KR101073944B1 - Image checking device, image checking method, and computer-readable recording medium for storing a image checking program - Google Patents

Abstract

A correlation value image is generated from an input image and a template image, and the correlation value image is separated into a positive correlation value image and a negative correlation value image based on whether or not the pixel value is greater than or equal to the threshold value, and the template image is divided into pixel values and threshold values. The positive template image and the negative template image are separated according to whether or not they are abnormal, and a plurality of positive / negative separated correlation images are generated by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image. The collation determination is performed using the positive / negative separation correlation image. As such an input image and a template image, the polar coordinate converted input image and template image are used.

Description

Computer-readable recording medium recording image control device, image control method and image control program {IMAGE CHECKING DEVICE, IMAGE CHECKING METHOD, AND COMPUTER-READABLE RECORDING MEDIUM FOR STORING A IMAGE CHECKING PROGRAM}

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to an image contrast apparatus, an image contrast method, and an image contrast program for collating images by comparing characteristics of an image between an input image of a matching object and a plurality of pre-registered template images. An image contrast apparatus, an image contrast method, and an image contrast program capable of efficiently collating with a template image to improve the contrast ratio.

 Background Art [0002] Conventionally, an image control apparatus is known which compares an input image obtained by photographing a deposited money with a CCD (Charge Coupled Device) camera and a template image registered in advance, and determines the authenticity of such currency.

For example, Patent Literature 1 calculates a correlation value by comparing an input image of a coin and a template image of a coin, and in a portion of a predetermined area or more of all images to be contrasted, such a correlation value represents a threshold value. In the case of falling over, an image contrast technique is disclosed in which the hardening of the input image is determined to be genuine hardening.

[Patent Document 1]

JP 2003-187289 A

However, in the case of using this conventional technique, since image contrast is performed using only correlation values exceeding a threshold value, for example, due to the deviation of an image or the occurrence of noise due to image conversion, etc., in contrasting an input image and a template image. When the correlation value was taken at a relatively low value, there was a problem that it was difficult to improve the contrast ratio of the image contrast by judging that the input hardening was a counterfeit hardening, even if the hardening was a real hardening.

In addition, such a problem is not only a problem which occurs only in the image contrast of hardening, but also a problem which arises similarly in the image contrast of bills, the image contrast of components and products in FA (Factory Automation), etc., for example.

This invention is made in view of the said subject (problem), The image control apparatus, the image control method, and the image control program which can improve the image contrast precision of an item other than a currency and money, and can improve the contrast ratio of image contrast are provided. It aims to provide.

In order to solve the above problems and achieve the object, the image control device according to the invention of claim 1 compares the image by comparing the characteristics of the image between the input image of the verification target and a plurality of template images registered in advance. An image control apparatus, comprising: a correlation value that generates a correlation value image from the input image and the template image, and separates the correlation value image into a positive correlation value image and a negative correlation value image based on whether or not the pixel value is equal to or greater than a threshold value Image separation means, and template image separation means for separating the template image into a positive template image and a negative template image according to whether or not the pixel value is equal to or greater than a threshold value, the positive correlation value image, the negative correlation value image, and the positive image. A plurality of template images and combinations of the negative template images It characterized in that it includes a support EXECUTIVE / negative-separated correlation positive / negative-separated correlation-image generating means for generating an image and a contrast determining means for determining the control using the positive / negative-separated correlation images.

Further, in the image control apparatus according to the invention of claim 2, in the invention of claim 1, the positive / negative separation correlation image generating unit includes a value obtained by calculating a product of each pixel of the positive correlation value image and the positive template image. A negative feature region image having a pixel value is generated by using a positive feature region image as a value and a value obtained by calculating a product for each pixel of the negative correlation value image and the positive template image.

Further, in the image control apparatus according to the invention of claim 3, in the invention of claim 1, the positive / negative separation correlation image generating unit calculates a value obtained by calculating a product of each positive pixel of the positive correlation image and the negative template image. A negative background region image having a pixel value is generated by using a positive background region image serving as a pixel value and a value calculated for each pixel of the negative correlation value image and the negative template image.

In the image control apparatus according to the invention of claim 4, in the invention of claim 1, the positive / negative separation correlation image generating unit calculates a product for each pixel of the positive correlation value image and the positive template image. A positive feature region image with the pixel value, a negative feature region image with the pixel value as a value obtained by calculating a product of each pixel of the negative correlation value image and the positive template image, and the positive correlation image and the negative template image A positive background region image having a pixel value as a pixel value and a negative background region image having a pixel value as a value obtained by calculating a product of each pixel of the negative correlation value image and the negative template image; Characterized in that.

In the image control apparatus according to the invention of claim 5, in the invention of claim 2, 3 or 4, the positive / negative separation correlation image generating means is the attention in the negative region image generated using the negative correlation value image. A pixel value of at least one of the peripheral pixels is larger than a pixel value of the pixel of interest, in contrast to a pixel and surrounding pixels of a corresponding pixel corresponding to the pixel of interest in a positive region image generated using the positive correlation value image. In this case, an expansion process for moving the pixel of interest to the corresponding pixel is performed.

In the image control apparatus according to the invention of claim 6, in the invention of claim 1, the input image and the template image are edge images which are image-converted by edge extraction processing using an edge extraction operator.

The image control apparatus according to the invention of claim 7 is characterized in that in the invention of claim 6, the edge image is a normalized edge image in which the edge strength of the extracted edge is normalized.

Moreover, in the image control apparatus which concerns on invention of Claim 8, in the invention of Claim 1, the said template image is an average image which averaged the image about each individual object of the said comparison object.

Further, in the image control apparatus according to the invention of claim 9, in the invention of claim 1, the correlation value image uses a normalized correlation value obtained by normalizing a correlation value for each pixel of the input image or the template image as a pixel value. It is an image.

Further, in the image contrast apparatus according to the invention of claim 10, in the invention of claim 1, the collation determining means divides the positive / negative separation correlation image into blocks and calculates the total sum of pixel values in each block as a block value. And a check value is calculated by adding a product of the block value and the weighting coefficient to all the positive / negative separated correlation images to calculate a check value.

The image control apparatus according to the invention of claim 11 is characterized in that, in the invention of claim 1, the check determining unit calculates the value of the weighting coefficient by linear discriminant analysis.

In the image control apparatus according to the invention of claim 12, in the invention of claim 1, the object to be compared is money.

The image matching method according to the invention of claim 13 is an image matching method in which an image is compared by comparing a feature of an image between an input image of a matching object and a plurality of template images registered in advance. And a correlation value image separation step of generating a correlation value image from the template image, and separating the correlation value image into a positive correlation value image and a negative correlation value image based on whether or not a pixel value is equal to or larger than a threshold value. A template image separation step of separating a positive template image and a negative template image according to whether or not the pixel value is equal to or larger than a threshold value, the positive correlation value image, the negative correlation value image, the positive template image, and the negative template image; A plurality of positive / negative separation correlation images by a combination of It characterized in that first and last verification determining step of performing a verification determination by using a positive / negative-separated correlation-image generating step, the positive / negative-separated correlation images.

The image contrast program according to the invention of claim 14 is a program that causes a computer to execute an image collation method for collating images by comparing characteristics of an image between an input image of a collation target object and a plurality of template images registered in advance. A correlation value image separation step of generating a correlation value image from the input image and the template image, and separating the correlation value image into a positive correlation value image and a negative correlation value image based on whether or not a pixel value is equal to or larger than a threshold value And a template image separation step of separating the template image into a positive template image and a negative template image based on whether or not the pixel value is equal to or larger than a threshold value, the positive correlation value image, the negative correlation value image, and the positive template image; A plurality of negative template images If EXECUTIVE / negative-separated correlation images, and the positive / negative-separated correlation-image generating step of generating the contrast and the determining step of performing a verification determination by using the positive / negative-separated correlation images, wherein execution of the computer.

The image matching apparatus according to the invention of claim 15 is an image matching apparatus for matching an image by comparing the characteristics of the image between an input image of a matching object and a plurality of template images registered in advance. And polar coordinate conversion image generating means for generating a ρ-θ input image and a ρ-θ template image correcting the rotational shift of both images after polar coordinate conversion of the template image, and the ρ-θ input image and the ρ- a correlation value image separation means for generating a correlation value image from the θ template image, and separating the correlation value image into a positive correlation value image and a negative correlation value image depending on whether or not the pixel value is equal to or larger than a threshold value; Template image separation means for separating an image into a positive template image and a negative template image depending on whether or not the pixel value is equal to or larger than a threshold value. Positive / negative separation correlation image generating means for generating a plurality of positive / negative separation correlation images by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image, and the positive And a contrast determination means for performing a contrast determination using a negative separation correlation image.

Further, in the image matching apparatus according to the invention of claim 16, in the invention of claim 15, the positive / negative separation correlation image generating unit calculates a product for each pixel of the positive correlation value image and the positive template image. A positive feature region image having a pixel value is generated by generating a positive feature region image having a pixel value as a pixel value and a value obtained by calculating a product of each negative pixel between the negative correlation value image and the positive template image.

Further, in the image control apparatus according to the invention of claim 17, in the invention of claim 15, the positive / negative separation correlation image generating unit calculates a product for each pixel of the positive correlation value image and the negative template image. A positive background region image having a pixel value, and a negative background region image having a pixel value as a value obtained by calculating a product for each pixel of the negative correlation value image and the negative template image are generated.

Further, in the image control apparatus according to the invention of claim 18, in the invention of claim 15, the positive / negative separation correlation image generating unit calculates a product for each pixel of the positive correlation value image and the positive template image. A positive feature region image having a pixel value, a negative feature region image having a pixel value as a value obtained by calculating a product of each pixel of the negative correlation value image and the positive template image, the positive correlation image and the negative template A positive background region image having a value obtained by calculating a product of each pixel with an image as a pixel value, and a negative background region image having a value obtained by calculating a product of each pixel of the negative correlation value image and the negative template image as a pixel value. It is characterized by generating.

Further, in the image control apparatus according to the invention of claim 19, in the invention of claim 16, 17 or 18, the positive / negative separation correlation image generating means is the attention in the negative region image generated using the negative correlation value image. A pixel value of at least one of the peripheral pixels is larger than a pixel value of the pixel of interest, in contrast to a pixel and surrounding pixels of a corresponding pixel corresponding to the pixel of interest in a positive region image generated using the positive correlation value image. In this case, an expansion process for moving the pixel of interest to the corresponding pixel is performed.

In the image control apparatus according to the invention of claim 20, in the invention of claim 1, the ρ-θ input image and the ρ-θ template image are edge images obtained by image conversion by edge extraction processing using an edge extraction operator. It is characterized by that.

In the image control apparatus according to the invention of claim 21, in the invention of claim 20, the edge image is a normalized edge image in which the edge strength of the extracted edge is normalized.

Moreover, in the image control apparatus which concerns on invention of Claim 22, in the invention of Claim 1, the said template image is an average image which averaged the image about each individual object of the said comparison object.

Further, in the image control apparatus according to the present invention of claim 23, in the invention of claim 1, the correlation value image is a normalized correlation value obtained by normalizing a correlation value for each pixel of the ρ-θ input image and the ρ-θ template. It is characterized by the image which makes into a pixel value.

Further, in the image control apparatus according to the invention of claim 24, in the invention of claim 1, the matching and judging means blocks the positive / negative separation correlation image to calculate a total of pixel values in each block as a block value. And a check value is calculated by adding a product of the block value and the weighting coefficient to all the positive / negative separated correlation images to calculate a check value.

The image control apparatus according to the present invention of claim 25 is characterized in that, in the invention of claim 1, the contrast determining means calculates the value of the weighting coefficient by linear discriminant analysis.

In the image control apparatus according to the present invention of claim 26, in the invention of claim 1, the polar coordinate conversion image generating means rotates the two images by causing the ρ-θ input image or the ρ-θ template image to move in parallel. It characterized in that to correct.

In the image control apparatus according to the invention of claim 27, in the invention of claim 1, the circular object is cured.

The image contrast method according to the invention of claim 28 is an image contrast method in which an image is contrasted by comparing features of an image between an input image of a matching object and a plurality of template images registered in advance. And a polar coordinate conversion image generating step of generating a ρ-θ input image and a ρ-θ template image correcting the rotational shift of both images after the polar coordinate conversion of the template image, and the ρ-θ input image and the ρ- a correlation value image separation step of generating a correlation value image from the θ template image, and separating the correlation value image into a positive correlation value image and a negative correlation value image depending on whether or not the pixel value is equal to or larger than a threshold value; A template image separation step of separating the template image into a positive template image and a negative template image depending on whether or not the pixel value is equal to or larger than a threshold value; A positive / negative separation correlation image generation process of generating a plurality of positive / negative separation correlation images by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image, and the positive / And a collation determination step of performing collation determination using a negative separation correlation image.

