JPH113423A - Image collator, collating method and computer readable collate program recording medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image collator and an image collating method which have high identification power and a high collation rate used for fingerprints, etc. SOLUTION: An image collator 20 is provided with 1st and 2nd image correlation operating parts 21 and 22 which perform two kinds of different image operations to two images, 1st and 2nd similarity decision parts 23 and 24 which determine the similarity of the images based on each of the image operation results and a similarity outputting part 25 that is connected to them. A correlation value that is acquired by the 1st image correlation operating part 21 is compared with two different thresholds which are preliminarily set by the 1st similarity deciding part 23, and when it is between both thresholds, a 2nd correlation value is further calculated by the 2nd image correlation operating part 22, and the 2nd similarity decision part 24 makes a decision. Thus, the part 25 performs more accurate similarity decision.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a device for comparing images by judging the degree of similarity between two images, and more particularly to an image for identifying a person by comparing images such as a fingerprint and a retina. The present invention relates to a matching device, an image matching method, and an image matching by a computer.

[0002]

2. Description of the Related Art A person collating apparatus for identifying an individual by collating images such as a fingerprint and a retina is widely used for entry / exit management to an important area or the like. These apparatuses compare an image pattern of a reference image registered in advance with an image pattern of an input collation image, and determine whether or not the user is the person based on whether or not the images are the same.

As a pattern matching technique for matching image patterns, there is a method using an image correlation operation. As such an image correlation operation, there is a method using matched filtering, but there is a disadvantage that discriminating power is low. On the other hand, there is a Fourier phase correlation method disclosed by Kawata et al. In "Image Data Processing for Scientific Measurements" (CQ Publishing Co.), and an example of applying this technique to fingerprint collation is Nakajima et al. Principle of the Method and Its Application to Fingerprint Verification ”
Times image sensing symposium). The Fourier phase correlation method has a high discriminating power because it performs discrimination by emphasizing the feature of the shape of the image pattern as compared with the matched filtering method.

[0004]

However, even in the above-mentioned fingerprint collation apparatus using the Fourier phase correlation method, the collation rate is 93.1%, and the discrimination power is still insufficient as security identification means. Was. This is because the peak of the correlation value indicating the similarity obtained by the correlation operation can be small for the person due to rotation or distortion of the image itself, and even if another person is similar to a part of the image, this peak is high. It depends on things that may come out. In other words, there is an overlap in the distribution of the correlation peaks, which lowers the matching rate, so that a sufficient matching rate cannot be achieved.

It is an object of the present invention to provide an image matching apparatus and an image matching method which have a high discriminating power and a high matching rate used for fingerprint matching and the like.

[0006]

An image collating apparatus according to the present invention is an image collating apparatus for judging the similarity between two images, wherein (1) a first signal between signals representing two inputted images. A first image correlation operation unit for performing an image correlation operation and outputting a first correlation value; and (2) comparing the first correlation value with two thresholds having different preset values, and If the correlation value of 1 is smaller than both of the two thresholds, it is determined that the two images are different. If the first correlation value is larger than both of the two thresholds, the two images are the same. A first similarity determination unit that outputs a determination result when the first correlation value is a value between two threshold values and outputs a permission signal that permits a subsequent comparison process; And (3) a second image phase different from the first image correlation operation between the signals representing the two images when the permission signal is output. A second image correlation operation unit that performs an operation and outputs a second correlation value, and (4) a second image correlation operation unit that determines a similarity between the two images based on the second correlation value and outputs a determination result. And (5) a similarity output unit that outputs the similarity between the two images based on the determination result output of the first or second similarity determination unit. Features.

According to this, by comparing the first correlation value obtained by the first image correlation calculation with two preset threshold values, the correlation value is larger than any of the threshold values. That is, when the similarity is sufficiently high, the two images are determined to be the same, and when the correlation value is smaller than any of the thresholds, that is, when the similarity is sufficiently low, the two images are determined to be different. . If the correlation value is between the two threshold values, the processing by the second image correlation calculation unit and the similarity determination unit is further performed. The second image correlation operation is the first image correlation operation.
Therefore, a second correlation value different from the first correlation value is obtained. That is, the second image correlation operation corresponds to performing image comparison using an index different from that of the first image correlation operation. Therefore, the second correlation value may indicate a value sufficient to perform the similarity determination even between the images for which the similarity determination is difficult with the first correlation value.

