than or equal to a preset threshold, the first Wherein the image second image is obtained so as to perform a and a determining procedure is the same. The image matching method according to the present invention further comprises a first conversion procedure, a first matching procedure, a maximum matching rate extracting procedure, and obtaining a minimum matching rate for each of the matching areas from the matching rates in a memory. A minimum matching rate extraction procedure to be stored; an arithmetic procedure of obtaining a difference between the maximum matching rate and the minimum matching rate for each of the matching areas, obtaining a matching rate difference average which is an average value of the differences, and storing the average in a memory; A determining step of determining that the first image and the second image are the same when the average of the coincidence rate differences is equal to or greater than a preset threshold value. Further, the image matching method of the present invention includes the first
, A first matching procedure, a maximum matching rate extracting procedure, a minimum matching rate extracting procedure, and a quotient obtained by dividing the maximum matching rate by the minimum matching rate is obtained for each of the matching areas. A calculating procedure for obtaining a matching rate quotient average which is an average value and storing the same in a memory; And a determination procedure for determining that there is a certain one. Further, the image matching method of the present invention includes a first conversion procedure, a first matching procedure, a maximum matching rate extracting procedure, a minimum matching rate extracting procedure, and an average value of the maximum matching rate for each matching area. A calculation procedure for obtaining a certain maximum match rate average and a minimum match rate average which is an average value of the minimum match rates for each of the matching areas and storing the same in a memory; and a first threshold for setting the maximum match rate average in advance The first image and the second image when the minimum match rate average is smaller than a predetermined second threshold (first threshold ≧ second threshold). And a determination procedure for determining that the images are the same. Further, the image matching method of the present invention includes the first
, A first matching procedure, a minimum matching rate extracting procedure, an arithmetic procedure of obtaining an average of the minimum matching rates, which is an average value of the minimum matching rates for each of the matching areas, and storing the average in a memory. When the rate average is smaller than a preset threshold value, a determination procedure for determining that the first image and the second image are the same is executed.

In one configuration example of the image matching method according to the present invention, before the calculation procedure, the maximum matching rates for each matching area obtained by the maximum matching rate extracting procedure are designated in descending order from the largest matching rate. A selection procedure of extracting the maximum matching rate of the number and storing it in the memory is executed, and the calculation procedure is configured to obtain the average of the maximum matching rates using only the maximum matching rate extracted by the selection procedure. Things. Also,
One configuration example of the image matching method according to the present invention includes, prior to the calculation procedure, a maximum number of maximum matching rates specified in advance in descending order from the maximum matching rates for each matching area obtained in the maximum matching rate extracting procedure. Are extracted and stored in a memory, and a predetermined number of minimum matching rates are extracted in ascending order from the minimum matching rates for each matching region obtained in the minimum matching rate extracting procedure and stored in the memory. Executing a selection procedure, the operation procedure uses only the maximum match rate and the minimum match rate extracted by the selection procedure, the match rate difference average, the match rate quotient mean, the maximum match rate average, the The minimum match rate average is obtained. Further, one configuration example of the image matching method according to the present invention is arranged such that, before the calculation procedure, the minimum number of predetermined numbers is set in ascending order from the minimum matching rate for each matching area obtained in the minimum matching rate extracting procedure. A selection procedure for extracting a match rate and storing the same in a memory is executed, and the arithmetic procedure is configured to obtain the minimum match rate average using only the minimum match rate extracted by the selection procedure. .

In one embodiment of the image matching method according to the present invention, before each procedure, at least one of a translation process and a rotation process is performed on the first image at the first initial position. Is repeatedly executed at predetermined intervals within a preset range, and the second image after processing is stored in the memory.
, And for each processing of this second conversion procedure,
A second collation procedure of comparing and collating the first image after the processing with the second image to obtain a coincidence rate and storing the same in a memory; Storing a parallel movement amount and a rotation angle of the first image from a position to a current position or both of the parallel movement amount and the rotation angle in a memory; The position obtained by adding the parallel movement amount, the rotation angle, or the parallel movement amount and the rotation angle stored in the first initial position to the second initial position is defined as a second initial position.
Is moved, and at least one of a translation process and a rotation process is performed on the first image. In one configuration example of the image matching method of the present invention, a range preset in the first conversion procedure is smaller than a range preset in the second conversion procedure. . Also, one configuration example of the image matching method of the present invention includes a translation amount per one time executed on the first image in the second conversion procedure;
The rotation angle or the translation amount and the rotation angle are set to be larger than the translation amount, the rotation angle, or the translation amount and the rotation angle per one execution performed on the first image in the first conversion procedure. It was made. In one configuration example of the image matching method of the present invention, the area of the matching area for which the matching rate is determined in the second matching procedure is larger than the total area of the matching areas for which the matching rate is determined in the first matching procedure. It is designed to be smaller.

Also, one configuration example of the image collating method of the present invention detects each reference point of the first image and the second image at a first initial position before each procedure. A reference point detection procedure stored in a memory;
And calculating a parallel movement amount, a rotation angle, or both of the parallel movement amount and the rotation angle of the first image necessary to make the positions of the respective reference points of the image coincide with each other and storing the same in a memory. Then, the first conversion procedure sets a position obtained by adding the parallel movement amount, the rotation angle or the parallel movement amount and the rotation angle obtained in the correction amount calculation procedure to the first initial position as a second initial position. After the first image is moved to the second initial position, at least one of a translation process and a rotation process is performed on the first image. Further, in one configuration example of the image matching method of the present invention, before each procedure, a reference point of each of the first image and the second image at a first initial position is detected and stored in a memory. A reference point detecting procedure to be performed, and a translation amount, a rotation angle or a translation amount and a rotation angle of the first image necessary for matching positions of the respective reference points of the first image and the second image. And a correction amount calculating step of obtaining both of them and storing the result in the memory. The first image is moved to the first initial position with the position added to the first position as a new first initial position, and then the first image is translated and rotated. At least one of the processes is executed.

[0018]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS In the present invention, a process of translating, rotating, or translating and rotating a first image in order to correct a displacement between a first image and a second image, and a process after the process. While repeating the process of comparing and collating the first image and the second image to determine the degree of similarity (matching rate), the maximum matching rate is determined for each of the plurality of matching areas, and the matching rate of each matching area is determined. The point of averaging is different from the conventional image matching device.

[First Embodiment] A case where an image is a fingerprint will be described below as an embodiment of the present invention. FIG. 1 is a block diagram illustrating a configuration of an image matching device according to a first embodiment of the present invention. This image matching device includes an image input device 100, an image database 200, and a processing device 300.

The image input device 100 detects unevenness of a fingerprint of a finger placed on a sensor (not shown) of the image input device 100 by a sensor, and performs analog / digital conversion (A / D / A / D conversion) on a signal output from the sensor. Image processing such as D conversion) and binarization. The output of the image input device 100 is a binary image in which the convex portion of the fingerprint is represented by a pixel having black luminance (black pixel), and the concave portion of the fingerprint is represented by a pixel having white luminance (white pixel). The convex portion of the fingerprint may be a white pixel, and the concave portion of the fingerprint may be a black pixel. Hereinafter, the image output by the image input device 100 is referred to as an inspection image.

The image input device 100 detects the capacitance formed between the electrode of the small sense unit two-dimensionally arranged on the LSI chip and the skin of the finger touched via the insulating film. , A capacitance detection type fingerprint sensor for sensing a concave and convex pattern of a fingerprint, an A / D converter for A / D converting an output signal of the sensor, and binarization of output data of the A / D converter. It comprises a processor for executing image processing and a storage device such as a semiconductor memory for storing image data. Regarding the capacitance detection type fingerprint sensor, see, for example, the document “M. Tartagni and R. Guerrieri,“ A fingerprint se
nsor based on the feedback capacitive sensing schem
e ", IEEE J. Solid-State Circuits, vol.33, pp.133-142, J
an, 1998 ".

The image database 200 stores fingerprint images acquired in advance as registration data. The image database 200 is configured by a storage device such as a hard disk device or a nonvolatile memory. Less than,
The image stored in the image database 200 is called a registered image.

The processing device 300 compares the inspection image output from the image input device 100 with the registered image output from the image database 200, and determines whether the two images are the same fingerprint image or different fingerprint images. Is determined. In order to improve the determination accuracy (collation accuracy), the processing device 300 includes a conversion unit 10, an area designation unit 20,
It comprises a matching means 30, a maximum matching rate extracting means 40, a calculating means 50, and a judging means 60. The difference from the conventional example is that an area specifying means 20 and a calculating means 50 are provided. Such a processing device 300 can be realized by a processor and a storage device such as a semiconductor memory.

The conversion means 10 translates (shifts) each pixel of the input inspection image by a predetermined amount from an initial position (the position at the time of input from the image input device 100) in accordance with a translation amount designation signal 401 described later. The inspection image thus output is output. The shift operation by the conversion means 10 will be described. First, a coordinate system is set for the inspection image, a primary conversion is performed to translate the coordinates of each pixel determined by the coordinate system in parallel, and finally, each pixel after the primary conversion is converted. By reconstructing the image based on the coordinates of, an inspection image that is translated is generated.

The area designating means 20 outputs a collation area designation signal 402 for sequentially designating a plurality of collation areas obtained by dividing the inspection image or the registered image one by one. The area specifying means 20 specifies the collation area by specifying the coordinates of the lower left vertex and the coordinates of the upper right vertex of each collation area by the collation area specification signal 402.

The collating means 30 compares and collates the luminance value of each pixel at the same position between the inspection image output from the converting means 10 and the registered image output from the image database 200, and outputs a black image having the same luminance value. The number of pixels is totaled in the collation area, and the total number of matching pixels is divided by the total number of black pixels of the registered image, thereby obtaining the degree of similarity (coincidence rate) 403 between the inspection image and the registered image. The number of matching pixels × 2
The coincidence rate 403 may be set to / (total number of black lines in the registered image + total number of black pixels in the inspection image).

Further, if the moving amount of the inspection image from the initial position to the current position (the position after the parallel movement by the conversion unit 10) is within a predetermined range, the verification unit 30 In order to execute the parallel movement and the calculation of the coincidence rate again, a parallel movement amount designation signal 401 for designating the movement amount of the inspection image per one time is output to the conversion means 10. The conversion means 10 converts the parallel movement amount designation signal 401
The inspection image is translated by the amount specified by.