The image contrast program according to the invention of claim 29 is a program for causing a computer to execute an image contrast method for collating images by comparing the characteristics of the images between an input image of a matching object and a plurality of template images registered in advance. And a polar coordinate conversion image generating step of generating a ρ-θ input image and a ρ-θ template image correcting rotational shifts of both images after polar coordinate conversion of the input image and the template image, and the ρ-θ input. A correlation value image separation step of generating a correlation value image from the image and the ρ-θ template image, and separating the correlation value image into a positive correlation value image and a negative correlation value image based on whether or not the pixel value is a threshold value; The ρ-θ template image is divided into a positive template image and a negative template image based on whether or not the pixel value is equal to or larger than a threshold value. A positive / negative separated correlation image for generating a plurality of positive / negative separated correlation images by a combination of a template image separation process and the positive correlation value image and the negative correlation value image and the positive template image and the negative template image A computer performs a generation process and a verification judgment step of performing a verification decision using the positive / negative separation correlation image.

According to the invention of claim 1, a correlation value image is generated from an input image and a template image, and the correlation value image is divided into a positive correlation value image and a negative correlation value image based on whether or not the pixel value is equal to or larger than a threshold value, and the template The image is divided into a positive template image and a negative template image according to whether or not the pixel value is equal to or greater than a threshold value, and a plurality of positive / negative images are formed by combining a positive correlation value image and a negative correlation value image with a positive template image and a negative template image. Since a negative separation correlation image is generated and a collation determination is performed using the positive / negative separation correlation image, a correlation value of not only a portion having a high correlation between the input image and a template image but also a portion having a low correlation is used. At the same time, as well as the features of the template image, Image contrast is also performed using the background portion, so that image contrast with high precision can be performed, and the contrast ratio of the image can be improved.

Moreover, according to invention of Claim 2, the positive characteristic area | region image which makes the pixel value the value which computed the product of each pixel of a positive correlation value image and a positive template image, and the pixel per pixel of a negative correlation value image and a positive template image Since a negative feature region image having a product value is generated as a pixel value, an image is generated by using an area image having a feature in the portion where the feature should appear and an area image having no feature in the region where the feature should appear. By contrast, image contrast with high precision can be performed, and the contrast ratio of an image can be improved.

According to the invention of claim 3, a positive background region image having a pixel value as a value obtained by calculating a product of each pixel of a positive correlation value image and a negative template image, and each pixel of the negative correlation value image and the negative template image Since it is configured to generate a negative background region image whose pixel value is a product value, image contrast is performed by using a region image having a background in a part where a background should be present and a region image having no background in a part where a background should be present. By doing so, it is possible to perform image contrast with high precision, and the effect of improving the contrast ratio of an image can be obtained.

According to the invention of claim 4, the product of the positive feature region image whose pixel value is the value calculated for each pixel of the positive correlation value image and the positive template image, and the product of each pixel of the negative correlation value image and the positive template image Negative feature region image using calculated value as pixel value, positive background region image using pixel value as calculated value for each pixel of positive correlation value image and negative template image, negative correlation value image and negative template image It is configured to generate a negative background region image whose pixel value is the product of the product of each pixel, and thus the region image where the feature is to be shown and the feature where the feature is not to be shown. There must be an area image and an area image with a background in the part where the background should be By using an area image without background parts performing image contrast, it is possible to perform high-precision screen image contrast, it has the effect of being able to improve the production rate of the large picture.

Further, according to the invention of claim 5, the pixel of interest in the negative region image generated using the negative correlation value image and the surrounding pixel of the corresponding pixel corresponding to the pixel of interest in the positive region image generated using the positive correlation value image In contrast, when the pixel value of at least one peripheral pixel is larger than the pixel value of the pixel of interest, an expansion process for moving the pixel of interest to the corresponding pixel is performed. The effect of the point can be eliminated, and image contrast with high precision can be performed, and the contrast ratio of an image can be improved.

In addition, according to the invention of claim 6, since the input image and the template image are configured to be edge images that are image-converted by edge extraction processing using an edge extraction operator, by contrasting the feature portions of each extracted image, high accuracy is achieved. Image contrast can be performed, and the contrast ratio of an image can be improved.

According to the invention of claim 7, the edge image is configured to be a normalized edge image in which edge strengths of the extracted edges are normalized, so that image contrast with high precision can be performed without removing the influence of individual differences of the contrasting object. Has the effect of improving the contrast ratio.

Further, according to the invention of claim 8, the template image is configured to be an average image obtained by averaging the images of the individual objects of the matching object, so that even if the object of the matching object has a unique shape, image contrast with high accuracy can be performed. It is possible to improve the contrast ratio of the image.

Further, according to the invention of claim 9, since the correlation value image is configured to be an image having a normalized correlation value obtained by normalizing the correlation value for each pixel of the input image or the template image as the pixel value, the variation of the correlation value is suppressed and the accuracy is increased. Can achieve a high image contrast, and can improve the contrast ratio of an image.

Further, according to the invention of claim 10, the collation judging means divides the positive / negative separation correlation image into blocks and calculates the total sum of the pixel values of each block as the block value, and multiplies the product of the block value with the weighting factor for all positive / Since the contrast value is calculated by adding the negative separated correlation image to perform the contrast determination, the weight of the area where the feature is likely to appear and the weight of the area where the feature is difficult to appear can be adjusted and simplified by the calculation procedure. In addition, efficient image contrast can be performed, and the contrast ratio of the image can be improved.

Further, according to the invention of claim 11, since the collation determination means is configured to calculate the value of the weighting coefficient by linear discriminant analysis, it is possible to obtain an appropriate weighting coefficient based on the learning sample, so that image contrast with high precision is achieved. Can be performed, and the contrast ratio of the image can be improved.

In addition, according to the invention of claim 12, since the collation object is configured to be money, an image contrast with high precision can be performed with respect to the collation of money, which has the effect of improving the contrast ratio of the image.

According to the invention of claim 13, a correlation value image is generated from an input image and a template image, and the correlation value image is divided into a positive correlation value image and a negative correlation value image depending on whether or not the pixel value is equal to or larger than a threshold value. The template image is divided into a positive template image and a negative template image according to whether or not the pixel value is equal to or greater than a threshold value, and the plurality of positive image images are combined by combining a positive correlation image and a negative correlation image with a positive template image and a negative template image. Since a positive / negative separated correlation image is generated and contrast determination is performed using the positive / negative separated correlation image, a correlation value of not only a portion having a high correlation between the input image and a template image but also a portion having a low correlation At the same time, as well as features of the template image In addition, by performing the image contrast processing using the background portion, it is possible to perform image contrast with high accuracy, which has the effect of improving the contrast ratio of the image.

According to the invention of claim 14, a correlation value image is generated from an input image and a template image, and the correlation value image is divided into a positive correlation value image and a negative correlation value image depending on whether or not the pixel value is equal to or larger than a threshold value. The template image is divided into a positive template image and a negative template image according to whether or not the pixel value is equal to or greater than a threshold value, and the plurality of positive image images are combined by combining a positive correlation image and a negative correlation image with a positive template image and a negative template image. Since a positive / negative separated correlation image is generated and contrast determination is performed using the positive / negative separated correlation image, a correlation value of not only a portion having a high correlation between the input image and a template image but also a portion having a low correlation At the same time, as well as features of the template image In addition, by performing the image contrast processing using the background portion, there is an effect that image contrast with high accuracy can be performed.

According to the invention of claim 15, after the polar coordinate conversion of the input image and the template image, the ρ-θ input image and the ρ-θ template image are corrected by correcting the rotational deviation of both images, and the ρ-θ input image and Generates a correlation value image from the p-theta template image, separates the correlation value image into a positive correlation value image and a negative correlation value image based on whether or not the pixel value is greater than or equal to the threshold value, and converts the p-theta template image to the pixel value The positive template image and the negative template image are separated according to whether or not the threshold value is larger than the threshold value, and a plurality of positive / negative separation correlation images are formed by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image. Is generated, and a collation determination is made using the positive / negative separation correlation image. By using the correlation value of not only the portion where the correlation between the input image and the template image is high, but also the portion where the correlation is low, image contrast is performed using not only the template portion but also the background portion. This has the effect of allowing contrast.

According to the invention of claim 16, a positive feature region image having a pixel value as a value obtained by calculating a product of each pixel of the positive correlation value image and the positive template image and the pixel of the negative correlation value image and the positive template image Since a negative feature region image having a product value is generated as a pixel value, an image is generated by using an area image having a feature in the portion where the feature should appear and an area image having no feature in the region where the feature should appear. The contrast has an effect that image contrast with high precision can be performed.

According to the invention of claim 17, the positive background region image having the pixel value as a value obtained by calculating the product of each pixel of the positive correlation value image and the negative template image, and the pixel of the negative correlation value image and the negative template image Since it is configured to generate a negative background region image whose pixel value is a product value, image contrast is performed by using a region image having a background in a part where a background should be present and a region image having no background in a part where a background should be present. By doing so, there is an effect that image contrast with high precision can be performed.

According to the invention of claim 18, a positive feature region image having a pixel value as a value obtained by calculating a product of each pixel of the positive correlation value image and the positive template image and the pixel of the negative correlation value image and the positive template image Negative feature area image using pixel value as a product value, positive background area image using pixel value as a value calculated for each pixel of positive correlation value image and negative template image, negative correlation value image and negative template Since the negative background area image is generated by using the pixel value as the value calculated for each pixel with the image, the area image where the feature should appear and the feature where the feature should appear. Area image with background, area image with background, and background Image contrast is performed by using an area image having no background in the part to be divided, and thus an image contrast with high precision can be performed.

Further, according to the invention of claim 19, the pixel of interest in the negative region image generated using the negative correlation value image and the surrounding pixel of the corresponding pixel corresponding to the pixel of interest in the positive region image generated using the positive correlation value image In contrast, when the pixel value of at least one peripheral pixel is larger than the pixel value of the pixel of interest, an expansion process for moving the pixel of interest to the corresponding pixel is performed. The effect of the point can be eliminated, and image contrast with high precision can be performed.

According to the invention of claim 20, since the ρ-θ input image and the ρ-θ template image are configured to be edge images that are image-converted by edge extraction processing using an edge extraction operator, the features of each extracted image are determined. By contrast, it is possible to perform image contrast with high precision.

According to the invention of claim 21, since the edge image is configured to be a normalized edge image in which edge strengths of the extracted edges are normalized, an image contrast with high precision can be performed by excluding the influence of individual differences of the contrasting object. Has

Further, according to the invention of claim 22, since the template image is configured to be an average image obtained by averaging the images of the individual objects of the matching object, even if the object of the matching object has an inherent shape, image contrast with high accuracy is achieved. It has the effect that it can be done.

In addition, according to the invention of claim 23, since the correlation value image is configured to be an image having a normalized correlation value obtained by normalizing the correlation value for each pixel of the ρ-θ input image and the ρ-θ template image as a pixel value. It is effective to suppress image irregularities and to perform image contrast with high precision.

Further, according to the invention of claim 24, the collation determination means divides the positive / negative separation correlation image into blocks and calculates the sum of the pixel values in each block as a block value, and calculates the product of the block value and the weighting factor for all the positive / negative values. Since the contrast value is calculated by adding the negative separated correlation image to perform the contrast determination, the weight of the area where the feature is likely to appear and the weight of the area where the feature is difficult to appear can be adjusted and simplified by the calculation procedure. It has the effect that efficient image contrast can be performed.

Further, according to the invention of claim 25, since the collation determination means is configured to calculate the value of the weighting coefficient by linear discriminant analysis, an appropriate weighting coefficient based on the learning sample can be obtained, so that image contrast with high precision is achieved. Has the effect that

Further, according to the invention of claim 26, since the polar coordinate conversion image generating means is configured to correct the rotational shift of both images by shifting the ρ-θ input image or the ρ-θ template image in parallel, It has the effect of reducing the image contrast by high efficiency.

In addition, according to the invention of claim 27, since the circular object is configured to be cured, it has an effect that image contrast with high precision can be performed with respect to currency matching.