On the other hand, the image matching method according to the present invention is an image matching method for judging the similarity between two images. (1) A first image correlation operation is performed between two images to obtain a first correlation. And (2) comparing the first correlation value with two thresholds having different preset values so that the first correlation value is equal to both of the two thresholds. If less,
The two images are determined to be dissimilar, and if the first correlation value is greater than both of the two threshold values, the two images are determined to be the same and the first correlation value is between the two threshold values. When the first similarity determination step permits the subsequent comparison processing, and (3) when the first similarity determination step permits the subsequent comparison processing, A second image correlation operation that performs a second image correlation operation different from the first image correlation operation and outputs a second correlation value; and (4) a second image correlation operation based on the second correlation value. A second similarity determination step of determining similarity, and (5) a similarity output step of outputting a similarity between two images based on a determination result in the first or second similarity determination step; It is characterized by having.

According to this, when it is difficult to determine similarity with the first correlation value obtained by the first image correlation operation, a second image correlation operation different from the first image correlation operation is performed. , A second correlation value different from the first correlation value is obtained, and image similarity determination is performed using the second correlation value.

A computer-readable recording medium storing an image collation program to be executed by a computer according to the present invention includes: (1) performing a first image correlation operation between two pieces of input image information, And (2) comparing the first correlation value with two thresholds having different preset values, and when the first correlation value is smaller than the first threshold on the smaller side. Is determined that the two images are dissimilar, and if the first correlation value is larger than the second threshold value on the larger side, it is determined that the two images are the same and a determination result is output. If the correlation value of 1 is a value between the two threshold values, the procedure proceeds to a subsequent comparison process; and (3) a second image difference between the two pieces of image information, which is different from the first image correlation operation. A step of performing image correlation calculation to obtain a second correlation value; (4)
Comparing the second correlation value with a preset third threshold to determine the similarity between the two images and output the determination result. It is characterized by being.

By causing a computer to read the image matching program stored in the recording medium, a computer program for executing the above-described image matching method is introduced into the computer, and the above-described image matching can be performed by the computer.

Further, in each of the image matching apparatus, the image matching method, and the image matching program, the first image correlation operation is a Fourier phase correlation operation, and the second image correlation operation is a Fourier amplitude correlation operation. Good.

In this case, the first correlation value indicates the similarity based on the characteristic of the shape of the image obtained by the Fourier phase correlation operation, and the second correlation value indicates the characteristic of the intensity of the image obtained by the Fourier amplitude correlation operation. The similarity based on is shown.

[0014]

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will be described below with reference to the drawings. In the drawings, the same components are denoted by the same reference numerals, and redundant description will be omitted. In the following description, as a specific embodiment of the present invention, a fingerprint matching device that matches a fingerprint to confirm a person will be described as an example.

FIG. 1 is a system configuration diagram of the image matching means of the present invention, and FIG. 2 is a block configuration diagram of a fingerprint matching device including the image matching means of FIG.

First, the system configuration will be described with reference to FIG. Input image signals constituting two images perform first and second image correlation operation units 2 that perform different image correlation operations between these image signals and output correlation values.
1 and 22 are input. The first image correlation operation is a Fourier phase correlation operation, and the second image correlation operation is a Fourier amplitude correlation operation. The details of each operation will be described later. Each of these image correlation calculators 21 and 22 includes a corresponding first and second similarity determiner 23,
24, and determines and outputs a similarity based on the correlation value. These similarity determination units 23 and 24 are connected to a similarity output unit 25 that finally outputs a similarity. Further, the first similarity determination unit 23 and the second image correlation calculation unit 2
2 is connected, and a permission signal for permitting the processing of the second image correlation calculation unit 22 under certain conditions described later is sent from the first similarity determination unit 23 to the second image correlation calculation unit 22. .

Next, referring to FIG.
The configuration of the fingerprint collation device 100 using the above will be described. The fingerprint matching device 100 includes an imaging unit 50 for inputting a fingerprint image.
And an operation processing unit 60 for performing image comparison, a display unit 6 for displaying processing results, and a panel unit 7 including switches and the like for performing various settings and operations.