The maximum coincidence rate extracting means 40 includes a matching means 30.
Calculates and outputs the maximum matching rate (maximum matching rate) 404 from among the matching rates 403 output by. The calculating means 50 calculates and outputs an average value (maximum matching rate average) 405 of the maximum matching rates 404 for each matching area output from the maximum matching rate extracting means 40.

The judging means 60 compares the maximum coincidence rate average 405 with a preset threshold value, and judges that the fingerprint image is the same when the maximum coincidence rate average 405 is equal to or larger than the threshold value. If the average matching rate 405 is smaller than the threshold value, it is determined that the fingerprint image is different.

FIG. 2 shows the collating operation of the image collating apparatus according to the present embodiment. First, the image input device 100 detects a fingerprint of a finger placed on a sensor and generates an inspection image (Step S1). The processing device 300 includes the image input device 100
(Step S2), and when a registered image is input from the image database 200 (step S2).
3) The inspection image is translated by the conversion means 10 (step S4).

The area designating means 20 outputs a collation area designation signal 402 for designating one of a plurality of collation areas set in the inspection image or the registered image (step S5). The matching unit 30 compares and compares the inspection image output from the conversion unit 10 with the registered image output from the image database 200, and specifies the number of black pixels having the same brightness value by the matching region specifying signal 402. Aggregation is performed in the collation area, and the matching rate 403 in the collation area is obtained (step S6).

Next, the maximum matching rate extracting means 40 determines whether the matching rate 403 output from the matching means 30 is the maximum value in the matching area designated by the matching area designation signal 402 (step S7). If it is the maximum value, the maximum value is stored as the maximum matching rate 404 in the collation area designated by the collation area designation signal 402 (step S8).

Subsequently, the area designating means 20 determines whether or not there is an area for which comparison and comparison has not been completed among the plurality of comparison areas (step S9). In this case, the collation area designation signal 40 for designating one collation area to be compared and collated next from the unfinished areas.
2 is output (step S5). In this way, the processing of steps S5 to S8 is performed for each matching area for which comparison and matching has not been completed, and the maximum matching rate 404 is calculated for each matching area.

When the comparison / collation has been completed in all the comparison areas and there is no longer an area for which comparison / comparison has not been completed (step S).
No. 9), the collation means 30 determines whether the movement amount of the inspection image from the initial position to the current position is within a preset range (step S10), and determines whether the movement amount is within the preset range. For example, a parallel movement amount designation signal 401 is output to the conversion means 10. The conversion means 10 translates the inspection image by the amount designated by the translation amount designation signal 401 (step S5). Thus, as long as the movement amount of the inspection image from the initial position to the current position is within a preset range, the processing of steps S4 to S9 is repeated.

If the matching rate 403 output from the matching means 30 is higher than the stored maximum matching rate 404 (YES in step S7), the matching rate 403 is stored as a new maximum matching rate 404. Needless to say (step S8). When the moving amount of the inspection image exceeds a preset range (N in step S10)
O), the calculating means 50 calculates the maximum matching rate average 405, which is the average value of the maximum matching rates 404 for each matching region output from the maximum matching rate extracting means 40 (step S1).
1).

The judgment means 60 compares the maximum coincidence rate average 405 with a preset threshold value (step S12). Judgment, maximum match rate average 405
Is smaller than the threshold value, it is determined that the fingerprint image is different.

The processing device 300 extracts the maximum matching rate 404 corresponding to each collation area while repeatedly executing the comparison collation by the collation means 30 for each collation area designated by the area designation means 20. The difference from the conventional image matching device is that the maximum matching rate is obtained with the optimum amount of parallel movement while the comparison and matching in each matching area designated by the area designating means 20 and the parallel movement by the converting means 10 are repeatedly executed. And average the results.

FIG. 3 shows an example of the collation performed by the conventional collation apparatus. For the same inspection image and registered image as the inspection image and the registered image shown in FIG. FIG. 3A is a schematic diagram showing a state in which the registered image is compared with the registered image, and FIG. 3A shows the comparison and collation between the collated region A and the inspection image among the divided divided regions of the registered image. (B) shows the comparison and collation between the collation area B and the inspection image. When the collation area A of the registered image and the inspection image are optimally aligned by the parallel movement of the inspection image, the matching rate 403 in the collation area A becomes the maximum, and similarly, the collation area B of the registered image and the inspection image become similar. When aligned optimally, the match rate 4 in the matching area B
03 is the maximum.

As described above, by performing the optimal alignment in each of the collation areas A and B, the portions which did not coincide with each other in the image collation apparatus of FIG. 23 (the portions surrounded by broken lines in FIGS. 3 and 25) coincide with each other. Can be done. Therefore,
Partial inconsistency caused by skin distortion or the like is corrected, and the maximum matching rate average 405 in the case of matching (identity matching) between fingerprint images obtained from the same finger is compared with the matching between fingerprint images obtained from different fingers (other person matching). ) Can be increased. As described above, in the present embodiment, it is possible to enlarge the difference between the maximum matching rate averages 405 between the person verification and the other person verification, thereby improving the verification accuracy.

[Second Embodiment] FIG. 4 shows a second embodiment of the present invention.
FIG. 2 is a block diagram illustrating a configuration of an image matching device according to an embodiment of the present invention, in which the same components as those in FIG. 1 are denoted by the same reference numerals. The difference between the present embodiment and the first embodiment is that the processing apparatus 301 includes the minimum matching rate extraction unit 41, the arithmetic unit 51 instead of the arithmetic unit 50, and the determination unit 60 instead of the determination unit 60. This is a point that means 61 is provided.

The minimum matching rate extraction means 41 is
Calculates and outputs the minimum matching rate (minimum matching rate) 406 from the matching rates 403 output by. The calculating means 51 calculates the maximum matching rate 404 output from the maximum matching rate extracting means 50 and the minimum matching rate 40 output from the minimum matching rate extracting means 41.
6 (matching rate difference) is calculated for each matching area specified by the matching area designating signal 402, and then the average value (matching rate difference average) 407 of each matching rate difference calculated for each matching area is calculated. And output. Note that the maximum matching rate 40
4 and the average of the minimum coincidence rates 406, which are the average values of the minimum coincidence rates 406 for each matching region, and calculate the difference between the average of the maximum coincidence rate and the average of the minimum coincidence rates. The average may be 407.

The determining means 61 compares the average matching rate difference 407 with a preset threshold value, and determines that the fingerprint image is the same when the average matching rate difference 407 is equal to or greater than the threshold value. If the average rate difference 407 is smaller than the threshold value, it is determined that the fingerprint image is different.

FIG. 5 shows the collating operation of the image collating apparatus according to the present embodiment. The processing in steps S1 to S8 is exactly the same as in the first embodiment. The minimum matching rate extraction means 41 determines whether the matching rate 403 output from the matching means 30 is the minimum value in the matching area designated by the matching area designation signal 402 (step S13). This minimum value is stored as the minimum matching rate 406 in the matching area designated by the matching area designation signal 402 (step S1).
4).

Since the processing in step S9 is exactly the same as that in the first embodiment, the processing in steps S5 to S8, S13, and S14 is performed for each comparison area for which comparison and comparison has not been completed.
A maximum matching rate 404 and a minimum matching rate 406 are calculated for each matching area. Also, since the processing in step S10 is exactly the same as in the first embodiment, as long as the movement amount of the inspection image from the initial position to the current position is within a preset range, steps S5 to S8, S13, The processing of S14 and S9 is repeated.

When the moving amount of the inspection image exceeds the preset range (NO in step S10), the calculating means 51 calculates an average of matching rate difference 407 from the maximum matching rate 404 and the minimum matching rate 406. Output (Step S1
5). The judging means 61 compares the average 407 of coincidence rate differences with a preset threshold value (step S1).
6) If the average match rate difference 407 is greater than or equal to the threshold value, the fingerprint image is determined to be the same, and if the average match rate difference 407 is smaller than the threshold value, the fingerprint image is determined to be different.

FIG. 6 is a diagram showing the relationship between the translation amount of the inspection image by the conversion means 10 and the coincidence rate 403 outputted by the collation means 30. FIG. FIG. 6B shows the case of verification (identification).
Indicates a case of collating fingerprint images obtained from different fingers (other person collation). In FIG. 6, C is the match rate difference in the case of the identity verification, and D is the match rate difference in the case of the others verification.

Comparing the coincidence rate differences C and D reveals that there is a clear difference in the coincidence rate difference between the case of identity verification and the case of others verification. The reason will be described with reference to an enlarged schematic diagram of a fingerprint shown in FIG. 7 (a) and 7 (b) show a case of identity verification, and FIG. 7 (b) shows a case where the inspection image of FIG. 7 (a) is translated in the direction of the arrow. FIG. 7C and FIG. 7D show the case of the other person collation.
FIG. 7D shows a case where the inspection image of FIG. 7C is translated in the direction of the arrow.

In the case of identity verification, the periodicity of the convex portion (black pixel) of the fingerprint is the same between the registered image and the inspection image.
FIG. 7B shows the vicinity of the position where the maximum matching rate of FIG.
, There is a tendency that the minimum matching rate can also be obtained. On the other hand, in the case of another person verification in which the periodicity of the convex portion of the fingerprint is different between the registered image and the inspection image, FIGS. 7C and 7D
As described above, the number of matching pixels due to the intersection of the registered image and the inspection image increases, and there is no tendency as described in the case of the identity verification.
This difference is the reason why the difference in the judgment index (matching rate difference) between the personal verification and the others verification becomes large. As mentioned above,
In the present embodiment, the difference between the judgment indices between the personal verification and the other person verification can be enlarged as compared with the case of the first embodiment, and the verification accuracy can be further improved.

[Third Embodiment] FIG. 8 shows a third embodiment of the present invention.
FIG. 13 is a flowchart illustrating a collation operation of the image collation device according to the embodiment of the present invention. Also in the present embodiment, the configuration as an image matching device is substantially the same as that of the second embodiment, and therefore, the description will be made using the reference numerals in FIG.