According to the invention of claim 28, after converting the polarity of the input image and the template image, the ρ-θ input image and the β-θ template image are corrected, and the ρ-θ input image and A correlation value image is generated from the p-theta template image, and the correlation value image is separated into a positive correlation value image and a negative correlation value image according to whether or not the pixel value is equal to or greater than a threshold value, and the p-theta template image is divided into pixel values. The positive template image and the negative template image are separated according to whether or not the threshold value is larger than the threshold value, and a plurality of positive / negative separation correlation images are formed by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image. Is generated, and a collation determination is made using the positive / negative separation correlation image. By using the correlation value of not only the portion where the correlation between the input image and the template image is high, but also the portion where the correlation is low, image contrast is performed using not only the template portion but also the background portion. This has the effect of allowing contrast.

According to the invention of claim 29, after converting the input image and the template image by polar coordinates, the ρ-θ input image and the ρ-θ template image are corrected by correcting the rotational deviation of both images, and the ρ-θ input image and A correlation value image is generated from the p-theta template image, and the correlation value image is separated into a positive correlation value image and a negative correlation value image according to whether or not the pixel value is equal to or greater than a threshold value, and the p-theta template image is divided into pixel values. The positive template image and the negative template image are separated according to whether or not the threshold value is larger than the threshold value, and a plurality of positive / negative separation correlation images are formed by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image. Is generated, and a collation determination is made using the positive / negative separation correlation image. By using the correlation value of not only the portion where the correlation between the input image and the template image is high, but also the portion where the correlation is low, image contrast is performed using not only the template portion but also the background portion. This has the effect of allowing contrast.

1 is a functional block diagram showing the configuration of an image control apparatus according to a first embodiment;

2 is an explanatory diagram for explaining a processing outline of the image cropping unit shown in FIG. 1;

3 is an explanatory diagram for explaining a Sobel operator used in the edge extraction unit shown in FIG. 1;

4 is an explanatory diagram for explaining a processing outline of an edge extracting unit shown in FIG. 1;

5 is an explanatory diagram for explaining a processing outline of a matching determination of the matching processing unit shown in FIG. 1;

6 is an explanatory diagram for explaining an image input to a positive / negative separation correlation determining unit according to the first embodiment;

7 is an explanatory diagram for explaining a feature region and a background region generated by the positive / negative separated correlation image generating unit according to the first embodiment;

8 is a flowchart of normalized correlation value image positive / negative separation processing according to the first embodiment;

9 is a flowchart of a template image positive / negative separation process according to the first embodiment;

10 is a flowchart showing a processing procedure of a positive / negative separated correlation image generating unit according to the first embodiment;

11 is an explanatory diagram for explaining an image generation procedure corresponding to each region shown in FIG. 7;

12 is an explanatory diagram for explaining an image mask used in the expansion processing unit according to the first embodiment;

13 is a flowchart showing a processing procedure of the expansion processing unit according to the first embodiment;

14 is an explanatory diagram for explaining an image generated by the expansion processing unit according to the first embodiment;

15 is an explanatory diagram for explaining block division of an image used in a contrast value calculating unit according to the first embodiment;

16 is an explanatory diagram for explaining a modification of the expansion processing according to the first embodiment;

17 is an explanatory diagram for explaining an image mask used in the modification shown in FIG. 16;

18 is a flowchart showing a processing procedure of an expansion processing unit in the modification shown in FIG. 16;

19 is a functional block diagram showing a configuration of an image control apparatus according to a second embodiment;

20 is an explanatory diagram for explaining a process outline of the image cropping unit shown in FIG. 19;

21 is an explanatory diagram for explaining a Sobel operator used in the edge extraction unit shown in FIG. 19;

FIG. 22 is an explanatory diagram for explaining a process outline of an edge extracting unit shown in FIG. 19; FIG.

23 is an explanatory diagram for illustrating a processing outline of polar coordinate transformation according to the second embodiment;

24 is an explanatory diagram for explaining a processing outline of the rotation angle detector shown in FIG. 19;

25 is an explanatory diagram for explaining respective area images according to the second embodiment;

26 is a flowchart of normalized correlation value image positive / negative separation processing according to the second embodiment;

27 is a flowchart of template image positive / negative separation processing according to the second embodiment;

28 is a flowchart showing the processing procedure of the positive / negative separated correlation image generating unit according to the second embodiment;

FIG. 29 is an explanatory diagram for explaining an image generation procedure corresponding to each region shown in FIG. 25;

30 is an explanatory diagram for illustrating an image mask used in the expansion processing unit according to the second embodiment;

31 is a flowchart showing a processing procedure of an expansion processing unit according to the second embodiment;

32 is an explanatory diagram for explaining an image generated by the expansion processing unit according to the second embodiment;

33 is an explanatory diagram for explaining block division of an image used in a contrast value calculating unit according to the second embodiment;

34 is an explanatory diagram for explaining a modification of the inflation process according to the second embodiment;

FIG. 35 is an explanatory diagram for explaining an image mask used in the modification shown in FIG. 34; FIG.

FIG. 36 is a flowchart showing the processing procedure of the expansion limb in the modification shown in FIG.

[Description of Drawings]

1: Image control unit 10: Image input unit

11: input image 20: image cutting unit

21: Projection in the horizontal direction 22: Projection in the vertical direction

23: Cutting image (rear surface) 24: Cutting image (surface)

30: edge extraction unit

30a: Sobel operator (for calculating horizontal edges)

30b: Sobel operator (for vertical edge calculation)

31: Edge extraction image 32: Edge normalization image (back side)

33: edge normalized image (surface) 40: matching processing unit

50: registration image storage unit 51: template image

51a: t + image 51b: t- image

100: positive / negative separation correlation determination unit

110: normalized correlation value calculation unit

111: normalized correlation value image 111a: r + image

111b: r-picture

120: positive / negative separation correlation image generation unit

121: A + area image 122: A- area image

123: B + area image 124: B- area image

130: expansion processing unit 130a: positive area image mask

130b: negative area image mask 130c: input image mask

130d: Template image mask 131: Expansion completed A + area image

132: expansion completion A-area image 133: expansion completion B + area image

134: expansion completed B-area image

135: expansion completion normalized correlation image

135a: expansion completed r + image 135b: expansion completed r + image

140: control value calculator 141: block division (A + region)

142: block division (A- area) 143: block division (B + area)

144: block division (B-area) 201: image control device

210: image input unit 211: input image

220: Image cutting unit 221: Projection in the horizontal direction

222: vertical projection 223: cropped image

230: edge extraction unit

230a: Sobel operator (for calculating horizontal edges)

230b: Sobel operator (for vertical edge calculation)

231: edge extraction image 232: edge normalization image

233: Polar coordinate complete edge normalized image

240: matching processing unit 240a: polar coordinate conversion unit

240b: rotation angle detection unit 240c: front and rear determination unit

250: Registered image storage unit 251: Template image

251a: t + image 251b: t- image

300: positive / negative separation correlation determination unit

310: normalized correlation value calculation unit

311 normalized correlation value image 311a r + image

311b: r-picture

320: positive / negative separation correlation image generation unit

321: A + area image 322: A- area image

323: B + area image 324: B- area image

330: Expansion treatment unit 330a: Positive area image mask

330b: Negative area image mask 330c: Input image mask

330d: Template image mask 331: Expansion completed A + area image

332: Expansion completed A- area image 333: Expansion completed B + area image

334: Expansion completed B-area image

335 expansion expansion normalized correlation image

335a: Expansion completed r + image 335b: Expansion completed r + image

340: control value calculator

EMBODIMENT OF THE INVENTION Below, with reference to an accompanying drawing, Embodiment 1-2 of the image control apparatus, image contrast method, and image contrast program concerning this invention are demonstrated in detail. In Example 1, the image contrast using the Cartesian coordinate system will be described, and in Example 2, the image contrast using the polar coordinate system will be described.

Example 1

Fig. 1 is a functional block diagram showing the configuration of the image control apparatus according to the first embodiment. As shown in the figure, this image control apparatus 1 includes an image input unit 10, an image cutting unit 20, an edge extracting unit 30, a matching processing unit 40, and a registered image storage unit. 50 and a positive / negative separation correlation determination unit 100, and such a positive / negative separation correlation determination unit 100 includes a normalized correlation value calculation unit 110 and a positive / negative separation correlation image. The generation unit 120, the expansion processing unit 130, and the control value calculating unit 140 are provided.

The image input unit 10 is an input unit for introducing a hardened input image to be collated into the apparatus, and outputs the input image to the image cropping unit 20. Specifically, the image input unit 10 treats the input image as an aggregate of a predetermined number of pixels. For example, the input image is recognized as a grayscale image having a density value of 256 gradations, and output to the image cropping unit 20 as a rectangular image of a predetermined size.

The image cropping unit 20 obtains such a rectangular image from the image input unit 10, cuts out only the image in the square region circumscribed to the cured image, and outputs the cut image to the edge extraction unit 30.

2 is an explanatory diagram for explaining the processing outline of the image cropping unit 20. As shown in the figure, the image cropping unit 20 scans the input image 11 acquired from the image input unit 10 in the horizontal direction, accumulates the density values of all the pixels, and generates a horizontal projection 21. . Further, the input image 11 is scanned in the vertical direction to generate the vertical projection 22 in the same order. The image cropping section 20 scans the horizontal projection 21 and the vertical projection 22 to calculate the rising and falling coordinates of the accumulated concentration values. And as shown with the four broken lines of the said figure, the area | region enclosed in each calculated coordinate is cut out as the cutout image 23, and this cutout image 23 is output to the edge extraction part 30. FIG.

Returning to the description of FIG. 1, the edge extraction unit 30 will be described. The edge extraction unit 30 acquires the cutting image 23 from the image cutting unit 20, and in order to avoid the influence based on individual differences such as brightness and color tone of the cutting image 23, the cutting image ( 23) Calculate concentration change (edge strength). Moreover, in order to suppress the nonuniformity of the calculated edge intensity, edge intensity normalization is performed. Specifically, edge strength is calculated by performing the edge extraction process using the Sobel operator on the cutout image 23, and the calculation result is normalized. In addition, in Example 1, although the Sobel operator is used, edge extraction can also be performed using a Robert operator or the like.

3 is an explanatory diagram for explaining a Sobel operator. As shown in the figure, the edge extracting section 30 calculated edge strength using two Sobel operators: horizontal edge calculating 30a and vertical edge calculating 30b. Specifically, each Sobel operator 30a and 30b is scanned with respect to all the pixels of the cutout image 23, and a horizontal edge calculation result Gx and a vertical edge calculation result Gy are acquired. After calculating the edge intensity G in each pixel, the edge intensity is normalized (E).

As shown in equation (1), the edge intensity G in each pixel is expressed as the sum of the absolute value of the horizontal edge calculation result Gx and the absolute value of the vertical edge calculation result Gy. As shown in Equation (2), the normalized edge intensity E in each pixel is a product of the integer c and the edge intensity G, in which a predetermined value is set for each type of curing, over the entire pixel. It is divided by the total of the edge strengths (G).

By normalizing the edge strength in this manner, it is possible to suppress the occurrence of nonuniformity of the edge strength between the myths where edges tend to come out and the circulation that hardly comes out of the edges. Without this, various hardening contrasts can be performed with high accuracy.

4 is an explanatory diagram for explaining an outline of an edge extraction process (image conversion process) performed by the edge extraction unit 30. As shown in the figure, the cutout image 23 is converted into an edge extraction image 31 by an edge strength calculation process using a Sobel operator. And the edge extraction image 31 is image-converted into the edge normalization image 32 by the edge intensity normalization process using Formula (1) and Formula (2). The edge extraction unit 30 outputs this edge normalized image 32 to the matching processing unit 40.

Each pixel value of the edge extraction image 31 shown in the figure takes a value of 0 to 255, for example, 0 corresponds to black, and 255 takes a grayscale value corresponding to white. In the edge extraction image 31 of the figure, a white portion is an extracted edge portion, and a black portion is a background portion. Each pixel value of the edge normalized image 32 takes a value of 0 to 255, for example, a gray scale value in which 0 corresponds to black and 255 corresponds to white. In addition, in the edge normalized image 32 of the drawing, the point where the white portion corresponds to the edge portion and the black portion corresponds to the background is the same as that of the edge extraction image 31.