Here, the image pickup unit 50 is disposed so as to sandwich a fingertip provided with a fingerprint, and the LED 2 is a light emitting element serving as a light source for irradiating a fingerprint image, and one end is an input end face on which a finger is disposed. At the same time, a fiber optic plate (FOP) 3 and FOP 3 constructed by bundling a number of optical fibers transmitting a fingerprint image illuminated by the LED 2
And a CCD 4 that is a two-dimensional image sensor that converts a fingerprint image transmitted by the FOP 3 into a corresponding electric signal using two-dimensionally arranged pixels.

On the other hand, the arithmetic processing unit 60 includes the frame memory 5 for digitizing the output image signal of the CCD 4 for one image and accumulating and storing the image signal, the above-mentioned image collating means 20, and performs actual arithmetic processing. CPU that controls the operation of
8, a RAM 9 for storing a reference image used for comparison with the input image during the comparison process, and a hard disk 10 for storing and recording the reference image. As the CPU 8, it is preferable to use an arithmetic device having a high arithmetic capability such as a digital signal processor (DSP).

Next, the operation of the fingerprint collation apparatus 100 will be described with reference to FIGS. FIG. 3 is a processing flow chart when registering a reference image of the fingerprint matching apparatus 100, and FIG. 4 is a processing flow chart when matching.

First, for fingerprint collation, it is necessary to register a fingerprint to be referred to at the time of collation (hereinafter referred to as a reference fingerprint) in advance. This registration operation will be described with reference to FIG. First, the user operates the switches on the panel 7 shown in FIG. 2 to set the registration mode, set the ID number (S1).
0), a finger is placed on the input end face of FOP3, and a switch on panel 7 is operated to start fingerprint reading. Then, the LEDs 2a and 2b are turned on to illuminate the fingerprint image of the fingertip, and the illuminated fingerprint image is sent to the CCD via the FOP3.
4 is incident. The CCD 4 converts the incident fingerprint image into a corresponding two-dimensional matrix of electric signals and outputs the electric signals (S
11). This image is temporarily stored in the frame memory 5 (S12). The CPU 8 reads out the image electric signal stored in the frame memory 5 and performs a fast Fourier transform (FFT) (S13). The result of this calculation is given as matrix RE
Expressed as F,

[0022]

(Equation 1) It is represented by Here, A _ref and B _ref are matrix values representing the real part and the imaginary part of the Fourier transform result of the reference fingerprint image, respectively. The calculation result is stored in the HD 10 in association with the ID number (S14).

Next, the collating operation will be described with reference to FIG. At the time of collation, the switch group on the panel 7 is operated to set the collation mode, and the ID number is set (S2
0), a finger is placed on the input end face of FOP3, and a switch on panel 7 is operated to start fingerprint reading. Then, similarly to the above-mentioned reference mode, C is transmitted via FOP3.
The fingerprint image is input to the CD 4, and the CCD 4 converts the incident fingerprint image into a corresponding electric signal and outputs it (S21).
This image is temporarily stored in the frame memory 5 (S
22). The CPU 8 reads out the image electric signal stored in the frame memory 5 and performs a fast Fourier transform (FFT).
Is performed (S23). When this calculation result is represented by a matrix TAG,
Similar to equation (1),

[0024]

(Equation 2) It is represented by Here, A _tag and B _tag are matrices representing the real part and the imaginary part of the matrix TAG. This calculation result TAG is C
The data is output from the PU 8 to the RAM 9 and stored in a predetermined position (S24).

On the other hand, the CPU 8 reads out the Fourier transform result of the fingerprint image for reference based on the ID number input from the HD 10 and stores the result in the RAM 9 at a position different from the storage position of the above-described calculation result TAG. (S25).

Next, the first image correlation calculator 2 of the CPU 8
1 (see FIG. 1) performs a Fourier phase correlation operation, which is a first image correlation operation, between the Fourier transform results of the fingerprint image for collation and reference.

First, a phase limiting process for extracting only phase information is performed by limiting the amplitude of each signal in each matrix of each Fourier calculation result to 1 (S26). The resulting phase information images REF _po and TAG _po are represented as follows.

[0028]

(Equation 3)

Subsequently, normalization processing of the power spectrum of each image information is performed (S27). This is a process for keeping the sum of the power spectrum at a constant value in order to compare the maximum value of the correlation peak finally obtained by the image correlation operation under a constant condition without depending on the image. Normally, the amplitude of each signal is set to 1
Therefore, the sum of the power spectrum is equal to the number of terms in the matrix, but when additional processing is added, the sum of the power spectrum is not constant depending on the image. It is preferred to do so. The sum REF _sum and TAG _sum of the power spectrum of each phase information image REF _po and TAG _po are represented as follows.