The difference between the present embodiment and the second embodiment is that instead of the calculating means 51 of the processing unit 301 obtaining the average of the matching rate differences, the maximum matching rate 404 is changed to the minimum matching rate 406.
(Computation rate quotient) is calculated for each collation area specified by the collation area designation signal 402, and then the average value (coincidence rate quotient) of each quotient calculated for each collation area is calculated. The determination means 61 uses the average of the coincidence rate quotient outputted from the arithmetic means 51 as a judgment index and compares this average of the coincidence rate quotient with a preset threshold value. Note that the maximum matching rate average, which is the average value of the maximum matching rates 404 for each matching area, and the minimum matching rate average, which is the average value of the minimum matching rates 406 for each matching area, are calculated. The quotient divided by the rate average may be used as the coincidence rate quotient average.

Next, the operation of the image collating apparatus according to this embodiment will be described. Steps S1 to S8, S13, S14, S
9. The processing in S10 is exactly the same as in the second embodiment. When the moving amount of the inspection image from the initial position to the current position exceeds a preset range (NO in step S10), the calculating unit 51 calculates a matching rate quotient average from the maximum matching rate 404 and the minimum matching rate 406. Calculate (Step S
17).

The judging means 61 compares this quotient average of coincidence rates with a preset threshold value (step S1).
8) If the quotient of coincidence rates is equal to or greater than the threshold, the fingerprint images are determined to be the same, and if the quotient of coincidence is smaller than the threshold, the fingerprint images are determined to be different. In the present embodiment, when the minimum matching rate 406 between the inspection image and the registered image is smaller than the maximum matching rate 404 by two digits or more, it is possible to increase the difference between the judgment indices between the personal verification and the others verification, The matching accuracy can be improved.

[Fourth Embodiment] FIG. 9 shows a fourth embodiment of the present invention.
5 is a block diagram illustrating a configuration of an image matching device according to an embodiment of the present invention, in which the same components as those in FIG. 4 are denoted by the same reference numerals. The difference between the present embodiment and the second embodiment is that the arithmetic unit 52 of the processing device 302 uses the maximum matching rate average 408, which is the average value of the maximum matching rates 404 for each matching area, and the minimum matching rate 408 for each matching area. The point at which the average of the minimum coincidence rates 409, which is the average value of the coincidence rates 406, is calculated, and the determination unit 62 compares each of the maximum average of the coincidence rates 408 and the minimum average of the minimum coincidence rates 409 with a preset threshold value. Maximum match rate average 40
8 is greater than or equal to the first threshold value and the minimum matching rate average is 409
Is the second threshold (first threshold ≧ second threshold)
Only when the size is smaller than that, the inspection image and the registered image are determined to be the same fingerprint image.

FIG. 10 is a flowchart showing the collating operation of the image collating apparatus of the present embodiment, and the same processes as those in FIG. Steps S1 to S8,
The processes in S13, S14, S9, and S10 are exactly the same as those in the second embodiment. When the moving amount of the inspection image from the initial position to the current position exceeds the preset range (NO in step S10), the calculating unit 52 determines the maximum matching rate, which is the average value of the maximum matching rates 404 for each matching area. The average rate 408 and the average minimum match rate 409, which is the average value of the minimum match rates 406 for each matching area, are calculated and output (step S19).

The judging means 62 calculates the maximum coincidence rate average 408 output from the calculating means 52 and the first preset coincidence rate.
(Step S20), and when the average maximum matching rate 408 is smaller than the first threshold, it is determined that the fingerprint image is different. If the maximum matching rate average 408 is equal to or greater than the first threshold value in step S20,
2 is the minimum matching rate average 40 output from the calculating means 52
9 is compared with a preset second threshold value (step S21). When the minimum matching rate average 409 is smaller than the second threshold value, it is determined that the fingerprint images are the same, and the minimum matching rate average is determined. If 409 is greater than or equal to the second threshold, it is determined that the fingerprint image is different.

FIG. 11 is a diagram showing the relationship between the amount of translation of the inspection image by the conversion means 10 and the coincidence rate 403 outputted by the collation means 30. FIG. FIG. 11 shows the case of verification (identification).
(B) shows a case of collation (other person collation) between fingerprint images acquired from different fingers. The range of the matching rate in which the threshold value can be set in the conventional matching device of FIG. 23 is E shown in FIG. On the other hand, in the present embodiment, since not only the maximum matching rate but also the minimum matching rate are added to the determination index, the threshold value can be set within the range of E and F shown in FIG.

As described above, in the present embodiment, the range of the matching rate in which the threshold can be set can be expanded, and the matching accuracy can be improved. When the first threshold value and the second threshold value are set to the same value, this value is set within the range of E shown in FIG. Larger than the average of the maximum matching rate in the case of identity verification, or within the range of F (exactly, less than the average of the minimum matching rate in the case of others verification, and less than the average of the minimum matching rate in the case of identity verification) Large range).
When the first threshold value and the second threshold value are set to different values, the first threshold value is set within the range of E, and the second threshold value is set within the range of F. Should be set to.

[Fifth Embodiment] FIG. 12 is a block diagram showing a configuration of an image matching apparatus according to a fifth embodiment of the present invention. The same components as those in FIG. I have. The difference between the present embodiment and the fourth embodiment is that the processing device 303 does not have the maximum matching rate extraction unit 50 and that the processing unit An arithmetic unit 53 for calculating a minimum matching rate average 409, which is an average value. The point is that a determination means 63 for determining that there is is provided.

FIG. 13 is a flowchart showing the collating operation of the image collating apparatus of the present embodiment, and the same processes as those in FIG. 5 are denoted by the same reference numerals. Steps S1 to S6
The processes in S13, S14, S9, and S10 are the same as those in the second embodiment. When the moving amount of the inspection image from the initial position to the current position exceeds a preset range (NO in step S10), the calculating unit 53 sets the minimum matching rate which is the average value of the minimum matching rates 406 for each matching area. Rate average 40
9 is calculated and output (step S22).

The judging means 63 compares the minimum matching rate average 409 outputted from the calculating means 53 with a preset threshold value (step S23). Are determined to be the same fingerprint image, and if the minimum matching rate average 409 is equal to or greater than the threshold value, the fingerprint image is determined to be different.

In this embodiment, the maximum coincidence rate extracting means 5
Since the processing can be simplified by reducing 0, the processing time can be reduced as compared with the fourth embodiment. Note that, in the present embodiment, the threshold value is set within the range of F shown in FIG. ) Must be set.

[Sixth Embodiment] FIG. 14 is a block diagram showing the configuration of an image matching apparatus according to a sixth embodiment of the present invention. The same components as those in FIG. I have. The difference between the present embodiment and the first embodiment is that the processing device 304 includes a selection unit 70.
The selection means 70 selects m (m
Is a natural number of 2 or more and the number of matching areas is 4).
04 (n is a natural number, n <m)
Is extracted and output as the maximum matching rate 410.

FIG. 15 is a flowchart showing the collating operation of the image collating apparatus of the present embodiment, and the same processes as those in FIG. Steps S1 to S10
Is exactly the same as in the first embodiment. When the moving amount of the inspection image from the initial position to the current position exceeds a preset range (NO in step S10), the selection / selection unit 70 selects n from the maximum matching rates 404 for each collation area in descending order. The maximum matching rate 404 is extracted and output as the maximum matching rate 410, and the calculating means 50 calculates the maximum matching rate average 405, which is the average value of the n maximum matching rates 410 output from the selection means 70. (Step S24).

The determining means 60 compares the maximum coincidence rate average 405 with a preset threshold value (step S25). If the maximum coincidence rate average 405 is equal to or larger than the threshold value, the same fingerprint image is determined. Judgment, maximum match rate average 405
Is smaller than the threshold value, it is determined that the fingerprint image is different.

In the present embodiment, the matching area in which the distortion of the skin is large and which is difficult to correct is excluded from the calculation target for calculating the maximum matching rate average 405, so that the maximum matching rate average 405 in the case of the identity verification is reduced. It can be even larger. By further increasing the maximum matching rate average 405, the matching accuracy can be improved as compared with the first embodiment.

In the present embodiment, the processing device 304
Of the processing apparatus 30 of the first embodiment
Although 0 is used, the configurations of the second to fifth embodiments may be used. In the case of the second to fourth embodiments, m maximum matching rates 40 calculated for each collation area
In addition to extracting the n maximum matching rates 404 in descending order from the maximum matching rate 404 by the selection unit, the m minimum matching rates 406 calculated for each collation area are set to 1 (m is a natural number, l <M) The minimum matching rate 406 is extracted by the selection means, and the extracted maximum matching rate 404 and minimum matching rate 406 are extracted.
May be given to the calculation means 51 and 52.

In the case of the fifth embodiment, 1 (where l is a natural number and l <m) minimum matching rates 406 are selected in ascending order from among the m minimum matching rates 406 calculated for each collation area. May be extracted by the selection means, and the extracted minimum matching rate 406 may be provided to the calculation means 53.

[Seventh Embodiment] FIG. 16 is a block diagram showing the configuration of an image matching apparatus according to a seventh embodiment of the present invention. The same components as those in FIG. I have. The difference between the present embodiment and the first embodiment is that, first, the processing unit 305 uses the conversion unit 1 as a unit for roughly correcting the relative positional shift between the registered image and the inspection image.
1, a matching unit 31, and a storage unit 80.
The point is that the amount of movement when the maximum matching rate is obtained is stored in the storage means 80 while repeating the parallel movement by 1 and the comparison and collation by the collation means 31. Second, the movement amount is output from the storage unit 80 to the conversion unit 12 corresponding to the conversion unit 10 of the first embodiment, and the second initial position based on the movement amount and the first initial position is output. Is that the translation of the inspection image by the conversion means 12 is started.

The conversion means 11 converts each pixel of the input inspection image from a first initial position (the position at the time of input from the image input device 100) to a translation amount designation signal 41 described later.
An inspection image translated by a certain amount according to 1 is output. The matching unit 31 compares and compares the luminance value of each pixel at the same position between the inspection image output from the conversion unit 11 and the registered image output from the image database 200, and determines the number of black pixels having the same luminance value. Are calculated in the second collation area, and the calculated number of matching pixels is divided by the number of black pixels of the registered image to obtain a matching rate 412 between the inspection image and the registered image.