Returning to the description of FIG. 1, the matching processing unit 40 will be described. The matching processing unit 40 acquires the edge normalization image 32 from the edge extraction unit 30, and performs a matching process with the template image stored in the registered image storage unit 50. Specifically, the template image is rotated by a predetermined angle to obtain a rotation angle Φ at which the degree of coincidence M between the template image and the edge normalized image 32 at each rotation angle is maximum. Such agreement degree (M) is

Calculate by

As shown in Equation 3, the degree of agreement M (Φ) at each rotation angle Φ is the density value tΦ (x, y) of each pixel of the template image rotated by the angle Φ, and the edge normalized image 32. The product of the density values s (x, y) of each pixel is summed over the phone.

5 is an explanatory diagram for illustrating an outline of a matching decision of the matching processing unit 40. As shown in the figure, the value of M (Φ) becomes a graph of a mountain shape having a maximum value at a certain rotation angle. The matching processing unit 40 acquires the value of Φ such that M (Φ) becomes the maximum (vertical portion of the mountain), and rotates the template image stored in the registered image storage unit 50 by an angle Φ. Then, the edge normalized image 32 and the template image of rotation completion are output to the positive / negative separation correlation determination unit 100.

FIG. 6 is an explanatory diagram for explaining an image example of an edge normalized processing completed image and a rotation completed template image output from the matching processing unit 40 to the positive / negative separation correlation determination unit 100. In the figure, an example of an image in the case where an image of a surface of 10 yen coin is input to the image control apparatus 1 is shown. That is, when the edge normalization process is performed on the cutout image 24, the edge normalization image 33 is generated, and the template image 51 of the rotation completion is generated by rotating the template image by the above matching process. . In the following description, such an edge normalized image (surface) 33 is used instead of the edge normalized image (back side) 32.

In addition, although the matching process part 40 demonstrated the acquisition of the rotation angle which maximizes a matching degree by rotating a template image, it rotates the edge normalization image 32, without rotating a template image, and the rotation angle which becomes a maximum matching degree. May be obtained.

Returning to the description of FIG. 1, the registered image storage unit 50 will be described. The registered image storage unit 50 stores a plurality of template images corresponding to various kinds of registrations registered in advance, and provides these template images to the matching processing unit 40. An average image obtained by synthesizing a plurality of images of the same kind of curing is used for such a template image in order to suppress unevenness due to individual differences in curing. By using such an average image, the correlation value between the concave-convex pattern portion inherent to each coin, such as the manufacturing year, and the corresponding portion of the template image, becomes a correlation value with respect to the average image (average value), and thus has an influence upon contrasting. Becomes difficult. That is, even though it is genuine hardening, it can prevent that it is judged as counterfeit hardening by a different manufacture year.

Such a template image is registered in the registration storage unit 50 after the edge normalization processing is performed in the same manner as the input image in order to check the input image subjected to the edge normalization processing. In addition, in the registration storage unit 50, a plurality of images subjected to edge normalization processing are registered on the average image on the front and back surfaces of various money.

The positive / negative separation correlation determination unit 100 receives, from the matching processing unit 40, an edge normalized image 33 (hereinafter referred to as an "input image 33") and a rotation completed template image 51 (hereinafter referred to as "input image 33") shown in FIG. Template image 51 "), and matching these images to determine whether or not the hardening of the input image 33 is real hardening, and outputs such a determination result.

The normalized correlation value calculation unit 110 calculates a correlation value for each pixel to which the input image 33 and the template image 51 correspond, and normalizes such a correlation value to generate a normalized correlation value image. Specifically, for each pixel whose coordinate value is (x, y), the density value s (x, y) of the input image 33 and the density value t (x, y) of the template image 51 are used.

The normalized correlation value r (x, y) of each pixel is calculated by In addition, the normalized correlation value r (x, y) in each pixel shown by Formula (4) takes the value of -1.0- + 1.0, for example. In the formula (4), n represents the number of pixels.

Then, the normalized correlation value calculating unit 110 determines the positive normalized correlation value image (r + image) and the negative normalized correlation value image (r-image) according to whether or not the pixel value of the normalized correlation value image is O or more. To separate. The template image 51 is divided into a positive template image (t + image) and a negative template image (t-image) depending on whether each pixel value is equal to or larger than a predetermined threshold value Tt.

In addition, the pixel value of an r + image takes the value of 0.0-1.0, for example, and the pixel value of an r- image takes the value of 0.0-1.0, for example by taking the absolute value of each pixel value. In addition, the pixel value of a t + image and a t- image takes the value of 0 or 1, for example. That is, the t + image and the t- image have a role as an image mask used for image conversion of each normalized correlation value image.

Herein, the meaning of each image will be described. The r + image represents a correlated (similar) pixel between the images to be collated, and if there is a strong correlation, such a pixel takes a large value. In addition, the r-image indicates that there is no correlation (not similar) between the images to be contrasted, and when there is a strong negative correlation, such a pixel takes a large value. The t + image represents the edge portion of the template image, and the edge portion has a value of 1 and the background portion has a value of zero. The t-image represents the background portion (the portion other than the edge) of the template image, and the background portion has a value of 1 and the edge portion has a value of zero.

The positive / negative split correlation image generation unit 120 generates a positive / negative split correlation image by a combination of an r + image, an r− image, a t + image, and a t− image generated by the normalized correlation value calculation unit 110. . Specifically, an A + area image from an r + image and a t + image, an A- area image from an r- image and a t + image, a B + area image from an r + image and a t- image, and a B- from an r- image and a t- image Create each area image.

The meaning of each area image is explained here. 7 is an explanatory diagram for explaining such four areas. As shown in the figure, the A + region image is an region image in which an r + image and a t + image are superimposed, and indicates that the edge portion is correlated with the edge portion, that is, the edge is located where the edge should come out. Corresponds to the positive feature region image of. The A-area image is an area image in which an r- image and a t + image are overlapped, and indicates that the edge portion does not correlate with the edge portion, that is, the edge does not appear where the edge should come out, and is a negative feature region in the claims. Corresponds to the image. The B + region image is a region image in which an r + image and a t- image are overlapped, indicating that the edge is not present where the edge should not appear, i.e., the positive background region in the claims. Corresponds to the image. A B-area image is an area image in which an r-image and a t-image are overlapped, and indicates that the edge is not present in the background part, i.e., the edge is located where the edge should not come out, and the negative background in the claims. Corresponds to the area image.

Returning to the description of FIG. 1, the expansion processing unit 130 will be described. The expansion processing unit 130 moves the pixels of the A-area image to the A + area image and moves the pixels of the B-area image to the B + area image by using a predetermined image mask. Such expansion processing is performed because an isolated point having a negative correlation value in the form of noise appears in the normalized correlation value. In other words, by performing such an expansion process, it is possible to suppress the influence of such an isolated point on the determination result of the control value.

The reference value calculation unit 140 divides each of the A + region image, the A- region image, the B + region image, and the B- region image into four blocks in four in the horizontal direction and four in the vertical direction. ,

The control value Z is obtained by equation (5). Here, coefficients (a _jj , b _jj , c _jj and d _jj ) are obtained by linear discriminant analysis using learning samples. In addition, A + _jj , A- _jj , B + _jj, and B- _{jj which} are block values of each area image represent the sum total of the pixel values in each block.

If the control value Z is equal to or greater than the threshold value, the control value calculator 140 checks the hardening of the input image 33 against real hardening, and if it is smaller than the threshold value, checks the counterfeit hardening and the contrast. Next, such a determination result is output.

In the following, the processing of the positive / negative separation correlation determination unit 100 shown in FIG. 1 will be described in more detail. First, the normalized correlation value positive / negative separation processing performed by the normalized correlation value calculation unit 110 will be described with reference to FIGS. 8 and 11. FIG. 8 is a flowchart of normalized correlation value positive / negative separation processing, and FIG. 11 is an explanatory diagram for explaining an image generation procedure in the positive / negative separation correlation determination unit 100.

As shown in FIG. 11, the normalization correlation value calculation part 110 produces | generates the normalization correlation value image 111 from the input image 33 and the template image 51 first. The normalized correlation value positive / negative separation process is performed by using the generated normalized correlation value image 111 as an input. The normalized correlation value image 111 is subjected to the r + image 11la which is a positive correlation value image, The image is separated into an r-image 111b which is a negative correlation value image.

As shown in FIG. 8, in a normalization correlation value positive / negative separation process, it moves to the viewpoint pixel of the normalization correlation value image 111 first (step S501). Such viewpoint pixels are, for example, pixels of x = 0 and y = 0. Then, the normalized correlation value r (x, y) of such a pixel is calculated using Equation (4) (step S502), and if the calculated r (x, y) is 0 or more (YES in step S503), Let the pixel value be the pixel value of the same coordinate of r + image 111a (step S504). On the other hand, if the calculated r (x, y) is less than 0 (NO in step S503), the absolute value of the pixel value of such a pixel is taken as the pixel value of the same coordinate of the r-image 111b (step S505). .

If the positive / negative separation process has not yet been completed for all the pixels of the normalized correlation value image 111 (step S506 negative), the process moves to the next pixel of interest (step S507), and the processes subsequent to step S502 are repeated. do. On the other hand, when the positive / negative separation process is finished for all the pixels (YES in step S506), the process ends. By such normalized correlation value positive / negative separation processing, the r + image 111a and the r- image 111b are generated as an image having a pixel having a pixel value of 0.0 to 1.0. In the first embodiment, the pixel value of the pixel of the r-image 111b is described as taking a pixel value of 0.0 to 1.0, but such a pixel value may take a value of -1.0 to 0.0.

Next, the template image positive / negative separation processing performed by the normalized correlation value calculation unit 110 will be described with reference to FIGS. 9 and 11. 9 is a flowchart of a template image positive / negative separation process. As shown in Fig. 11, in the template image positive / negative separation process, a process of separating the template image 51 into a t + image 51a which is a positive template image and a t- image 51b that is a negative template image is performed. do.

As shown in FIG. 9, in a template image positive / negative separation process, it moves to the viewpoint pixel of the template image 51 first (step S601). Such viewpoint pixels are, for example, pixels of x = 0 and y = 0. And if the density value of such a pixel is more than predetermined threshold Tt (YES in step S602), the pixel value of the same coordinate of t + image 51a shall be 1 (step S603). On the other hand, if this density value is smaller than the predetermined threshold value Tt (step S602 negative), the pixel value at the same coordinate of the t-image 51b is 1 (step S604).

If the positive / negative separation process has not yet been completed for all the pixels of the template image 51 (step S605 is negative), the process moves to the next pixel of interest (step S606), and the processes following step S602 are repeated. On the other hand, when the positive / negative separation process is finished for all the pixels (YES in step S605), the process ends. By such a template image positive / negative separation process, the t + image 51a is generated as a binary image having an edge portion of 1 and a background portion of 0. The t-image 51b has an edge portion of 0 and a background portion of 0. It is generated as a binary image of 1.

Next, the positive / negative separation correlation image generation process performed by the positive / negative separation correlation image generation unit 120 will be described with reference to FIGS. 10 and 11. 10 is a flowchart of positive / negative separation correlation image generation processing.

As shown in Fig. 11, in the positive / negative split correlation image generation process, the r + image 111a, r- image 111b, t + image 51a, and t- image generated by the normalized correlation value calculation unit 110 are shown. An A + area image 121, an A- area image 122, a B + area image 123, and a B- area image 124 are generated using the 51b as an input image.

For example, in the case where the r + image 111a and the t + image 51a are used as the input image, as shown in FIG. 10, the image is first moved to the viewpoint pixel of each image (step S701). If the pixel value of the t + image 51a in such a pixel is 1 (YES in step S702), the pixel value of the A + region image 121 is the pixel value of the r + image 111a (step S703). On the other hand, when the pixel value of the t + image 51a in such a pixel is not 1 (i.e., 0) (step S702 negative), the pixel value of the A + area image 121 is called 0 (step S704). .

If the area image generation processing has not been completed for all the pixels (step S705 is negative), the processing moves to the next pixel of interest (step S706), and the process of step S702 or less is repeated. On the other hand, when the area image generation processing is finished for all the pixels (YES in step S705), the A + area image 121 is generated, so the process ends.

Similarly, the A-area image 122 is obtained from the r-image 111b and the t + image 51a, and the B-area image 123 is obtained from the r + image 111a and the t-image 51b. And a B-region image 124 are generated from the t-image 51b, respectively.