[0030]

(Equation 4) Therefore, the standardized phase image information RE
F " _po and TAG" _po are represented as follows.

[0031]

(Equation 5)

Subsequently, these standardized phase image information
REF " _po and TAG" _po are complex-combined to obtain a combined image G (S
28). The composite image G is represented as follows.

[0033]

(Equation 6)

By inversely Fourier-transforming this G,
A phase-only correlation is obtained (S29). This corresponds to the first correlation value. This is an index indicating the similarity of the shape between the reference image and the matching image. The more similar the image shape, the higher the correlation peak height, and the more dissimilar the image shape, the higher the correlation peak height. Becomes smaller.

Next, based on the first correlation value, the CPU
The first similarity determination unit 23 in FIG. 8 (see FIG. 1) determines the similarity between the reference image and the comparison image. The first mentioned above
From the characteristics of the correlation value of the above, the height when the height of the correlation peak indicates that the images are similar within an allowable range is set as the first threshold value, The height at which there is a certain value is set in advance as a second threshold value. At this time, the first threshold value is larger than the second threshold value. Therefore, the first similarity determination unit 23
Compares the height of the correlation peak with the first and second thresholds (S30). When the value exceeds the first threshold value, the matching image is determined to be the same as the reference image, and when the value is less than the second threshold value, the matching image is determined to be different from the reference image. When the height of the correlation peak is between the first and second thresholds, it is difficult to determine the similarity, and the second image correlation based on the second image correlation calculation is further performed.
Is determined. For this purpose, the permission signal for performing the second comparison process is transmitted to the second image
Emit against. The permission signal can be represented, for example, by turning on / off a flag in a register indicating whether or not the processing of the second image correlation operation unit 22 is possible.

Next, the second image correlation calculation and the second judgment processing will be described. When the permission signal is on, the second image correlation calculator 22 shown in FIG.
The second correlation value is obtained by the image correlation calculation of (2). This second
The Fourier amplitude correlation operation is used for the image correlation operation of (1). In this operation, first, as described above, the Fourier transform results REF and TAG stored in the RAM 9 are read out, and the k matrixes f1 to fk and g1
Divide into ~ gk. Then, for each region, the sum totals I _fi and I _gi of the amplitudes of the Fourier transform result are as follows.
(1 ≦ i ≦ k) is obtained (S31).

[0037]

(Equation 7)

Then, the absolute value of the difference between the intensities I _fi and I _gi of each region is integrated for all the regions (S32).

[0039]

(Equation 8)

The division into regions has the effect of canceling out the effect of a shift or rotation in the image obtained as a result of the Fourier transform as compared with the case where the intensity difference is calculated for all pixels, so that the comparison can be performed reliably. . The distribution of the amplitude of the Fourier transform result, the so-called Fourier amplitude plane distribution, is similar for the same image and naturally different for different images. Therefore, the result of equation (12) indicates a small value for the same image and a large value for different images, and is an index indicating the similarity of the images. This is based on the amplitude information of the Fourier-transformed image, which has been neglected by homogenization in the above-described Fourier phase correlation calculation, so that similarity determination can be performed even between images for which similarity determination is difficult only by phase correlation. It can be an index to perform. This is the second correlation value.

In the second similarity determination section 24 (see FIG. 1) in the CPU 8, the second correlation value is set to a predetermined third correlation value.
(S33), and when the second correlation value exceeds the third threshold value, it is determined that the matching image is dissimilar to the reference image, and the third threshold value is determined. If it is less than the value, it is determined that they are the same.

When the permission signal is off, the similarity output section 25 in the CPU 8 turns on the permission signal based on the determination output of the first similarity determination section 23 when the permission signal is off. At this time, based on the determination output of the second similarity determination unit 24, when the reference image and the collation image are the same,
It is determined that the person is the person (S34), and if the persons are not similar, it is determined that the person is another person (S35), and signals corresponding to the respective persons are output.

Based on this output signal, the CPU 8 displays the judgment result on the display unit 6 and outputs a signal according to the judgment result to the outside. In response to this signal, it is possible to unlock a door for entering a predetermined area, permit use of a specific device, and the like.