Further, the collation means 31 outputs a movement amount signal 4 representing the movement amount of the inspection image from the first initial position to the current position (the position after the parallel movement by the conversion means 11).
13 is output to the storage means 80, and if this movement amount is within the first range set in advance, the conversion means 11 Then, a parallel movement amount designation signal 411 for designating a movement amount per inspection image is output. Conversion means 11
Translates the inspection image by an amount designated by the translation amount designation signal 411. Here, the first range is the same as the range set in the matching unit 30 of the first embodiment.

The storage means 80 stores the movement amount signal 413 output from the matching means 31 when the coincidence rate 412 output from the matching means 31 becomes maximum, and uses this as the movement amount signal 414. Output to The conversion unit 12 moves the inspection image input from the image input device 100 to the second initial position, using a position obtained by adding the movement amount represented by the movement amount signal 414 to the first initial position as a second initial position. After that, an inspection image is output in which each pixel of the inspection image is translated by a predetermined amount in accordance with the translation amount designation signal 402.

The matching unit 32 obtains the coincidence rate 403 between the inspection image output from the conversion unit 12 and the registered image output from the image database 200 in the same manner as the matching unit 30. Further, the collation unit 32 determines that the movement amount of the inspection image from the second initial position to the current position (the position after the parallel movement by the conversion unit 12) is within the second range set in advance. And outputs a parallel movement amount designation signal 402 for designating a movement amount per inspection image. Conversion means 1
No. 2 translates the inspection image by an amount designated by the translation amount designation signal 402. Here, the second range is set to a range smaller than the first range.

FIG. 17 is a flowchart showing the collating operation of the image collating apparatus of the present embodiment, and the same processes as those in FIG. 2 are denoted by the same reference numerals. First, the image input device 1
00 detects a fingerprint of a finger placed on the sensor and generates an inspection image (step S1). The processing device 305
When an inspection image is input from the image input device 100 (step S26) and a registered image is input from the image database 200 (step S27), the inspection image is translated by the conversion unit 11 (step S28). Note that the initial translation amount may be zero. Then, the matching means 31
The inspection image output from the conversion unit 11 and the registered image output from the image database 200 are compared and collated to determine a coincidence rate 412 (step S29).

The storage means 80 determines whether or not the coincidence rate 412 output from the collation means 31 is the maximum value (step S30). If the coincidence rate 412 is the maximum value, the movement amount signal 413 is stored (step S31). The collation unit 31 determines whether the movement amount of the inspection image from the first initial position to the current position is within a first range set in advance (step S32), and determines whether the movement amount is within the first range. If so, a parallel movement amount designation signal 411 is output to the conversion means 11. Thus, as long as the movement amount of the inspection image from the first initial position to the current position is within the first range, the processing of steps S28 to S31 is repeated.

When the moving amount of the inspection image exceeds the first range (NO in step S32), the storage means 80
The stored movement amount signal 413 is output as the movement amount signal 414. When the moving amount of the inspection image exceeds the first range, the conversion unit 12 sets the position obtained by adding the moving amount represented by the moving amount signal 414 to the first initial position as a second initial position. The inspection image input from the input device 100 is moved to the second initial position (Step S33).
Then, the conversion unit 12 translates the inspection image according to the translation amount designation signal 402 (step S3).
4).

The processing in step S5 is exactly the same as in the first embodiment. The matching unit 32 compares and matches the inspection image output from the conversion unit 12 with the registered image output from the image database 200, and specifies the number of black pixels having the same brightness value by the matching region designation signal 402. Aggregation is performed within the collation area, and the matching rate 403 is obtained (step S3).
5). The processing of steps S7 to S9 is exactly the same as in the first embodiment.

The collating means 32 determines whether the movement amount of the inspection image from the second initial position to the current position is within a second range set in advance (step S36), and determines whether the movement amount is equal to the second movement amount. If it is within the range, a translation amount designation signal 402 is output to the conversion means 12. Thus, as long as the movement amount of the inspection image from the second initial position to the current position is within the second range, steps S34, S5, S35, S7-S
Step 9 is repeated. The processing in steps S11 and S12 is exactly the same as in the first embodiment.

In the present embodiment, in order to roughly correct the relative displacement between the registered image and the inspection image, the maximum matching rate is obtained within the first range, and the position at which the maximum matching rate is obtained is determined by the second position. Since the process described in the first embodiment is executed within the second range smaller than the first range as the initial position of the second position, the range set in the matching unit 30 of the first embodiment and the second range If the range of 1 is the same, only the maximum matching rate needs to be obtained within the first range, so that the processing time can be reduced as compared with the first embodiment.

[Eighth Embodiment] In the seventh embodiment, the parallel movement amount designation signal 411 output by the collation means 31 for roughly correcting the relative displacement between the registered image and the inspection image is obtained. Although the difference of the translation amount designation signal 402 output from the collation means 32 for comparison and collation for each collation region is not mentioned, the translation amount per one time indicated by the translation amount designation signal 411 is designated by the translation amount designation. The movement amount per one time represented by the signal 402 may be larger. Also in this case, the configuration as the image matching device and the matching operation thereof are as follows.
This is as shown in FIGS.

As described above, in this embodiment, the amount of movement of the inspection image at one time when the relative displacement between the registered image and the inspection image is roughly corrected is increased. And the number of times of comparison and collation by the collation means 31 can be reduced, and the processing time can be shortened as compared with the seventh embodiment.

[Ninth Embodiment] In the ninth embodiment, the difference between the second matching area where the matching means 31 calculates the matching rate 412 and the matching area where the matching means 32 calculates the matching rate 403 will be described. Although not mentioned, the area of the second collation area preset in the collation means 31 is determined by the area designation means 20.
May be smaller than the total area of all of the m matching regions designated by. Also in this case, the configuration as the image collating device and the collating operation thereof are as shown in FIGS.

Thus, in this embodiment, when the relative displacement between the registered image and the inspection image is roughly corrected, the number of pixels to be compared and collated can be reduced, so that the processing time can be shortened.

In the seventh to ninth embodiments, the processing apparatus 300 according to the first embodiment is used as a base of the processing apparatus 305, but the second embodiment to the sixth embodiment.
The configuration of the embodiment may be used. In the case of the second to sixth embodiments, the conversion means 11, the collation means 31 and the storage means 80 are added, the conversion means 12 is used in place of the conversion means 10, and the collation means is used in place of the collation means 30. Means 32 may be used.

[Tenth Embodiment] FIG. 18 is a block diagram showing the configuration of an image matching apparatus according to a tenth embodiment of the present invention. The same components as those in FIG. I have. The difference between the present embodiment and the first embodiment is that the processing device 306 uses a reference point detecting means 85 and a correction amount calculating means 90, and outputs a moving amount from the correction amount calculating means 90 to the converting means 13 corresponding to the converting means 10 of the first embodiment. Conversion means 13 from the initial position
Is the point at which the parallel movement of the inspection image is started.

The reference point detecting means 85 is provided in the image input device 10.
For each of the inspection image output from 0 and the registered image output from the image database 200, a reference point of the image such as an approximate center point is detected, and the coordinates 415 of the reference point of the inspection image and the reference point of the registered image are detected. Is output. The search for the approximate center point is performed by sequentially approaching the center position using, for example, the number of intersections of each side of the rectangle with the parallel line (reference: Ito et al., "Classification of Fingerprint Images Focusing on Center Point") Law, “Religion Techniques, PRU89-79, pp15-22,
1989). In the case of a fingerprint image, the approximate center point is a position where the curvature of the fingerprint is large.

The correction amount calculating means 90 includes the reference point detecting means 8
5. The amount of parallel movement of the inspection image required to match the position of the reference point of the inspection image detected by step 5 with the position of the reference point of the registered image is obtained, and a movement amount signal 417 representing the amount of parallel movement is output. I do. The correction amount calculating means 90 can be realized by a method of calculating the difference between the position vector of the reference point of the inspection image and the position vector of the reference point of the registered image and using the difference as the translation amount.

The conversion means 13 sets the position obtained by adding the movement amount represented by the movement amount signal 417 to the first initial position as the second initial position, and sets the inspection image input from the image input device 100 to the second initial position. Then, an inspection image in which each pixel of the inspection image is translated by a predetermined amount in accordance with the translation amount designation signal 402 is output.

FIG. 19 is a flowchart showing the collating operation of the image collating apparatus of the present embodiment, and the same processes as those in FIG. 2 are denoted by the same reference numerals. First, the image input device 1
00 detects a fingerprint of a finger placed on the sensor and generates an inspection image (step S1). The processing device 306
When the inspection image is input from the image input device 100 (step S37), and the registered image is input from the image database 200 (step S38), the reference point detection unit 85 detects the respective reference points of the inspection image and the registered image. (Step S39).

The correction amount calculating means 90 obtains a parallel movement amount of the inspection image necessary to make the position of the reference point of the inspection image coincide with the position of the reference point of the registered image, and calculates the movement amount representing the parallel movement amount. An amount signal 417 is output (step S40).
Next, the conversion unit 13 sets the position obtained by adding the movement amount represented by the movement amount signal 417 to the first initial position as the second initial position, and sets the inspection image input from the image input device 100 to the second initial position. It is moved to the position (step S41). Then, the conversion unit 13 translates the inspection image according to the translation amount designation signal 402 (step S42).
The processing of steps S5 to S12 is exactly the same as in the first embodiment.

In this embodiment, in order to roughly correct the relative displacement between the registered image and the inspection image, the reference point detecting means 85 detects the respective reference points of the inspection image and the registered image. Since the processing described in the first embodiment is executed with the position where the points match as the second initial position, the processing time can be reduced as compared with the first embodiment.

In this embodiment, the processing device 306
Of the processing apparatus 30 of the first embodiment
Although 0 is used, the configurations of the second to sixth embodiments may be used. In the case of the second to sixth embodiments, the reference point detecting means 85 and the correction amount calculating means 90 may be added, and the converting means 13 may be used instead of the converting means 10.

[Eleventh Embodiment] As shown in FIG. 20, the seventh embodiment and the tenth embodiment may be combined. 20, the same components as those in FIGS. 16 and 18 are denoted by the same reference numerals. Conversion means 14
Sets the position obtained by adding the movement amount represented by the movement amount signal 417 to the first initial position as a new first initial position, and sets the inspection image input from the image input device 100 to this new first initial position. After the movement, each pixel of the inspection image is translated by a predetermined amount according to the translation amount designation signal 411 to output an inspection image.