Next, the expansion processing performed by the expansion processing unit 130 will be described with reference to FIGS. 12 to 14. FIG. 12 is an explanatory diagram for explaining the image mask used in the expansion process, FIG. 13 is a flowchart of the expansion process, and FIG. 14 is an explanatory diagram for explaining the image generated by the expansion process.

In such an expansion process, the isolated point (pixel) of the noise shape included in the negative area image (A-area image 122 and B-area image 124) is selected as the positive area image (A + area image 121). And B + area image 123]. By performing such a process, the precision of a contrast value can be raised.

As shown in FIG. 12, in such an expansion process, two image masks of the positive area image mask 130a and the negative area image mask 130b are used. Each image mask has P5 and M5 and eight areas surrounding these areas. For example, in the case of performing an expansion process from the A-region image 122 to the A + region image 121, M5 of the negative region image mask 130b is aligned with the pixel of interest of the A-region image 122, P5 of the positive area image mask 130a is matched with the pixel corresponding to the pixel of interest. Then, the pixel value of M5 and the pixel values of P1 to P9 are sequentially compared to perform expansion processing.

Next, a case where the expansion process is performed from the A-area image 122 to the A + area image 121 is taken as an example, and the processing procedure of such an expansion process will be described with reference to FIG. First, the process moves to the viewpoint pixels of the respective images 121 and 122 (step S801). Such a viewpoint pixel is a pixel of x = 0 and y = 0, for example. Then, 1 is set to n to sequentially convert the nine regions P1 to P9 of the positive region mask 130a (step S802). That is, when step S802 is completed, the area of the positive area image mask 130a as a target becomes P1.

When the value of Pn is compared with the value of M5, and the value of P1 is larger than the value of M5 (YES in step S803), the value of P5 is replaced with M5 and the value of M5 is set to 0 (step S805). That is, the pixel of M5 is moved to the pixel of P5. On the other hand, when the value of Pn is equal to or less than the value of M5 (step S803 negative), 1 is added to the value of n (step S804), and when the value of n is 9 or less (step S806 negative), step 803 is performed again. .

Thus, if any of the values of P1 to P9 is larger than the value of M5, the pixel of M5 is moved to P5. On the other hand, when all of the values of P1 to P9 are equal to or less than the value of M5 (YES in step S806), the pixel is not moved.

If the processing has not been completed for all the pixels of the A-region image 122 (step S807 is negative), the processing moves to the next pixel of interest (step S808), and the processing after step S802 is performed. On the other hand, when the processing is finished for all the pixels of the A-area image 122 (YES in step S807), this expansion processing ends.

As shown in Fig. 14, the A + area image 121, the A- area image 122, the B + area image 123, and the B- area image 124 are each expanded in the expanded A + area image by such an expansion process. 131, the expanded A-region image 132, the expanded B + region image 133, and the expanded B-region image 134 are converted into images. In addition, since the isolated point on the A-region image 122 moves to the A + region image 121, the edge portion of the expanded A + region image 131 increases in area compared with the A + region image 121. Doing. On the other hand, the edge portion of the expanded A-region image 132 is reduced in area compared with the A-region image 122.

Next, the control value calculation process performed by the control value calculator 140 will be described with reference to FIG. 15.

15 is an explanatory diagram for explaining the block division of the expansion-completed area images 131 to 134. As shown in the figure, the control value calculation processing unit 140 first divides each of the expanded-completed area images 131 to 134 into 16 blocks in four in the horizontal direction and four in the vertical direction. 141, an A-region image block 142, a B + region image block 143, and a B- region image block 144 are generated.

And the contrast value calculation part 140 calculates a contrast value Z using Formula (5). Here, each coefficient (a _jj , b _jj , c _jj, and d _jj ) in Equation (5) is assumed to obtain an optimal solution by linear discriminant analysis or the like using a learning sample. Specifically, because of the difference in the design of the uneven pattern of hardening, there are hardly hard edges and hard hard to come out, so these coefficients take different values for each type of hardening. By optimizing these coefficients with a learning sample, it is possible to perform image contrast with high accuracy.

The contrast value calculator 140 calculates the contrast Z using the coefficients a _jj , b _jj , c _jj, and d _{jj in which the} optimum values are set, and the respective image blocks 141 to 144. When the control value is greater than or equal to the threshold, it is determined to be genuine hardening, and when smaller than the threshold, it is determined to be counterfeit hardening. In Example 1, the case where each image is divided into 16 blocks has been described, but the number of blocks can be any number.

In addition, in Equation (5), when the coefficients c _jj and d _jj are set to 0, the control value Z can be calculated only from the A + region image block 141 and the A- region image block 142. . And count (a _jj and When b _jj ) is set to 0, the control value Z can be calculated only from the B + region image block 143 and the B− region image block 144.

In this way, the contrast calculation unit 140 can perform image contrast efficiently by adjusting the number of image blocks or the value of each coefficient of the equation (5) according to the type of hardening or the capability of hardware.

In addition, although the contrast calculation part 140 which concerns on Example 1 was comprised so that block value may be divided | segmented into each area | region image, and the contrast value Z is calculated by Formula (5), it is not limited to this, but it is not limited to this method. It can also be configured to perform the contrast determination. For example, other methods such as a multilayer neural network, a support vector machine, and a quadratic identification function may be used.

Hereinafter, the case where expansion processing is performed before positive / negative separation of the normalized correlation value image 111 is demonstrated using FIGS. 16-18. FIG. 16 is an explanatory diagram for explaining an image generation procedure by such an expansion process, FIG. 17 is an explanatory diagram for explaining an image mask used for such an expansion process, and FIG. 18 is a flowchart of such an expansion process. .

In the above expansion process, after generation of each of the area images 121 to 124, a positive area image (for example, A + area image 121) is generated from a negative area image (for example, A-area image 122). ] To move the pixels. However, such an expansion process can be performed using the normalized correlation value image 111 before positive / negative separation and the template image 51 before positive / negative separation.

As shown in FIG. 16, the normalization correlation value calculation part 110 produces | generates the normalization correlation value image 111 from the input image 33 and the template image 51 first. In this expansion process, an expansion process is performed by using the generated normalized correlation value image 111 as an input to generate an expansion completion normalized correlation value image 135. The expansion normalized correlation value image 135 is separated into an expansion completed r + image 135a and an expansion completion r− image 135b. Then, the positive / negative separation correlation value image generating unit 120 is processed by inputting the expanded r + image 135a, the expanded r + image 135b, the t + image 51a and the t-image 51b. The expanded-complete A + area image 131, the expanded-complete A- area image 132, the expanded-complete B + area image 133, and the expanded-complete B- area image 134 are output.

As shown in FIG. 17, in this expansion process, two image masks, an input image mask 130c and a template image mask 130d, are used. Each image mask has S5 and T5 and eight areas surrounding these areas. For example, when expansion processing is performed using the template image 51 and the normalized correlation value image 111, S5 of the input image mask 130c is matched with the pixel of interest of the normalized correlation value image, and the template image mask 130d is used. Is set to a pixel corresponding to the pixel of interest. The expansion processing is performed by referring to and comparing the pixel values of the regions S1 to S9 and T1 to T9.

The processing procedure of such an expansion process is demonstrated using FIG. First, the process moves to the viewpoint pixels of the respective images 111 and 51 (step S901). Such viewpoint pixels are, for example, pixels of x = 0 and y = 0. When the value of S5 is negative, that is, when the normalized correlation value of the pixel described above is negative (step S902 negative), the nine regions S1 to S9 of the input image mask 130c and the template image mask 130d are negative. 1 is set to n in order to convert the nine areas T1 to T9 in the order (step S903).

If the value of Tn is greater than the threshold value Tt (YES at step S904), it is determined whether the value of Sn is at least 0 (step S905), and if the value of Sn is at least 0 (OK at step S905), This value of Sn is compared with the absolute value of S5 (step S906). If the value of Sn is larger than the absolute value of S5 (YES in step S906), the value of Sn is replaced with the absolute value of S5 (step S907).

That is, in the Sn around S5, when the value of Tn is larger than the threshold value Tt, the value of Sn is greater than or equal to 0, and the value of Sn is larger than the absolute value of S5, there exists a region Sn. It is determined that such a pixel of S5 is an isolation point, and the value of S5 is inverted by taking the absolute value of the value of S5. If the expansion process has not been completed for all the pixels of the normalized correlation value image 111 (step S910 is negative), the process moves to the pixel of interest (step S911), and the processes of step S902 or less are repeated. On the other hand, when the expansion processing is finished for all the pixels (YES in step S910), this expansion processing is finished.

On the other hand, if the value of Tn is equal to or less than the threshold Tt (step S904 negative), or the value of Sn is negative (step S905 negative), or the value of Sn is equal to or less than the absolute value of S5 (step S906 negative), If 1 is added (step S908) and n is 9 or less (step S909 negative), the process of step S904 or less is repeated. On the other hand, when n is larger than 9 (YES at step S909), the processing at step S910 is performed.

Thus, even when the expansion process is performed before the positive / negative separation of the normalized correlation value image 111, the expanded area images 131 to 134 can be obtained. In such a case, since the normalized correlation value image 111 before the positive / negative separation is used, the number of images subjected to the expansion process can be reduced as compared with the expansion process after the generation of the area images 121 to 124, More efficient expansion treatment can be performed.

As described above, in the image control apparatus, the image control method, and the image control program according to the first embodiment, the input image subjected to the feature extraction by performing the edge extraction processing and the edge normalization processing, and the template image subjected to the edge normalization processing in advance By contrast, a normalized correlation value image and a normalized correlation value image and a template image are generated, and a positive normalized correlation value image and a negative normalized correlation value image are respectively determined by whether or not the pixel value in each image is equal to or larger than a threshold value. After separating into a positive template image and a negative template image, a combination of such images produces a positive feature region image, a negative feature region image, a positive background region image, and a negative background region image. And from a negative feature region image to a positive feature region image An expansion process for shifting the pixels and moving the pixels from the negative background area image to the positive background area image is performed, the block image is divided into these expansion areas, and a contrast value is calculated by linear discriminant analysis or the like. Since the collation determination is performed, all pixels of the input image and the template image can be matched, and the correlation values of not only the feature region but also the background region can be excluded while the influence of the isolated point due to the correlation value calculation is excluded. Since the contrast can be reflected in the contrast, the image contrast with high accuracy can be performed, and the contrast ratio of the image can be improved.

In Example 1, the case where image contrast is performed for a hardened input image has been described. However, the present invention is not limited thereto, and for example, parts and products in the case of image contrast of bills, factory automation (FA), and the like. The same applies to image contrast of.

(Example 2)

By the way, in Example 1 mentioned above, the case where image contrast, such as hardening, is performed using a rectangular coordinate system was demonstrated. In Embodiment 2, image contrast using a polar coordinate system instead of such a rectangular coordinate system will be described. By using a polar coordinate system, image contrast of circular objects, such as hardening, can be performed more efficiently.

19 is a functional block diagram showing the configuration of the image control apparatus according to the second embodiment. As shown in the figure, the image control apparatus 201 includes an image input unit 210, an image cropping unit 220, an edge extracting unit 230, a matching processing unit 240, and a registered image storage unit 250. ) And a positive / negative separation correlation determination unit 300, and the matching processing unit 240 includes a polar coordinate conversion unit 240a, a rotation angle detection unit 240b, and a front and rear determination unit 240c. The positive / negative separation correlation determination unit 300 includes a normalized correlation value calculation unit 310, a positive / negative separation correlation image generation unit 320, an expansion processing unit 330, and a contrast value calculation. The part 340 is provided.

The image input unit 210 is an input unit for introducing a hardened input image to be collated into the apparatus, and outputs the input image to the image cropping unit 220. Specifically, the image input unit 210 treats the input image as an aggregate of a predetermined number of pixels. For example, the input image is recognized as a grayscale image having a density value of 256 gradations, and output to the image cropping unit 220 as a rectangular image of a predetermined size.

The image cropping unit 220 acquires such a rectangular image from the image input unit 210, cuts out only the image in the square region circumscribed to the cured image, and outputs the cut out image to the edge extracting unit 230.

20 is an explanatory diagram for explaining a processing outline of the image cropping unit 220. As shown in the figure, the image cropping unit 220 scans the input image 211 acquired from the image input unit 210 in the horizontal direction, accumulates the density values of all the pixels, and generates the horizontal projection 221. do. Further, the input image 211 is scanned in the vertical direction to generate the vertical projection 222 in the same order. The image cropping unit 220 scans the horizontal projection 221 and the vertical projection 222 to calculate the rising and falling coordinates of the accumulated concentration values. And the area | region enclosed in each coordinate calculated as shown by the four broken lines of the said figure is cut out as the cutout image 223, and this cutout image 223 is output to the edge extraction part 230. As shown in FIG.