The inventor of the present application performed a collation test for examining the recognition rate of the fingerprint collation device of the present embodiment. The results of the collation test are described below.

In the collation test, when a fingerprint image identical to the reference fingerprint image is input using a fingerprint image collected from 500 persons, that is, when the user is the person, the personal recognition rate and the personal recognition rate are correctly recognized. In addition, when a fingerprint image for verification different from the reference fingerprint image is input, that is, when the user is another person, both the exclusion rate of others and the others are correctly examined. Here, a fingerprint image input by a 256 × 256 pixel CCD is used for this collation test, and the area division in the Fourier amplitude correlation processing is performed in the circumferential direction from the center in six radial directions as shown in FIG. Forty-eight divided areas divided into eight parts were used.

FIG. 6 shows a comparison test result when only the Fourier phase correlation processing is used, and FIG.
FIG. Here, when the correlation value is equal to or more than the predetermined threshold value, it is determined that the user is the person, and when the correlation value is less than the predetermined threshold value, the person is determined to be another person. The horizontal axis in the figure is
The set threshold is shown, and the vertical axis indicates the personal recognition rate and the exclusion rate for the threshold. The higher the threshold is set, the lower the identity recognition rate is, while the other person exclusion rate is improved. The identity recognition rate and the exclusion rate are the same when the recognition rate (exclusion rate) is 98.25% and the threshold value (correlation value) at this time is 0.068.

Next, the result of the collation test of this embodiment will be described. From the results of FIG. 6, two values of 0.070 and 0.066 are used as the first and second threshold values of the phase correlation operation, respectively. In the above cases, the person was determined to be the person. When the correlation value is larger than 0.066 and smaller than 0.070, Fourier amplitude correlation calculation is performed. If the correlation value is smaller than the third threshold value I, it is determined that the person is the person, and if the correlation value is larger, the person is determined to be another person. FIG. 7 shows the relationship between the numerical value of the index I and the recognition rate at this time. The larger the index I, the higher the identity recognition rate, and conversely, the smaller the index I, the higher the others exclusion rate.

FIG. 8 is a graph comparing the relationship between the identity recognition rate and the other person exclusion rate in each case. In the case of only Fourier phase correlation processing, the threshold value of the correlation peak is set to 0.066.
6B, the identity recognition rate is 98.5 from FIG.
%, The other person exclusion rate is 97.5%. On the other hand, in the present embodiment, the index I, which is the third threshold value, is 840 from FIG.
By doing so, the other person exclusion rate can be improved to 98.2% while maintaining the personal recognition rate of 98.5%. Also, the case where the identity recognition rate is the same as the other person exclusion rate is 98.4%, which is higher than the case where only the Fourier phase correlation processing is performed.

Accordingly, it can be seen that by selecting an appropriate threshold, both the identity recognition rate and the exclusion rate can be improved.

In the above description, the embodiment applied to the fingerprint collating apparatus has been described. However, the image to be collated is not limited to the fingerprint image, but may be used for other person confirmation means such as a retinal pattern or a factory production line. It can be used for various image recognition such as positioning and defective product checking.

Although an example in which a specialized device is used for image collation has been described here, the image collating means of the present invention is realized by a computer program using a personal computer to which an image input device, a display device, and the like are connected. You may. In this case, the program is provided on a recording medium on which the program is recorded.

[0052]

As described above, according to the image collating apparatus or the collating method of the present invention, the two image signals are reliably imaged by the first correlation value obtained by the first image correlation operation. If the same or dissimilarity cannot be determined,
A second image correlation operation different from the image correlation operation of the above is performed, and the similarity between the images can be determined by obtaining a second correlation value that can reliably determine the sameness or dissimilarity of the images. The discriminating power of the image is improved as compared with the case where the correlation calculation is performed.

On the other hand, the image collating method of the present invention can be realized as a computer program, and this program is provided by a recording medium recording the program. Therefore, implementation on a personal computer or the like is easy.

Further, if a Fourier phase correlation operation is performed as the first image correlation operation and a Fourier amplitude correlation operation is performed as the second image correlation operation, a part of the Fourier phase correlation operation having a high discriminating power, for which it is difficult to determine the similarity, is obtained. Even between similar images, it is possible to determine non-similar images by Fourier amplitude correlation calculation.