FIG. 21 is a flowchart showing the collating operation of the image collating apparatus according to the present embodiment.
The same processes as those in FIG. 14 are denoted by the same reference numerals. The processing in steps S1 and S37 to S40 is exactly the same as in the tenth embodiment. Further, the processing in steps S26 and S27 is exactly the same as in the seventh embodiment.

The conversion means 14 sets the position obtained by adding the moving amount represented by the moving amount signal 417 to the first initial position as a new first initial position, and uses the inspection image input from the image input device 100 as this new initial position. Move to the first initial position (step S43). Then, the conversion means 14 translates the inspection image in accordance with the translation amount designation signal 411 (step S44).

Steps S29 to S32, S2, S3, S
33, S34, S5, S35, S7 to S9, S36, S
Steps S11 and S12 are exactly the same as those of the seventh embodiment.
Thus, in the present embodiment, the effect of increasing the processing speed according to the seventh embodiment and the effect of increasing the processing speed according to the tenth embodiment can be simultaneously obtained.

[Twelfth Embodiment] FIG. 22 is a block diagram showing a configuration of an image matching apparatus according to a twelfth embodiment of the present invention. The configuration of the image matching apparatus described in the first to eleventh embodiments can be realized by the computer 308. The computer 308 includes a CPU 500,
ROM (Read Only Memory) 501, RAM (Random Ac)
cess memory) 502, an auxiliary storage device 503 such as a flexible disk device, a large-capacity auxiliary storage device 504 such as a hard disk device, an interface device 505 serving as an interface with a display device (not shown) for outputting video, and a keyboard (not shown). Interface device 506 serving as an interface with the image input device 100
It has a configuration such as an interface device 507 serving as an interface with the device.

The image database 200 has a RAM 502
Alternatively, it can be realized by the auxiliary storage device 504.
In the apparatus shown in FIG. 22, the program for realizing the image collating method of the present invention is a flexible disk, a CD, or the like.
-Provided in a state recorded on a recording medium such as a ROM and a memory card. When this recording medium is inserted into the auxiliary storage device 503 of the computer 308, the program recorded on the medium is read. Then, the CPU 500 writes the read program in the RAM 502 or the auxiliary storage device 504, and executes the first to first programs in accordance with the program.
The processing as described in the first embodiment is executed. Thus, the same processing as in the first to eleventh embodiments can be realized.

In the first to twelfth embodiments, the case where the image to be collated is a fingerprint image has been described. However, an image having periodicity such as a nose pattern, an iris or a texture pattern of an animal is described. Then, the present invention can be applied. In particular, when the image is deformed due to the detection of the shape of a soft object such as an animal's nose by a contact-type detection device, the matching accuracy can be improved by applying the present invention.

[0099] The matching means 30 to 32 use a black pixel whose luminance value perfectly matches in the inspection image and the registered image as a matching pixel. The number of matches / the number of white pixels in the registered image, or the number of matched pixels × 2 / (the number of white pixels in the registered image + the number of white pixels in the inspection image) may be used as the match rate. In the above embodiment, the registered image and the inspection image are binary images, but a grayscale image of, for example, 256 gradations before the binarization processing may be used. In this case, a pixel whose difference in luminance value between the inspection image and the registered image is within a predetermined range may be set as a matching pixel.

The conversion means 10 to 14 translate the inspection image in parallel in order to correct the positional deviation between the inspection image and the registered image, but may rotate the inspection image. May be executed. When both the parallel movement and the rotation are performed, not only the positional deviation but also the angular deviation can be corrected. When performing both the translation and the rotation, for example, after rotating the inspection image by a fixed angle, the translation of the inspection image is repeated as described above, and the total amount of the translation exceeds the preset range. At this point, the inspection image may be rotated again by a certain angle, and the parallel movement of the inspection image may be repeated.

However, when performing both the parallel movement and the rotation, it is necessary to set the respective ranges of the parallel movement amount and the rotation angle in the matching means 30 to 32. Collation means 30
If the movement amount of the inspection image from the first or second initial position to the current position is within the range of a predetermined parallel movement amount, the parallel movements to specify the movement amount of the inspection image per one time. The movement amount designation signals 402 and 411 are output, and if the rotation angle of the inspection image from the first or second initial position to the current position is within the range of the preset rotation angle, one time of the inspection image is output. Outputs a rotation angle specification signal that specifies the rotation angle. When only rotation is performed, a rotation angle designation signal is output.

The conversion means 10 to 14 translate the inspection image in accordance with the translation amount designation signals 402 and 411, and rotate the inspection image in accordance with the rotation angle designation signal. Also,
When performing both translation and rotation, the matching unit 31
Outputs and stores a movement amount signal 413 indicating the movement amount of the inspection image from the first initial position to the current position and an angle signal indicating the rotation angle of the inspection image from the first initial position to the current position. The means 80 stores the movement amount signal 413 and the angle signal output from the matching means 31 when the coincidence rate 412 becomes the maximum, and outputs them to the conversion means 12.

In the case of performing both the parallel movement and the rotation, when the reference point detection means 85 obtains a plurality of reference points for each of the registered image and the inspection image, the correction amount calculation means 9
0 calculates the rotation angle of the inspection image necessary to make the position of the reference point of the inspection image coincide with the position of the reference point of the registered image, and outputs an angle signal representing the rotation angle. The conversion means 12 to 14 are stored in the storage means 80 or the correction amount calculation means 90.
A position obtained by adding the rotation angle represented by the angle signal output from the first initial position to the first initial position is set as a new first initial position or second initial position. Further, the processing of the conversion units 10 to 14 may be executed for a registered image instead of the inspection image.

Japanese Patent Application Laid-Open Nos. 5-242222 and 9-282458 disclose image collating apparatuses for the purpose of coping with image deformation due to distortion.
In the image matching apparatus disclosed in Japanese Patent Application Laid-Open No. 5-242222, a registered image is divided into matching areas, and the result of feature point matching is determined based on the number of matching areas exceeding a threshold. In this method, feature point collation is performed for each partial area to determine whether the fingerprints are the same or different. On the other hand, the present invention is fundamentally different in that the match / mismatch is not determined in the matching region alone, and the match / mismatch is determined using an average value of the matching rates of the matching regions as an index.

The image collating apparatus disclosed in Japanese Patent Application Laid-Open No. 9-282458 not only divides a registered image into collation regions, but also moves a collation region of an inspection image corresponding to the collation region of the registered image from a center position to a relative position. And a correlation value is determined between the corresponding collation regions to determine a match / mismatch. On the other hand, the present invention is fundamentally different in that one of the registered image and the inspection image is divided into a matching area, and the other image is used as a matching area to obtain a matching rate.

Japanese Patent Application Laid-Open No. H10-214343 discloses an image collating apparatus that cuts out a plurality of images from one image and performs collation. The image matching apparatus disclosed in Japanese Patent Application Laid-Open No. H10-214343 stores a distribution of coincidences when a cut-out image is compared and collated while scanning the other image, and determines a coincidence based on a totaling result of the coincidence distribution. -Determine mismatch. In this method, the coincidence distribution is stored in a storage device, and the positional relationship between the coincidence distributions needs to be stored in advance. On the other hand, according to the present invention, the inspection image and the registered image are compared and collated, and the coincidence / mismatch is determined by a simple process by storing only the coincidence rate corresponding to each collation region. Different.

[0107]

According to the present invention, by determining the maximum matching rate for each collation area, it is possible to perform optimal positioning in each collation area, and to locally correct the first image or the second image. Can be corrected. As a result, the first
Even if there is a local distortion in the image or the second image, the difference between the maximum matching rate averages between the personal verification and the others verification can be enlarged, and the verification accuracy can be improved.

Also, the first conversion means, the first collation means,
The difference between the maximum matching rate output from the maximum matching rate extracting means, the minimum matching rate extracting means, and the minimum matching rate output from the minimum matching rate extracting means is determined for each matching area. Calculating means for calculating the average of the matching rate differences, which is the average value, and the first image and the second image when the average of the matching rate differences output from the calculating means are equal to or greater than a predetermined threshold value.
By providing determination means for determining that the images are the same, local distortion of the first image or the second image can be corrected. In addition, by using the average of the matching rate differences as the determination index, the difference in the determination index between the personal verification and the other person verification can be expanded as compared with the case where only the maximum matching rate is used, so that the verification accuracy is further improved. Can be done.

Also, the first conversion means, the first collation means,
A quotient obtained by dividing the maximum matching rate output from the maximum matching rate extracting means, the minimum matching rate extracting means and the maximum matching rate extracting means by the minimum matching rate output from the minimum matching rate extracting means is obtained for each matching area. Calculating means for obtaining a matching rate quotient average which is an average value of the first and second images, and when the matching rate quotient average outputted from the calculating means is equal to or greater than a preset threshold value, the first image and the second image are the same. Is provided, it is possible to correct local distortion of the first image or the second image. Further, by using the quotient of coincidence rates as a judgment index, when the minimum coincidence rate between the first image and the second image is smaller than the maximum coincidence rate by two digits or more, the judgment between the identity verification and the others verification is made. Since the difference between the indices can be enlarged, the matching accuracy can be further improved.

Further, the first conversion means, the first collation means,
The maximum match rate average, which is the average value of the maximum match rates for each matching area output from the maximum match rate extracting means, the minimum match rate extractor, and the matching area output from the maximum match rate extractor, and the matching area output from the minimum match rate extractor. Calculating means for obtaining an average value of the minimum coincidence rates, which is an average value of the minimum coincidence rates, and an average of the maximum coincidence rates output from the arithmetic means being equal to or greater than a first threshold value set in advance and being output from the arithmetic means. The first image or the second image is provided by providing a determination unit that determines that the first image and the second image are the same when the minimum matching rate average is smaller than a preset second threshold value. Can correct the local distortion of the image. Further, since the range of the matching rate in which the threshold value can be set can be expanded, the matching accuracy can be further improved.

Also, the first conversion means, the first collation means,
The minimum matching rate extracting means, the calculating means for obtaining the average of the minimum matching rates for each collation region output from the minimum matching rate extracting means, and the minimum matching rate average output from the calculating means are preset. Correcting local distortion of the first image or the second image by providing a determination unit for determining that the first image and the second image are the same when the threshold value is smaller than the threshold value. Can be. Further, since the processing can be simplified by reducing the maximum matching rate extraction means, the processing time can be shortened.