Returning to the description of FIG. 19, the edge extraction unit 230 will be described. The edge extracting unit 230 acquires the cropped image 223 from the image cropping unit 220, and cuts the cropped image 223 to avoid an influence based on individual differences such as brightness and color tone of the cropped image 223. Calculate the concentration change (edge strength) of In addition, the edge strength is normalized in order to suppress the calculated nonuniformity of the edge strength. Specifically, the edge strength is calculated by performing an edge extraction process using the Sobel operator on the cut image 223, and the calculation result is normalized. In the second embodiment, the Sobel operator is used, but edge extraction can also be performed using a Robert operator or the like.

It is explanatory drawing for demonstrating the Sobel operator. As shown in the figure, the edge extracting unit 230 calculates edge strength using two Sobel operators: horizontal edge calculating 230a and vertical edge calculating 230b. Specifically, the Sobel operators 230a and 230b are scanned for all the pixels of the cropped image 223 to obtain a horizontal edge calculation result Gx and a vertical edge calculation result Gy. After calculating the edge intensity G in each pixel, the edge intensity is normalized (E).

As shown in equation (6), the edge intensity G in each pixel is expressed as the sum of the absolute value of the horizontal edge calculation result Gx and the absolute value of the vertical edge calculation result Gy. . As shown in Equation (7), the normalized edge intensity E in each pixel is an edge that spans all pixels by multiplying the integer c and the edge intensity G, in which a predetermined value is set for each type of curing. It is divided by the total of the strengths (G).

By normalizing the edge strength in this way, it is possible to suppress the occurrence of nonuniformity of the edge strength between myths where edges tend to come out and distribution that hardly comes out of the edges. The contrast can be performed with high precision.

22 is an explanatory diagram for explaining an outline of an edge extraction process (image conversion process) performed by the edge extraction unit 230. As shown in the figure, the cropped image 223 is converted into an edge extraction image 231 by an edge strength calculation process using the Sobel operator. And the edge extraction image 231 is image-converted into the edge normalization image 232 by the edge intensity normalization process using Formula (6) and (7). The edge extraction unit 230 outputs this edge normalized image 232 to the matching processing unit 240.

Each pixel value of the edge extraction image 231 shown in the figure takes a value of 0 to 255, for example, a gray scale value where 0 corresponds to black and 255 corresponds to white. In the edge extraction image 231 of the figure, a white portion is an extracted edge portion, and a black portion is a background portion. In addition, each pixel value of the edge normalized image 232 takes a value of 0 to 255, for example, a gray scale value in which 0 corresponds to black and 255 corresponds to white. In addition, in the edge normalized image 232 of the figure, the point where the white portion corresponds to the edge portion and the black portion corresponds to the background is the same as that of the edge extraction image 231.

Returning to the description of FIG. 19, the matching processing unit 240 will be described. The matching processing unit 240 acquires an edge normalization image 232 from the edge extraction unit 230, and acquires a template image of edge normalization and polar coordinate conversion completion from the registered image storage unit 250. The edge normalized image 232 is polar coordinate-transformed, the shift angle between the polar coordinate-converted image and the template image is detected by the parallel movement of the template image, and the front and back are determined to determine the edge normalized image. And the misalignment angle corrected template image are output to the positive / negative separation correlation determination unit 300. In the second embodiment, a case where the shift angle is detected by moving the template image in parallel will be described. However, the shift angle may be detected by parallel shifting the image obtained by polar coordinate conversion of the edge normalized image 232.

The polar coordinate conversion unit 240a is a processing unit for polar coordinate converting the edge normalized image 232. Specifically, the center point of the edge normalized image 232 is calculated, and this center point is used as the origin of the polar coordinate. Each pixel is specified by the rotation angle θ and the distance ρ from the center point, and polar coordinate transformation is performed by moving each pixel to the ρ-θ space. In such a conversion,

Use

Fig. 23 is an explanatory diagram for explaining the processing outline of such polar coordinate conversion. With the center point of the xy space (edge normalized image 232) as the origin and the coordinates of each pixel represented by (x, y), such x and y, and p and θ described above are represented by equation (8) and It has a relationship shown in equation (9). Therefore, by converting each pixel (x, y) in the edge normalized image 232 into (p, θ) satisfying the relationship between the equations (8) and (9), the polar coordinate converting unit 240a, The polar coordinate completed edge normalized image 233 is produced | generated.

Moreover, although the distance p from the center point takes the value of 10-100, and the rotation angle (theta) takes the value of 0-255 in the said figure, although the range of these values can be set arbitrarily. .

Returning to the description of FIG. 19, the rotation angle detector 240b will be described. The rotation angle detection unit 240b detects a shift angle between the polar coordinate-converted edge normalized image 233 and the template image previously polar-transformed by the same polar coordinate conversion process, and performs a process of correcting the shift angle of both images. . 24 is an explanatory diagram for explaining a processing outline of the rotation angle detector 240b.

As shown in the figure, the template image 251 is moved in parallel with the θ coordinate axis in the ρ-θ space. The coincidence degree M (Φ) between the shift angle Φ in the θ coordinate direction, the template image 251 and the edge normalized image 232 at each shift angle Φ is calculated, and this coincidence M (Φ) is maximum. Acquire a rotation angle Φ _max to be. In addition, the match degree M (Φ) is

Calculate by

As shown in Equation (10), the degree of coincidence M (Φ) at each shift angle Φ is a density value of each pixel of the template image 251 when the template image 251 is shifted by Φ. The product of t (k, θ-Φ) and the density value s (k, θ) of each pixel of the edge normalized image 232 is summed over each pixel. Here, k is a selection value which selected the distance from which the characteristic is easy to come out of the distance p from the center point in the edge normalization image 232. For example, such a k is selected by extracting 16 values of p that are easy to come out of characteristics among p (0 to 100) of the polar coordinate-converted edge normalized image 233 shown in FIG.

Such a value of M (Φ) is a graph of a mountain shape having a maximum value at a certain rotation angle as shown in FIG. 24. A rotation angle detection unit (240b) obtains the value of this Φ _max M (Φ) such that the maximum value (peak of the mountain-shaped portions). Thus, since the rotation angle detection part 240b corrects a shift angle by the parallel movement of the p-theta image which performed polar coordinate conversion, it can reduce a calculation amount compared with the method of correcting a shift angle by rotation of an xy image.

Returning to the description of FIG. 19, the front and back determination unit 240c will be described. First, the front and rear determination unit 240c determines the maximum value M of the above coincidence degree M (Φ) between the surface coordinate image and the back surface template image of the polar coordinate conversion completed and the polar coordinate conversion edge normalized image 233. Φ _max ) is calculated and the normalized correlation coefficient R is obtained from this M (Φ _max ). Specifically, this normalized correlation coefficient (R) is

Obtained by In addition, N in Formula (11) represents the number of pixels used as a determination object.

The front and back determination unit 240c selects the template image with the larger normalized correlation coefficient R, and outputs the positive / negative separation correlation determination unit 300 together with the polar coordinate-converted edge normalized image 233. do. For example, the normalized correlation coefficient R between the back template image and the polar coordinate-converted edge normalized image 233 is the normalized correlation coefficient R between the surface template image and the polar coordinate-converted edge normalized image 233. If larger than), the rear template image and the polar coordinate-converted edge normalized image 233 are output to the positive / negative separation correlation determination unit 300. Here, the template image output to the positive / negative separation correlation determining unit 300 is a template image in which the shift angle with the polar coordinate-converted edge normalized image 233 is corrected in parallel by an angle Φ _max .

Returning to the description of FIG. 19, the registered image storage unit 250 will be described. The registered image storage unit 250 stores a plurality of template images corresponding to various kinds of registrations registered in advance, and provides these template images to the matching processing unit 240. The average image which synthesize | combined two or more images of the same kind of hardening is used for such a template image in order to suppress the nonuniformity by the individual difference of hardening. By using such an average image, the correlation value between the concave-convex pattern portion inherent to each coin, such as the manufacturing year, and the corresponding portion of the template image, becomes a correlation value with respect to the average image (average value), and thus has an influence upon contrasting. Becomes difficult. That is, even though it is genuine hardening, it can prevent that it is a counterfeit hardening because a manufacture year differs.

Such a template image is registered in the registered image storage unit 250 after the polar coordinate conversion process and the edge normalization process are performed in the same manner as the input image in order to collide with the input image subjected to the polar coordinate conversion process and the edge normalization process. In addition, a plurality of template images on the front and back sides of various money are registered in the registered image storage unit 250.

Returning to the description of FIG. 19, the positive / negative separation correlation determination unit 300 will be described. The positive / negative separation correlation determination unit 300 performs a polar coordinate conversion completed edge normalized image 233 (hereinafter referred to as an “input image 233”) and a shift angle correction completed template image 251 from the matching processing unit 240 shown in FIG. (Hereinafter, referred to as a "template image 251") is obtained, and the images are collated to determine whether or not the hardening of the input image 233 is a real hardening, and outputs such a determination result.

The normalized correlation value calculating unit 310 calculates a correlation value for each pixel corresponding to the input image 233 and the template image 251, and normalizes such a correlation value to generate a normalized correlation value image. Specifically, for each pixel whose coordinate value is (k, θ), the density value s (k, θ) of the input image 233 and the density value t (k, θ- of the deviation angle corrected template image 251 Using Φ _max )

By using this, the normalized correlation value r (k, θ) of each pixel is calculated. In addition, the normalized correlation value r (k, θ) in each pixel shown in equation (12) takes a value of -1.0 to +1.0, for example. In the formula (12), n represents the number of pixels.

The normalized correlation value calculating unit 310 converts the positive normalized correlation value image (r + image) and the negative normalized correlation value image (r-image) according to whether or not the pixel value of the normalized correlation value image is zero or more. Separate. In the template image 251, a positive template image (t + image) is divided into a negative template image (t− image) depending on whether or not each pixel value is equal to or larger than a predetermined threshold value T _t .

In addition, the pixel value of an r + image takes the value of 0.0-1.0, for example, and the pixel value of an r- image takes the value of 0.0-1.0, for example by taking the absolute value of each pixel value. In addition, the pixel value of a t + image and a t- image takes the value of 0 or 1, for example. That is, the t + image and the t- image have a role as an image mask used for image conversion of each normalized correlation value image.

Herein, the meaning of each image will be described. The r + image represents a correlated (similar) pixel between the images to be collated, and if there is a strong correlation, such a pixel takes a large value. In addition, the r-image indicates that there is no correlation (not similar) between the images to be contrasted, and when there is a strong negative correlation, such a pixel takes a large value. The t + image represents the edge portion of the template image, and the edge portion has a value of 1 and the background portion has a value of zero. The t-image represents the background portion (the portion other than the edge) of the template image, and the background portion has a value of 1 and the edge portion has a value of zero.

The positive / negative split correlation image generation unit 320 generates a positive / negative split correlation image by a combination of an r + image, an r− image, a t + image, and a t− image generated by the normalized correlation value calculation unit 310. . Specifically, an A + area image from an r + image and a t + image, an A- area image from an r- image and a t + image, a B + area image from an r + image and a t- image, and a B- area from an r- image and a t- image Each image is generated.

The meaning of each area image is explained here. 25 is an explanatory diagram for explaining such four areas. As shown in the figure, the A + region image is a region image in which an r + image and a t + image are overlapped, indicating that the edge part is correlated with the edge portion, that is, the edge is located where the edge should appear, and is positive in the scope of the claims. Corresponds to the feature region image. An A-area image is an area image in which an r- image and a t + image are overlapped, indicating that the edge is not correlated with the edge portion, that is, the edge does not appear where the edge should come out, and is a negative feature region in the claims. Corresponds to the image. A B + area image is an area image in which an r + image and a t- image are overlapped, indicating that an edge is not present where the edge does not appear, that is, a background image in the claims. Corresponds to. A B-area image is an area image in which an r-image and a t-image are overlapped, and indicates that the edge is out where the edge should not be generated, that is, the background is negative. Corresponds to the image.