[Brief description of the drawings]

FIG. 1 is a system configuration diagram of an image matching unit of the present invention.

FIG. 2 is a block diagram of a fingerprint matching device including the image matching unit of FIG. 1;

FIG. 3 is an operation flowchart of the apparatus of FIG. 2 when registering a reference fingerprint.

FIG. 4 is an operation flowchart at the time of fingerprint collation of the apparatus of FIG. 2;

FIG. 5 is a region division diagram in Fourier amplitude correlation processing used in a collation test.

FIG. 6 is a diagram showing a fingerprint collation test result when only Fourier phase correlation processing is used.

FIG. 7 is a diagram showing a fingerprint collation test result of the apparatus of FIG. 2;

8 is a graph comparing the relationship between the personal recognition rate and the exclusion rate of others in the test results shown in FIGS. 6 and 7. FIG.

[Explanation of symbols]

2 LED, 3 FOP, 4 CCD, 5 frame memory, 6 display unit, 7 panel unit, 8 CPU, 9 R
AM, 10 HD, 20 image collating means, 21 first image correlation operation unit, 22 second image correlation operation unit, 23
First similarity determination unit, 24... Second similarity determination unit, 25
... Similarity output unit, 50 ... Imaging system, 60 ... Calculation processing unit, 1
00: Fingerprint collation device.

Claims (6)

[Claims] An image matching apparatus for determining the similarity between two images, wherein a first image correlation operation is performed between input signals representing the two images, and a first correlation value is output. Comparing the first correlation value with two predetermined thresholds having different values, and if the first correlation value is smaller than both of the two thresholds, The two images are determined to be dissimilar, and if the first correlation value is greater than both of the two thresholds, the two images are determined to be the same and a determination result is output; When the correlation value of 1 is a value between the two threshold values, a first similarity determination unit that outputs a permission signal for permitting a subsequent comparison process, and when the permission signal is output, , A second image phase different from the first image correlation operation between signals representing the two images A second image correlation operation unit that performs a correlation operation and outputs a second correlation value, and a second image that determines similarity between the two images based on the second correlation value and outputs a determination result And a similarity output unit that outputs a similarity between the two images based on a determination result output of the first or second similarity determination unit. Image collating device. 2. The apparatus according to claim 1, wherein the first image correlation operation is a Fourier phase correlation operation, and the second image correlation operation is a Fourier amplitude correlation operation. 3. An image matching method for determining a similarity between two images, comprising: performing a first image correlation operation between the two images to obtain a first correlation value; The first correlation value is compared with two threshold values having different preset values, and if the first correlation value is smaller than the first threshold value on the smaller side, the two images are dissimilar. When the first correlation value is larger than the second threshold value on the larger side, it is determined that the two images are the same, and the first correlation value is equal to the two threshold values. A first similarity determination step for permitting a subsequent comparison process, and a first similarity determination step for permitting a subsequent comparison process when the first similarity determination step permits a subsequent comparison process. A second image correlation operation different from the first image correlation operation is performed to obtain a second correlation value. An image correlation calculation step, a second similarity determination step of determining a similarity between the two images based on the second correlation value, and a determination result in the first or second similarity determination step. And a similarity output step of outputting a similarity between the two images based on the image matching method. 4. The method according to claim 3, wherein said first image correlation operation is a Fourier phase correlation operation, and said second image correlation operation is a Fourier amplitude correlation operation. 5. A procedure for performing a first image correlation operation between two pieces of input image information to obtain a first correlation value, and two thresholds having different first correlation values from each other. If the first correlation value is smaller than the first threshold value on the smaller side, the two images are determined to be dissimilar, and the second image on the side with the larger first correlation value is compared with the second image. If the value is larger than the threshold value, the two images are determined to be the same and a determination result is output. If the first correlation value is a value between the two threshold values, the subsequent comparison is performed. A procedure for shifting to processing, a procedure for performing a second image correlation operation different from the first image correlation operation between the two pieces of image information to obtain a second correlation value, The similarity between the two images is determined by comparing with a predetermined third threshold, and the determination result is obtained. And a computer-readable recording medium storing an image collation program to be executed by a computer comprising: 6. The computer according to claim 5, wherein the first image correlation operation is a Fourier phase correlation operation, and the second image correlation operation is a Fourier amplitude correlation operation. A computer-readable recording medium storing an image matching program.