Further, by providing sorting means for extracting a predetermined number of maximum matching rates in descending order from the maximum matching rate for each collation area output from the maximum matching rate extracting means, the skin distortion and the like can be obtained. By excluding the matching region where the correction is difficult and the calculation for calculating the average maximum matching rate is excluded, the average maximum matching rate in the case of the identity verification can be further increased. As a result, the matching accuracy can be further improved.

Also, a predetermined number of maximum matching rates are extracted in descending order from the maximum matching rates for each matching area output from the maximum matching rate extracting means, and the matching rates output from the minimum matching rate extracting means are extracted. By providing a selection unit that extracts a predetermined number of minimum matching rates in ascending order from the minimum matching rate for each area, skin distortions and the like are difficult to correct, and the matching areas are large, and the matching rate difference average, By excluding the average of the matching rate quotient, the average of the maximum matching rate, and the average of the minimum matching rate from the calculation, the average of the maximum matching rates in the case of the identity verification can be further increased. As a result, the matching accuracy can be further improved.

Further, by providing sorting means for extracting a predetermined number of minimum matching rates in ascending order from the minimum matching rates for each collation area output from the minimum matching rate extracting means, the distortion of the skin, etc. By excluding the matching region where the correction is difficult to correct from the calculation target for calculating the minimum matching rate average, the minimum matching rate average in the case of the identity verification can be further reduced. As a result, the matching accuracy can be further improved.

Further, a second conversion means, a second collation means, and a storage means are provided, and a range preset for the first conversion means is made smaller than a range preset for the second conversion means. The first conversion means sets the parallel movement amount, the rotation angle or the parallel movement amount and the rotation angle stored in the storage means to a first initial position as a second initial position, and sets the second initial position as a second initial position.
After moving the first image to the initial position of the first image, and performing at least one of the translation process and the rotation process on the first image, the relative image between the registered image and the inspection image is obtained. The second conversion means, the second collation means, and the storage means for roughly correcting the positional deviation need only determine the maximum matching rate, so that the processing time can be reduced.

Further, the first conversion means converts the amount of parallel movement, rotation angle or the amount of parallel movement and rotation angle to be executed by the second conversion means for the first image into the first image. By making the parallel movement amount, the rotation angle, or the parallel movement amount and the rotation angle larger at one time, the positional deviation between the registered image and the inspection image is roughly corrected. Since the amount of movement is increased, the number of times of parallel movement processing by the second conversion means and the number of times of comparison and matching by the second matching means can be reduced.
Processing time can be reduced.

Further, by making the area of the matching area for which the second matching means finds the matching rate smaller than the total area of each matching area for which the first matching means finds the matching rate, the registered image and the inspection image can be compared. When the relative displacement is roughly corrected, the number of pixels to be compared and matched can be reduced, so that the processing time can be reduced.

Further, reference point detecting means and correction amount calculating means are provided, and the first converting means converts the parallel movement amount, the rotation angle or the parallel movement amount and the rotation angle obtained by the correction amount calculating means into a first initial value. The position added to the position is defined as a second initial position.
After moving the first image to the second initial position, by executing at least one of a translation process and a rotation process on the first image, the registered image and the inspection image are compared. Since the reference points of the inspection image and the registered image are detected by the reference point detecting means in order to roughly correct the positional deviation, and the position where these reference points coincide is set as the second initial position, the processing time is reduced. Can be.

Further, reference point detecting means and correction amount calculating means are provided, and the second converting means converts the parallel movement amount, the rotation angle or the parallel movement amount and the rotation angle obtained by the correction amount calculating means into a first initial value. The position added to the position is set as a new first initial position, the first image is moved to the first initial position, and at least one of the translation processing and the rotation processing is performed on the first image. By executing one process, it is possible to simultaneously obtain the effect of speeding up processing by the second conversion unit, the second matching unit, and the storage unit, and the effect of speeding up processing by the reference point detection unit and the correction amount calculation unit. Can be.

[Brief description of the drawings]

FIG. 1 is a block diagram illustrating a configuration of an image matching device according to a first embodiment of the present invention.

FIG. 2 is a flowchart illustrating a collation operation of the image collation device according to the first embodiment of the present invention.

FIG. 3 is a diagram illustrating a state in which an inspection image and a registered image are compared and collated for each collation region in the first embodiment of the present invention.

FIG. 4 is a block diagram illustrating a configuration of an image matching device according to a second embodiment of the present invention.

FIG. 5 is a flowchart illustrating a collation operation of the image collation device according to the second embodiment of the present invention.

FIG. 6 is a diagram illustrating a relationship between a parallel movement amount of an inspection image and a coincidence rate according to the second embodiment of the present invention.

FIG. 7 is an enlarged schematic diagram of a fingerprint for explaining the principle of the second embodiment of the present invention.

FIG. 8 is a flowchart illustrating a collation operation of the image collation device according to the third embodiment of the present invention.

FIG. 9 is a block diagram illustrating a configuration of an image matching device according to a fourth embodiment of the present invention.

FIG. 10 is a flowchart illustrating a collation operation of the image collation device according to the fourth embodiment of the present invention.

FIG. 11 is a diagram illustrating a relationship between a parallel movement amount of an inspection image and a coincidence rate according to the fourth embodiment of the present invention.

FIG. 12 is a block diagram illustrating a configuration of an image matching device according to a fifth embodiment of the present invention.

FIG. 13 is a flowchart illustrating a collation operation of the image collation device according to the fifth embodiment of the present invention.

FIG. 14 is a block diagram illustrating a configuration of an image matching device according to a sixth embodiment of the present invention.

FIG. 15 is a flowchart illustrating a collation operation of the image collation device according to the sixth embodiment of the present invention.

FIG. 16 is a block diagram illustrating a configuration of an image matching device according to a seventh embodiment of the present invention.

FIG. 17 is a flowchart illustrating a collation operation of the image collation device according to the seventh embodiment of the present invention.

FIG. 18 is a block diagram illustrating a configuration of an image matching device according to a tenth embodiment of the present invention.

FIG. 19 is a flowchart illustrating a collation operation of the image collation device according to the tenth embodiment of the present invention.

FIG. 20 is a block diagram illustrating a configuration of an image matching device according to an eleventh embodiment of the present invention.

FIG. 21 is a flowchart illustrating a collation operation of the image collation device according to the eleventh embodiment of the present invention.

FIG. 22 is a block diagram illustrating a configuration of an image matching device according to a twelfth embodiment of the present invention.

FIG. 23 is a block diagram showing a configuration of a conventional fingerprint matching device.

FIG. 24 is a flowchart illustrating a collation operation of the fingerprint collation device in FIG. 23;

FIG. 25 is an enlarged schematic diagram of a fingerprint for explaining a problem of the conventional fingerprint matching device.

[Explanation of symbols]

100: image input device, 200: image database, 3
00, 301, 302, 303, 304, 305, 30
6, 307 processing unit, 10, 11, 12, 13, 14
... Conversion means, 20 ... Area designation means, 30, 31, 32 ...
Collation means, 40 ... maximum matching rate extracting means, 41 ... minimum matching rate extracting means, 50, 51, 52, 53 ... calculating means, 6
0, 61, 62, 63: determination means, 70: selection means, 80: storage means, 85: reference point detection means, 90: correction amount calculation means.

 ──────────────────────────────────────────────────続 き Continuing on the front page (72) Inventor Takahiro Hanedano 2-3-1 Otemachi, Chiyoda-ku, Tokyo Within Nippon Telegraph and Telephone Corporation (72) Inventor Katsuyuki Machida 2-3-3, Otemachi, Chiyoda-ku, Tokyo No. 1 Nippon Telegraph and Telephone Corporation F term (reference) 4C038 FF01 FF05 VA07 VA20 VB04 VB40 VC05 5B043 BA02 CA02 EA12 GA01 5L096 BA15 EA15 EA16 GA19 HA07 JA03

Claims (43)