Returning to the description of FIG. 19, the expansion processing unit 330 will be described. The expansion processing unit 330 moves the pixels of the A-area image to the A + area image by using a predetermined image mask and moves the pixels of the B-area image to the B + area image. Such expansion processing is performed because an isolated point having a negative correlation value in the form of noise appears in the normalized correlation value. In other words, by performing such an expansion process, it is possible to suppress the influence of such an isolated point on the determination result of the control value.

The control value calculator 340 divides each of the A + region image, the A- region image, the B + region image, and the B- region image into 64 blocks of 16 in the horizontal direction and 4 in the vertical direction. ,

By the equation (13), the contrast value Z is obtained. Here, the coefficients (a _jj , b _jj , c _jj, and d _jj ) are obtained by the linear discriminant analysis using the learning samples. In addition, A ⁺ _jj , A ^- _jj , B ⁺ _jj, and B ^- _{jj which} are block values of each area | region image represent the sum total of the pixel value in each block.

If the control value Z is greater than or equal to the threshold value, the control value calculator 340 determines that the hardening of the input image 233 is genuine hardening, and if it is smaller than the threshold value, the counterfeit hardening is determined. , Output the collation result.

Hereinafter, the processing of the positive / negative separation correlation determination unit 300 shown in FIG. 19 will be described in more detail. First, the normalized correlation value positive / negative separation processing performed by the normalized correlation value calculation unit 310 will be described with reference to FIGS. 26 and 29. FIG. 26 is a flowchart of normalized correlation value positive / negative separation processing, and FIG. 29 is an explanatory diagram for explaining an image generation procedure in the positive / negative separation correlation determining unit 300.

As shown in FIG. 29, the normalization correlation value calculation unit 310 first generates a normalization correlation value image 311 from the input image 233 and the template image 251. The normalized correlation value positive / negative separation processing is performed by using the generated normalized correlation value image 311 as an input. The normalized correlation value image 311 is subjected to the positive correlation image r + image 311a and negative. R-image 311b which is a correlation value image of?

As shown in FIG. 26, in a normalization correlation value positive / negative separation process, it moves to the viewpoint pixel of the normalization correlation value image 311 first (step S1501). Such a viewpoint pixel is a pixel of k = 0 and (theta) = 0, for example. Then, the normalized correlation value r (k, θ) of such a pixel is calculated using Equation (12) (step S1502), and if the calculated r (k, θ) is 0 or more (YES in step S1503), Let the pixel value be the pixel value of the same coordinate of r + image 311a (step S1504). On the other hand, if the calculated r (k, θ) is less than 0 (step S1503 negative), the absolute value of the pixel value of such a pixel is set to the pixel value of the same coordinate of the r-image 311b (step S1505).

If the positive / negative separation process has not yet been completed for all the pixels of the normalized correlation value image 311 (step S1506 negative), the process moves to the next pixel of interest (step S1507), and the processes of step S1502 or less are repeated. do. On the other hand, when the positive / negative separation process is finished for all the pixels (YES in step S1506), the process ends. By such normalized correlation value positive / negative separation processing, the r + image 311a and r- image 311b are generated as an image having a pixel having a pixel value of 0.0 to 1.0. In the second embodiment, the pixel value of the pixel of the r-image 311b assumes a pixel value of 0.0 to 1.0, but such a pixel value may take a value of -1.0 to 0.0.

Next, the template image positive / negative separation processing performed by the normalized correlation value calculation unit 310 will be described with reference to FIGS. 27 and 29. 27 is a flowchart of the template image positive / negative separation process. As shown in Fig. 29, in the template image positive / negative separation process, a process of separating the template image 251 into a t + image 251a which is a positive template image and a t-image 251b that is a negative template image is performed. .

As shown in FIG. 27, in the template image positive / negative separation process, the process first moves to the viewpoint pixel of the template image 251 (step S1601). Such a viewpoint pixel is a pixel of k = 0 and (theta) = 0, for example. If the density value of such a pixel is equal to or larger than the predetermined threshold T _t (YES in step S1602), the pixel value at the same coordinate of t + image 251a is 1 (step S1603). On the other hand, if this density value is smaller than the predetermined threshold value Tt (step S1602 negative), the pixel value at the same coordinate of the t-image 251b is 1 (step S1604).

If the positive / negative separation process has not yet been completed for all the pixels of the template image 251 (step S1605 negative), the process moves to the next pixel of interest (step S1606) and repeats the processing of step S1602 or less. . On the other hand, when the positive / negative separation process is finished for all the pixels (YES in step S1605), the process ends. By such a template image positive / negative separation process, the t + image 251a is generated as a binary image with an edge portion of 1 and a background portion of 0. The t-image 251b has an edge portion of 0 and a background portion. This 1 is generated as a binary image.

Next, the positive / negative separation correlation image generation process performed by the positive / negative separation correlation image generation unit 320 will be described with reference to FIGS. 28 and 29. 28 is a flowchart of positive / negative separation correlation image generation processing.

As shown in Fig. 29, in the positive / negative split correlation image generation process, the r + image 311a, r- image 311b, t + image 251a and t- image (generated by the normalized correlation value calculating unit 310) 251b) is used as an input image to generate an A + area image 321, an A- area image 322, a B + area image 323, and a B- area image 324.

For example, in the case where the r + image 311a and the t + image 251a are used as the input image, as shown in FIG. When the pixel value of the t + image 251a in such a pixel is 1 (YES in step S1702), the pixel value of the A + region image 321 is the pixel value of the r + image 311a (step S1703). On the other hand, when the pixel value of the t + image 251a in such a pixel is not 1 (that is, 0) (step S1702 is negative), the pixel value of the A + region image 321 is set to 0 (step S1704). .

If the area image generation processing is not completed for all pixels (step S1705 negative), the processing moves to the next pixel of interest (step S1706), and the processes of step S1702 or less are repeated. On the other hand, when the area image generation processing is finished for all the pixels (YES at step S1705), the A + area image 321 is generated, so the process ends.

Similarly, the A-area image 322 is obtained from the r- image 311b and the t + image 251a, and the B-area image 323 is formed from the r + image 311a and the t- image 251b. And a B-region image 324, respectively, from the t-image 251b.

Next, the expansion processing performed by the expansion processing unit 330 will be described with reference to FIGS. 30 to 32. 30 is an explanatory diagram for explaining an image mask used in the expansion process, FIG. 31 is a flowchart of the expansion process, and FIG. 32 is an explanatory diagram for explaining an image generated by the expansion process.

In such an expansion process, the isolated point (pixel) of noise shape included in the negative region image (A-region image 322 and B-region image 324) is a positive region image [A + region image 321]. ) And B + region image 323]. By performing such a process, the precision of a contrast value can be raised.

As shown in FIG. 30, in this expansion process, two image masks of the positive area image mask 330a and the negative area image mask 330b are used. Each image mask has P5 and M5 and eight areas surrounding these areas. For example, in the case of performing an expansion process from the A-area image 322 to the A + area image 321, M5 of the negative area image mask 330b is aligned with the pixel of interest of the A-area image 322, and thus the positive area. P5 of the image mask 330a is matched with the pixel corresponding to the pixel of interest. Then, the pixel value of M5 and the pixel values of P1 to P9 are sequentially compared to perform expansion processing.

Next, a case where the expansion processing is performed from the A-region image 322 to the A + region image 321 is taken as an example, and the processing procedure of such expansion processing will be described with reference to FIG. First, the process moves to the viewpoint pixels of the respective images 321 and 322 (step S1801). Such a viewpoint pixel is a pixel of k = 0 and (theta) = 0, for example. Then, 1 is set to n in order to convert the nine regions P1 to P9 of the positive region mask 330a in sequence (step S1802). That is, when step S1802 is completed, the area of the positive area image mask 330a as the target becomes P1.

Then, the value of Pn is compared with the value of M5, and when the value of P1 is larger than the value of M5 (YES in step S1803), the value of P5 is replaced with M5 and the value of M5 is set to 0 (step S1805). . That is, the pixel of M5 is moved to the pixel of P5. On the other hand, when the value of Pn is equal to or less than the value of M5 (step S1803 negative), 1 is added to the value of n (step S1804), and when the value of n is 9 or less (step S1806 negative), step S1803 is performed again.

Thus, if any of the values of P1 to P9 is larger than the value of M5, the pixel of M5 is moved to P5. On the other hand, when all of the values of P1 to P9 are equal to or less than the value of M5 (YES in step S1806), the pixel is not moved.

If the processing has not been completed for all the pixels of the A-area image 322 (step S1807 is negative), the processing moves to the next pixel of interest (step S1808) and the processing after step S1802 is performed. On the other hand, when the processing is finished for all the pixels of the A-area image 322 (YES at step S1807), this expansion processing is completed.

As shown in FIG. 32, the A + area image 321, the A- area image 322, the B + area image 323, and the B- area image 324 are each expanded in the A + area by such an expansion process. The image is converted into the image 331, the expanded A-region image 332, the expanded B + region image 333, and the expanded B-region image 334. In addition, since the isolated point on the A-region image 322 moves to the A + region image 321, the edge portion of the expanded A + region image 331 increases in area compared with the A + region image 321. Doing. On the other hand, the edge portion of the expanded A-region image 332 is reduced in area compared with the A-region image 322.

Next, the comparison value calculation processing performed by the comparison value calculation unit 340 will be described with reference to FIG. 33. FIG. 33 is an explanatory diagram for explaining an example of dividing a block of the expanded A + area image 331 into block divisions of the expanded area images 331 to 334. As shown in the figure, the control value calculation processing unit 340 first divides the expansion-completed A + area image 331 into 64 blocks of 16 in the horizontal direction and 4 in the vertical direction. Similarly, block division is performed for the expanded A-region image 332, the expanded B + region image 333, and the expanded B-region image 334 as well.

And the contrast value calculating part 340 calculates a contrast value Z using Formula (13). Here, each coefficient (a _jj , b _jj , c _jj and djj) of Equation (13) is assumed to obtain an optimal solution by linear discriminant analysis or the like using a learning sample. Specifically, hardening that tends to come out of edges or hardening that does not come out due to differences in the design of the uneven pattern of hardening, these coefficients take different values for each type of hardening. By optimizing these coefficients with a learning sample, it is possible to perform image contrast with high accuracy.

The contrast calculator 340 calculates the contrast Z using the coefficients (a _jj , b _jj , c _jj and d _jj ) in which the optimal values are set, and each image block. If it is more than the threshold, it is determined to be genuine hardening, and if it is smaller than the threshold, it is determined to be counterfeit hardening. In Example 2, the case where each image is divided into 64 blocks has been described. However, the number of blocks can be any number.

When the coefficients c _jj and d _jj are set to 0 in equation (13), the contrast value Z can be calculated only from the A + region image block and the A- region image block. If the coefficients a _jj and b _jj are set to 0, the control value Z can be calculated only from the B + region image block and the B- region image block.

In this way, the contrast value calculator 340 can perform image contrast efficiently by adjusting the number of image blocks and the value of each coefficient of the equation (13) according to the type of hardening and the capability of hardware.

In addition, although the control value calculation unit 340 according to the second embodiment is configured to calculate the control value Z by dividing each area image by the following equation (13), the present invention is not limited thereto. Other methods such as a multilayer neural network, a support vector machine, and a quadratic identification function may be used.

Hereinafter, the case where expansion processing is performed before the positive / negative separation of the normalized correlation value image 311 is demonstrated using FIGS. 34-36. FIG. 34 is an explanatory diagram for explaining an image generation procedure by such an expansion process, FIG. 35 is an explanatory diagram for explaining an image mask used for such an expansion process, and FIG. 36 is a flowchart of such an expansion process. to be.

The above-described expansion process is performed after the generation of each of the area images 321 to 324, from the negative area image (eg, A-area image 322) to the positive area image (eg, A + area image 321). )] Was moving the pixel. However, such an expansion process can be performed using the normalized correlation value image 311 before positive / negative separation and the template image 251 before positive / negative separation.

As shown in FIG. 34, the normalization correlation value calculation unit 310 first generates a normalization correlation value image 311 from the input image 233 and the template image 251. In this expansion process, an expansion process is performed by using the generated normalized correlation value image 311 to generate an expansion completion normalized correlation value image 335. The expansion normalized correlation value image 335 is separated into the expansion r + image 335a and the expansion r-image 335b. Then, the positive / negative separation correlation value image generating unit 320 is processed by inputting the r + image 335a of the expansion completion, the r-image 335b of the expansion completion, the t + image 251a and the t-image 251b. The expanded completed A + region image 331, the expanded completed A- region image 332, the expanded completed B + region image 333 and the expanded completed B- region image 334 are output.