[Claims] 1. An image matching device for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first step of repeatedly executing the processing every fixed amount within the range and outputting a processed first image;
Conversion means, and for each processing of the first conversion means, the first image and the second image output from the first conversion means are converted to the first image or the second image. First matching means for comparing and matching in a plurality of matching areas obtained by dividing an image to obtain a matching rate in each matching area; and a matching rate output from the first matching means for each of the matching areas. A maximum matching rate extracting means for obtaining a maximum matching rate; a calculating means for obtaining a maximum matching rate average which is an average value of the maximum matching rates for each matching region output from the maximum matching rate extracting means; The first image and the second image when the averaged maximum matching rate average is equal to or greater than a predetermined threshold value.
An image collating apparatus comprising: a determination unit that determines that the images are the same. 2. An image matching apparatus for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first step of repeatedly executing the processing every fixed amount within the range and outputting a processed first image;
Conversion means, and for each processing of the first conversion means, the first image and the second image output from the first conversion means are converted to the first image or the second image. First matching means for comparing and matching in a plurality of matching areas obtained by dividing an image to obtain a matching rate in each matching area; and a matching rate output from the first matching means for each of the matching areas. A maximum matching rate extracting means for obtaining a maximum matching rate; a minimum matching rate extracting means for obtaining a minimum matching rate for each of the matching regions from the matching rates output from the first matching means; Calculating means for determining a difference between the maximum matching rate output from the means and the minimum matching rate output from the minimum matching rate extracting means for each of the matching regions, and calculating a matching rate difference average which is an average value of the differences; The average of the matching rate differences output from this calculating means is predicted. Wherein said first image when the above set threshold second
An image collating apparatus comprising: a determination unit that determines that the images are the same. 3. An image collating apparatus for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first step of repeatedly executing the processing every fixed amount within the range and outputting a processed first image;
Conversion means, and for each processing of the first conversion means, the first image and the second image output from the first conversion means are converted to the first image or the second image. First matching means for comparing and matching in a plurality of matching areas obtained by dividing an image to obtain a matching rate in each matching area; and a matching rate output from the first matching means for each of the matching areas. A maximum matching rate extracting means for obtaining a maximum matching rate; a minimum matching rate extracting means for obtaining a minimum matching rate for each of the matching regions from the matching rates output from the first matching means; Calculating means for calculating a quotient obtained by dividing the maximum matching rate output from the means by the minimum matching rate output from the minimum matching rate extracting means for each of the matching areas, and obtaining a matching rate quotient average which is an average value of the quotients; The quotient of the coincidence rate output from the calculating means is set in advance. Image matching system, characterized in that it comprises a a determination means is the second image and the first image when the above threshold the same that is. 4. An image matching device for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first step of repeatedly executing the processing every fixed amount within the range and outputting a processed first image;
Conversion means, and for each processing of the first conversion means, the first image and the second image output from the first conversion means are converted to the first image or the second image. First matching means for comparing and matching in a plurality of matching areas obtained by dividing an image to obtain a matching rate in each matching area; and a matching rate output from the first matching means for each of the matching areas. A maximum matching rate extracting means for obtaining a maximum matching rate; a minimum matching rate extracting means for obtaining a minimum matching rate for each of the matching regions from the matching rates output from the first matching means; The maximum matching rate average, which is the average value of the maximum matching rates for each matching area output from the means, and the minimum matching rate average, which is the average value of the minimum matching rates for each matching area, output from the minimum matching rate extracting means. The calculation means to be obtained, and the maximum value output from the calculation means The average matching rate is equal to or greater than a predetermined first threshold value, and the minimum matching rate average output from the calculating means is set to a predetermined second threshold value (first threshold ≧ second threshold value). An image collating device, comprising: a determination unit that determines that the first image and the second image are the same when the threshold value is smaller than the threshold value. 5. An image matching apparatus for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first step of repeatedly executing the processing every fixed amount within the range and outputting a processed first image;
Conversion means, and for each processing of the first conversion means, the first image and the second image output from the first conversion means are converted to the first image or the second image. First matching means for comparing and matching in a plurality of matching areas obtained by dividing an image to obtain a matching rate in each matching area; and for each of the matching areas from the matching rate output from the first matching means. A minimum matching rate extracting means for obtaining a minimum matching rate, a calculating means for obtaining a minimum matching rate average which is an average value of the minimum matching rates for each matching region output from the minimum matching rate extracting means, and an output from the calculating means. An image matching apparatus, comprising: a determination unit that determines that the first image and the second image are the same when the obtained minimum matching rate average is smaller than a preset threshold value. 6. The image matching apparatus according to claim 1, wherein a predetermined number of maximum matching rates are extracted in descending order from the maximum matching rates for each matching area output from the maximum matching rate extracting means. The image matching device further comprising a selection unit, wherein the calculation unit obtains the average of the maximum matching ratios using only the maximum matching ratio extracted by the selection unit. 7. The image matching apparatus according to claim 2, wherein a predetermined number of maximum matching rates from the maximum matching rate for each matching area output from said maximum matching rate extracting means are set in descending order. A selection unit for extracting a matching rate and extracting a predetermined number of minimum matching rates in ascending order from the minimum matching rates for each matching region output from the minimum matching rate extracting means, The means, using only the maximum match rate and the minimum match rate extracted by the selection means, calculates the match rate difference average, the match rate quotient average, the maximum match rate average, and the minimum match rate average. Characteristic image matching device. 8. The image matching apparatus according to claim 5, wherein a predetermined number of minimum matching rates are extracted in ascending order from the minimum matching rates for each matching area output from the minimum matching rate extracting means. An image matching apparatus, further comprising a selection unit, wherein the calculation unit obtains the minimum match ratio average using only the minimum match ratio extracted by the selection unit. 9. The image matching apparatus according to claim 1, wherein at least one of a translation process and a rotation process is performed on the first image at the first initial position in advance. A second conversion means for repeatedly executing the processing every fixed amount within a set range and outputting a processed first image; and a second conversion means for each processing of the second conversion means. Second matching means for comparing and matching the first image output from the second image with the second image to obtain a matching rate; and when the matching rate output from the second matching means is maximum, The first position from a first initial position to a current position;
Storage means for storing both the translation amount of the image, the rotation angle, or both the translation amount and the rotation angle, wherein the first conversion means stores the translation amount, the rotation angle or the rotation amount stored in the storage means. After setting the position obtained by adding the translation amount and the rotation angle to the first initial position as a second initial position, moving the first image to the second initial position,
An image matching apparatus, which performs at least one of a translation process and a rotation process on the first image. 10. The image collating apparatus according to claim 9, wherein the range set in advance by said first conversion means is defined by said second conversion means.
An image matching apparatus characterized in that the area is smaller than a range set in advance by the conversion means. 11. The image collating apparatus according to claim 9, wherein the second conversion means executes a translation amount, a rotation angle or a translation amount and a rotation angle per one time performed on the first image. An image collating apparatus, wherein the first conversion means executes the first image with respect to the first image in a parallel movement amount, a rotation angle, or a parallel movement amount and a rotation angle larger than each other. 12. The image matching apparatus according to claim 9, wherein the area of the matching area for which the second matching means obtains the matching rate is calculated from the total area of each matching area for which the first matching means obtains the matching rate. An image collating apparatus characterized in that the size is also reduced. 13. The image collating apparatus according to claim 1, wherein a reference point is detected to detect respective reference points of the first image and the second image at a first initial position. Means for correcting the translation amount, the rotation angle, or both the translation amount and both the rotation angle and the rotation angle of the first image necessary for matching the positions of the respective reference points of the first image and the second image. Further comprising an amount calculating means, wherein the first conversion means calculates a position obtained by adding the parallel movement amount, the rotation angle or the parallel movement amount and the rotation angle obtained by the correction amount calculating means to the first initial position. As the second initial position, after moving the first image to the second initial position, executing at least one of a translation process and a rotation process on the first image An image matching device characterized by the above-mentioned. 14. The image collating apparatus according to claim 9, wherein reference point detecting means for detecting respective reference points of the first image and the second image at a first initial position; Correction amount calculating means for calculating the parallel movement amount, the rotation angle, or both the parallel movement amount and the rotation angle of the first image necessary for matching the positions of the respective reference points of the image and the second image. Wherein the second conversion means calculates a new first initial position by adding the parallel movement amount, the rotation angle or the parallel movement amount and the rotation angle obtained by the correction amount calculation means to the first initial position. As a position, after moving the first image to the first initial position, at least one of a translation process and a rotation process is performed on the first image. Image matching device. 15. An image matching method for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A maximum matching rate extraction procedure for obtaining a maximum matching rate for each of the matching areas from the matching rates and storing the maximum matching rate in a memory; and a maximum matching rate average that is an average value of the maximum matching rates for each of the matching areas. And the maximum matching rate Image matching method characterized by equalizing executes and the determining procedure is wherein the first image the second image is the same in the case of more than a preset threshold value. 16. An image matching method for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A matching procedure for obtaining a maximum matching rate for each of the matching areas from the matching rates and storing the maximum matching rate in a memory; and obtaining a minimum matching rate for each of the matching areas from the matching rates. Extracting the minimum match rate stored in the memory; and the maximum match rate A calculation procedure for obtaining a difference from the minimum matching rate for each of the matching regions, obtaining an average of the matching rate differences which is an average value of the differences, and storing the average in a memory; In the above case, a determination procedure for determining that the first image and the second image are the same is executed. 17. An image matching method for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A matching procedure for obtaining a maximum matching rate for each of the matching areas from the matching rates and storing the maximum matching rate in a memory; and obtaining a minimum matching rate for each of the matching areas from the matching rates. Extracting the minimum match rate stored in the memory; and the maximum match rate A calculation procedure of obtaining a quotient divided by the minimum matching rate for each of the matching areas, obtaining a matching rate quotient average which is an average value of the quotients, and storing the average in a memory; A determining step of determining that the first image and the second image are the same when the value is equal to or more than the value. 18. An image matching method for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A matching procedure for obtaining a maximum matching rate for each of the matching areas from the matching rates and storing the maximum matching rate in a memory; and obtaining a minimum matching rate for each of the matching areas from the matching rates. Extracting the minimum matching rate stored in the memory; Calculating a maximum matching rate average, which is an average value of the maximum matching rates, and a minimum matching rate average, which is an average value of the minimum matching rates for each of the matching regions, and storing the calculated average in the memory; If the average of the minimum coincidence rates is smaller than a predetermined second threshold (first threshold ≧ second threshold), the first An image collating method, wherein the image collating method and the second image are determined to be the same. 19. An image matching method for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A matching procedure for extracting a minimum matching rate from the matching rates for each of the matching areas and storing the minimum matching rate in a memory; and a minimum matching rate average that is an average value of the minimum matching rates for the matching areas. And the minimum matching rate Image matching method characterized by equalizing executes and the determining steps are the same the first image and the second image is smaller than a preset threshold. 20. The image matching method according to claim 15, wherein, before the calculation procedure, a predetermined number of the maximum matching rates from the maximum matching rates for each matching area obtained in the maximum matching rate extracting procedure are selected. Executing a selection procedure of extracting a maximum matching rate and storing the extracted maximum matching rate in a memory; and wherein the calculating step obtains the average of the maximum matching rates using only the maximum matching rate extracted by the selection procedure. Image matching method. 21. The image matching method according to claim 16, wherein before the calculation procedure, the largest matching rate for each matching region obtained in the maximum matching rate extracting procedure is in descending order. A predetermined number of maximum matching rates are extracted and stored in the memory, and a predetermined number of minimum matching rates are set in ascending order from the minimum matching rates for each matching area obtained in the minimum matching rate extracting procedure. Is executed, and the calculation procedure is performed by using only the maximum matching rate and the minimum matching rate extracted by the selecting procedure, and the average of the matching rate difference, the matching rate quotient is used. An image matching method, wherein an average, the maximum matching rate average, and the minimum matching rate average are obtained. 22. The image matching method according to claim 19, wherein, prior to the calculation procedure, a predetermined number of the minimum matching rates from the minimum matching rates for each matching area obtained in the minimum matching rate extracting procedure are selected. Executing a selection procedure of extracting a minimum matching rate and storing the extracted minimum matching rate in a memory, wherein the calculating step calculates the minimum matching rate average using only the minimum matching rate extracted by the selection procedure. Image matching method. 23. The image matching method according to claim 15, wherein prior to each step, the first image at the first initial position is selected from a translation process and a rotation process. A second conversion procedure of repeatedly executing at least one process within a predetermined range for each fixed amount, and storing the processed first image in a memory; and for each processing of the second conversion procedure. A second collation procedure for comparing and collating the processed first image and the second image to obtain a coincidence rate and storing the same in a memory; And storing a translation amount, a rotation angle, or both the translation amount and the rotation angle of the first image from the initial position to the current position of the first image in a memory. The translation amount, rotation angle or translation stored in the storage procedure The position where the amount and rotational angle was added to the first initial position as the second initial position, after moving the first image to the second initial position,
An image matching method, wherein at least one of a translation process and a rotation process is performed on the first image. 24. The image matching method according to claim 23, wherein the range preset in the first conversion procedure is the second range.
An image matching method characterized in that the range is smaller than a preset range in the conversion procedure. 25. The image matching method according to claim 23, wherein the translation amount, the rotation angle, or the translation amount and the rotation angle per one time performed on the first image in the second conversion procedure are determined. An image collating method, wherein the amount of translation and the angle of rotation, or the amount of translation and rotation, each time performed on the first image in the first conversion procedure are larger. 26. The image matching method according to claim 23, wherein the area of the matching area for which the matching rate is determined in the second matching procedure is calculated from the total area of each matching area for which the matching rate is determined in the first matching procedure. An image matching method characterized by reducing the size of the image. 27. The image matching method according to claim 15, wherein before each step, a reference point of each of the first image and the second image at a first initial position. A reference point detection procedure for detecting and storing the reference point in a memory; and a translation amount, a rotation angle, or a translation amount of the first image necessary for matching positions of the respective reference points of the first image and the second image. And a correction amount calculation procedure of obtaining both the translation amount and the rotation angle and storing the same in a memory. The first conversion procedure includes the translation amount, the rotation angle, or the translation amount obtained in the correction amount calculation procedure. A position obtained by adding the amount and the rotation angle to the first initial position is set as a second initial position, and the first image is moved to the second initial position. Execute at least one of a translation process and a rotation process An image matching method characterized in that: 28. The image matching method according to claim 23, wherein before each step, a reference point of each of the first image and the second image at a first initial position is detected and stored in a memory. A reference point detection procedure to be stored; a translation amount, a rotation angle or a translation amount and a rotation angle of the first image necessary for matching positions of the respective reference points of the first image and the second image. And a correction amount calculation procedure for obtaining and storing the same in a memory. The second conversion procedure includes calculating the translation amount, the rotation angle, or the translation amount and the rotation angle obtained in the correction amount calculation procedure. The position added to the first initial position is set as a new first initial position, and the first image is moved to the first initial position. Executing at least one of the rotation processes Image matching method. 29. An image matching program for determining whether the first image and the second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A maximum matching rate extraction procedure for obtaining a maximum matching rate for each of the matching areas from the matching rates and storing the maximum matching rate in a memory; and a maximum matching rate average that is an average value of the maximum matching rates for each of the matching areas. Calculation procedure to find in memory and Image collation program, characterized in that 致率 average to execute determining the determining step that the second image and the first image is the same in the case of more than a preset threshold value to the computer. 30. An image matching program for determining whether the first image and the second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing in a fixed amount within the range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A matching procedure for obtaining a maximum matching rate for each of the matching areas from the matching rates and storing the maximum matching rate in a memory; and obtaining a minimum matching rate for each of the matching areas from the matching rates. Extracting the minimum match rate stored in the memory; A calculation procedure for obtaining a difference between the matching rate and the minimum matching rate for each of the matching areas, obtaining an average of matching rate differences as an average value of the differences, and storing the average in a memory; An image matching program for causing a computer to execute a determination procedure for determining that the first image and the second image are the same when the threshold value is equal to or greater than a threshold value. 31. An image matching program for determining whether the first image and the second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A matching procedure for obtaining a maximum matching rate for each of the matching areas from the matching rates and storing the maximum matching rate in a memory; and obtaining a minimum matching rate for each of the matching areas from the matching rates. Extracting the minimum match rate stored in the memory; A quotient obtained by dividing the matching rate by the minimum matching rate is obtained for each of the matching regions, a matching rate quotient average, which is an average value of the quotients, is calculated and stored in a memory. An image comparison program for causing a computer to execute a determination procedure for determining that the first image and the second image are the same when the threshold value is equal to or greater than the threshold value. 32. An image comparison program for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A matching procedure for obtaining a maximum matching rate for each of the matching areas from the matching rates and storing the maximum matching rate in a memory; and obtaining a minimum matching rate for each of the matching areas from the matching rates. Extracting the minimum matching rate stored in the memory; Calculating a maximum matching rate average, which is an average value of the maximum matching rates for each region, and a minimum matching rate average, which is an average value of the minimum matching rates for each of the matching areas, and storing the calculated average in the memory Is greater than or equal to a preset first threshold value and the minimum match rate average is smaller than a preset second threshold value (first threshold value ≧ second threshold value). An image matching program for causing a computer to execute a determination procedure for determining that the first image and the second image are the same. 33. An image comparison program for determining whether a first image and a second image are the same, wherein at least one of a translation process and a rotation process is preset for the first image. A first conversion procedure of repeatedly executing the first image after processing within a range and storing the processed first image in a memory; and a first conversion procedure for each processing of the first conversion procedure. An image and the second image are compared and collated in a plurality of collation regions obtained by dividing the first image or the second image, and a match rate in each collation region is obtained and stored in a memory. A matching procedure for extracting a minimum matching rate from the matching rates for each of the matching areas and storing the minimum matching rate in a memory; and a minimum matching rate average that is an average value of the minimum matching rates for the matching areas. The calculation procedure for finding and storing in memory Image collation program, characterized in that 致率 average to execute and the determining steps are the same the first image and the second image is smaller than the preset threshold the computer. 34. The image matching program according to claim 29, wherein before the calculation procedure, a predetermined number of maximum matching rates from the maximum matching rates for each matching area obtained in the maximum matching rate extracting procedure are selected. Causing the computer to execute a selection procedure for extracting a maximum matching rate and storing the extracted maximum matching rate in a memory, wherein the calculating step obtains the maximum matching rate average using only the maximum matching rate extracted by the selection procedure. Image matching program. 35. The image matching program according to claim 30, wherein before the calculation procedure, the largest matching rate for each matching area obtained by the maximum matching rate extracting procedure is set in descending order. A predetermined number of maximum matching rates are extracted and stored in the memory, and a predetermined number of minimum matching rates are set in ascending order from the minimum matching rates for each matching area obtained in the minimum matching rate extracting procedure. And causing the computer to execute a selection procedure of extracting and storing the same in a memory. The calculation procedure uses only the maximum matching rate and the minimum matching rate extracted by the selection procedure, and calculates the matching rate difference average and the matching rate. An image matching program for calculating a rate quotient average, the maximum matching rate average, and the minimum matching rate average. 36. The image matching program according to claim 33, wherein before the calculation procedure, a predetermined number of the minimum matching rates from the minimum matching rates for each matching area obtained in the minimum matching rate extracting procedure are selected. Causing the computer to execute a selection procedure for extracting a minimum matching rate and storing the extracted minimum matching rate in a memory; and wherein the calculating step obtains the minimum matching rate average using only the minimum matching rate extracted by the selection procedure. Image matching program. 37. The image matching program according to claim 29, wherein before each step, the first image at the first initial position is selected from a translation process and a rotation process. A second conversion procedure of repeatedly executing at least one process within a predetermined range for each fixed amount, and storing the processed first image in a memory; and for each processing of the second conversion procedure. A second collation procedure for comparing and collating the processed first image and the second image to obtain a coincidence rate and storing the same in a memory; Storing the parallel movement amount, the rotation angle, or both the parallel movement amount and the rotation angle of the first image from the initial position to the current position of the first image in a memory, and causing the computer to execute the first conversion procedure. , The translation stored in the storage procedure , The position where the rotational angle, or translation amount and rotational angle was added to the first initial position as the second initial position, after moving the first image to the second initial position,
An image matching program for executing at least one of a translation process and a rotation process on the first image. 38. The image collation program according to claim 37, wherein the range preset in the first conversion procedure is the second range.
An image matching program characterized in that the size is smaller than a preset range in the conversion procedure. 39. The image collating program according to claim 37, wherein the translation amount, the rotation angle, or the translation amount and the rotation angle per one time performed on the first image in the second conversion procedure are determined. An image collating program, wherein the amount of translation and the angle of rotation or the amount of translation and rotation at one time executed on the first image in the first conversion procedure are larger than the amount of translation and rotation. 40. The image matching program according to claim 37, wherein the area of the matching area for which the matching rate is determined in the second matching procedure is calculated from the total area of the matching areas for which the matching rate is determined in the first matching procedure. An image collation program characterized by reducing the size of the image. 41. The image matching program according to claim 29, wherein, before each step, a reference point of each of the first image and the second image at a first initial position. A reference point detection procedure for detecting and storing the reference point in a memory; and a translation amount, a rotation angle, or a translation amount of the first image necessary for matching positions of the respective reference points of the first image and the second image. And a computer that executes a correction amount calculation procedure for obtaining both the translation amount and the rotation angle and storing the same in a memory. The first conversion procedure includes the translation amount, rotation angle, or rotation amount obtained in the correction amount calculation procedure. A position obtained by adding the parallel movement amount and the rotation angle to the first initial position is set as a second initial position, and the first image is moved to the second initial position. In contrast, at least one of the translation processing and the rotation processing An image comparison program for executing one process. 42. The image collating program according to claim 37, wherein before each step, a reference point of each of the first image and the second image at a first initial position is detected and stored in a memory. A reference point detection procedure to be stored; a translation amount, a rotation angle or a translation amount and a rotation angle of the first image necessary for matching positions of the respective reference points of the first image and the second image. And a correction amount calculation procedure for obtaining and storing both in a memory. The second conversion procedure includes: a translation amount, a rotation angle or a translation amount and a rotation angle obtained in the correction amount calculation procedure. Is added to the first initial position as a new first initial position, the first image is moved to the first initial position, and then translated with respect to the first image. At least one of processing and rotation processing An image matching program characterized by executing a process. 43. A recording medium on which the image collation program according to claim 29 is recorded.