As shown in FIG. 35, in this expansion process, two image masks, an input image mask 330c and a template image mask 330d, are used. Each image mask has S5 and T5 and eight areas surrounding these areas. For example, in the case where expansion processing is performed using the template image 251 and the normalized correlation value image 311, the template image mask 330d is aligned with S5 of the input image mask 330c to the pixel of interest of the normalized correlation value image. ) Is matched to the pixel corresponding to the pixel of interest. The expansion processing is performed by referring to and comparing the pixel values of the regions S1 to S9 and T1 to T9.

The processing procedure of such an expansion process is demonstrated using FIG. First, the process moves to the viewpoint pixels of the respective images 311 and 251 (step S1901). Such viewpoint pixels are, for example, pixels of k = 0 and θ = 0. When the value of S5 is negative, that is, when the normalized correlation value of the pixel described above is negative (step S1902 negative), the nine regions S1 to S9 of the input image mask 330c and the template image mask 330d are negative. In order to convert the nine regions T1 to T9 in the sequence, 1 is set to n (step S1903).

If the value of Tn is greater than the threshold value T _t (YES at step S1904), it is determined whether the value of Sn is at least 0 (step S1905), and if the value of Sn is at least 0 (YES at step S1905). Then, the value of Sn is compared with the absolute value of S5 (step S1906). If the value of Sn is larger than the absolute value of S5 (YES in step S1906), the value of Sn is replaced with the absolute value of S5 (step S1907).

That is, in the Sn around S5, when the value of Tn is larger than the threshold value T _t , and the value of Sn is greater than or equal to 0, and the value of Sn is larger than the absolute value of S5, there exists a region Sn. It is determined that such a pixel of S5 is an isolated point, and the value of S5 is inverted by taking the absolute value of the value of S5. If the expansion process has not been completed for all the pixels of the normalized correlation value image 311 (step S1910 negative), the process moves to the pixel of interest (step S1911), and the processes of step S1902 or less are repeated. On the other hand, when the expansion processing is finished for all the pixels (YES at step S1910), the expansion processing is completed.

On the other hand, when the value of Tn is equal to or less than the threshold value T _t (step S1904 negative), or the value of Sn is negative (step S1905 negative), or the value of Sn is equal to or less than the absolute value of S5 (step S1906 negative), n 1 is added (step S1908), and if n is 9 or less (step S1909 negative), the process of step S1904 or less is repeated. On the other hand, when n is larger than 9 (YES at step S1909), the process at step S1910 is performed.

Thus, even when the expansion process is performed before the positive / negative separation of the normalized correlation value image 311, the expanded area images 331 to 334 can be obtained. In such a case, since the normalized correlation value image 311 before the positive / negative separation is used, the number of images subjected to expansion processing can be reduced as compared with the expansion processing after the generation of the area images 321 to 324. More efficient expansion treatment can be performed.

As described above, in the image control apparatus, the image control method, and the image control program according to the second embodiment, polar coordinate conversion input images in which feature extraction is performed by performing edge extraction processing and edge normalization processing, and polar coordinates in which edge normalization processing has been previously performed. Contrast the conversion template image, correct the shift angles of the two images to generate a normalized correlation value image, and positively correlate the normalized correlation value image and the template image by whether the pixel value in each image is equal to or larger than a threshold value. The normalized correlation value image and the negative normalized correlation value image are separated into a positive template image and a negative template image, and then a positive feature region image, a negative feature region image, and a positive Create background area images and negative background area images, and also create negative The block processing is performed by dividing the expanded area image by performing the expansion process for moving the pixel from the image area image to the positive feature area image and the movement of the pixel from the negative background area image to the positive background area image. Since the contrast value is calculated by linear discriminant analysis, the collation judgment is performed, so that all pixels of the input image and the template image are collated and the influence of the isolated point due to the correlation value calculation is eliminated. Since not only the region but also the correlation value of the background region can be reflected in the contrast with good balance, the image contrast with high precision can be performed, and the contrast ratio can be improved.

In addition, in Example 2, the case where image contrast is performed with respect to a hardened input image is demonstrated, but this invention is not limited to this, For example, image contrast of medals used in amusement facilities, etc., FA (Factory Automation) It can also be applied to the image control of circular parts and circular products. In addition, the control object does not necessarily need to be circular, and the present invention can also be applied to hardening or parts having a point symmetrical shape such as a regular octagon or a hexagonal hexagon.

As described above, the image control apparatus, the image control method, and the image control program according to the present invention are useful for image control of articles, and are particularly suitable for collation of money such as bills and coins.

Claims (29)

In the image control apparatus which contrasts an image by comparing the characteristic of an image between the input image of a matching object, and the some template image registered in advance, A correlation value image is generated in which a correlation value for each corresponding pixel of the input image and the template image is a pixel value, and the correlation value image is negatively correlated with a positive correlation value image depending on whether or not a pixel value is equal to or larger than a threshold value. Correlation value image separating means for separating into value images, Template image separation means for separating the template image into a positive template image and a negative template image depending on whether or not the pixel value is equal to or larger than a threshold value; Positive / negative separated correlation image generating means for generating a plurality of positive / negative separated correlation images by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image; And contrast determination means for performing contrast determination using the positive / negative separation correlation image. The method of claim 1, The positive / negative separated correlation image generating means includes a positive feature region image having a pixel value as a value obtained by calculating a product of each positive pixel between the positive correlation value image and the positive template image, and the negative correlation value image and the positive template. And a negative feature region image whose pixel value is a value obtained by calculating a product of each pixel with an image. The method of claim 1, The positive / negative separated correlation image generating means includes a positive background region image having a pixel value as a value obtained by calculating a product of the positive correlation value image and the negative template image for each pixel, and the negative correlation value image and the negative template. And a negative background area image whose pixel value is a value obtained by calculating a product of each pixel with an image. The method of claim 1, The positive / negative separated correlation image generating means includes a positive feature region image having a pixel value as a value obtained by calculating a product of each positive pixel between the positive correlation value image and the positive template image, and the negative correlation value image and the positive template. A negative feature region image whose pixel value is a value calculated from each pixel with an image; a positive background region image whose pixel value is a value obtained by calculating a product of each pixel of the positive correlation value image and the negative template image; And a negative background area image having a pixel value as a value obtained by calculating a product of each negative pixel with the negative correlation value image and the negative template image. The method according to any one of claims 2 to 4, The positive / negative split correlation image generating means corresponds to a pixel of interest in a negative area image generated using the negative correlation value image and a pixel of interest in a positive area image generated using the positive correlation value image. And an expansion processing for moving the pixel of interest to the corresponding pixel when the pixel value of the at least one peripheral pixel is larger than the pixel value of the pixel of interest in contrast to the pixel of the pixel of the pixel. . The method of claim 1, And the input image and the template image are edge images which are image converted by an edge extraction process using an edge extraction operator. The method of claim 6, And the edge image is a normalized edge image obtained by normalizing edge strengths of the extracted edges. The method of claim 1, And said template image is an average image obtained by averaging images of the individual objects of the matching object. The method of claim 1, And said correlation value image is an image whose pixel value is a normalized correlation value obtained by normalizing a correlation value for each corresponding pixel of said input image and said template image. The method of claim 1, The matching and judging means divides the positive / negative separated correlation image into blocks and calculates the sum of pixel values in each block as a block value, and multiplies the block value and the weighting coefficient with respect to all the positive / negative separated correlation images. And adding a check value to calculate a check value. The method of claim 10, And said contrast determining means calculates a value of said weighting coefficient by linear discriminant analysis. The method of claim 1, The contrast control device is characterized in that the currency. In the image contrast method of contrasting an image by comparing the characteristics of the image between an input image of a matching object and a plurality of template images registered in advance, A correlation value image is generated, wherein the correlation value for each corresponding pixel of the input image and the template image is a pixel value, and the correlation value image is negatively correlated with the positive correlation value image by whether or not the pixel value is equal to or larger than a threshold value. A correlation value image separation process for separating into value images, A template image separation step of separating the template image into a positive template image and a negative template image depending on whether or not the pixel value is equal to or larger than a threshold value; A positive / negative separation correlation image generation step of generating a plurality of positive / negative separation correlation images by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image; And a collation determination step of performing collation determination using the positive / negative separation correlation image. A computer-readable recording medium having recorded thereon an image control program for contrasting an image by comparing characteristics of an image between an input image of a matching object and a plurality of template images registered in advance, A correlation value image is generated in which a correlation value for each corresponding pixel of the input image and the template image is a pixel value, and the correlation value image is negatively correlated with a positive correlation value image depending on whether or not a pixel value is equal to or larger than a threshold value. A correlation value image separation process for separating into value images, A template image separation step of separating the template image into a positive template image and a negative template image depending on whether or not the pixel value is equal to or larger than a threshold value; A positive / negative separation correlation image generation step of generating a plurality of positive / negative separation correlation images by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image; A computer-readable recording medium having recorded thereon an image control program for causing a computer to perform a contrast determination process of performing a contrast determination using the positive / negative separated correlation image. In the image control apparatus which contrasts an image by comparing the characteristics of an image between an input image of a circular object and a plurality of template images registered in advance, Polar coordinate conversion image generating means for generating a ρ-θ input image and a ρ-θ template image correcting the rotational deviation of both images after polar coordinate conversion of the input image and the template image; A correlation value image is generated in which the correlation value for each corresponding pixel of the ρ-θ input image and the ρ-θ template image is a pixel value, and the correlation value image is positive depending on whether or not the pixel value is equal to or larger than a threshold value. Correlation value image separation means for separating the correlation value image and the negative correlation value image; Template image separation means for separating the ρ-θ template image into a positive template image and a negative template image according to whether or not the pixel value is equal to or larger than a threshold value; Positive / negative separated correlation image generating means for generating a plurality of positive / negative separated correlation images by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image; And contrast determination means for performing contrast determination using the positive / negative separation correlation image. The method of claim 15, The positive / negative separated correlation image generating means includes a positive feature region image having a pixel value as a value obtained by calculating a product of each positive pixel between the positive correlation value image and the positive template image, and the negative correlation value image and the positive template. And a negative feature region image whose pixel value is a value obtained by calculating a product of each pixel with an image. The method of claim 15, The positive / negative separated correlation image generating means includes a positive background region image having a pixel value as a value obtained by calculating a product of the positive correlation value image and the negative template image for each pixel, and the negative correlation value image and the negative template. And a negative background area image whose pixel value is a value obtained by calculating a product of each pixel with an image. The method of claim 15, The positive / negative separated correlation image generating means includes a positive feature region image having a pixel value as a value obtained by calculating a product of each positive pixel between the positive correlation value image and the positive template image, and the negative correlation value image and the positive template. A negative feature region image whose pixel value is a value calculated from each pixel with an image; a positive background region image whose pixel value is a value obtained by calculating a product of each pixel of the positive correlation value image and the negative template image; And a negative background area image having a pixel value as a value obtained by calculating a product of each negative pixel with the negative correlation value image and the negative template image. The method according to any one of claims 16 to 18, The positive / negative split correlation image generating means corresponds to a pixel of interest in a negative area image generated using the negative correlation value image and a pixel of interest in a positive area image generated using the positive correlation value image. And an expansion processing for moving the pixel of interest to the corresponding pixel when the pixel value of the at least one peripheral pixel is larger than the pixel value of the pixel of interest in contrast to the pixel of the pixel. . The method of claim 15, And the ρ-θ input image and the ρ-θ template image are edge images that are image-converted by edge extraction processing using an edge extraction operator. The method of claim 20, And the edge image is a normalized edge image obtained by normalizing edge strengths of the extracted edges. The method of claim 15, And said template image is an average image obtained by averaging images of individual objects of said circular object. The method of claim 15, And said correlation value image is an image whose pixel value is a normalized correlation value obtained by normalizing a correlation value for each corresponding pixel of said ρ-θ input image and said ρ-θ template image. The method of claim 15, The matching and judging means divides the positive / negative separated correlation image into blocks and calculates the sum of pixel values in each block as a block value, and multiplies the block value and the weighting coefficient with respect to all the positive / negative separated correlation images. And adding a check value to calculate a check value. The method of claim 24, And said contrast determining means calculates a value of said weighting coefficient by linear discriminant analysis. The method of claim 15, And the polar coordinate conversion image generating means corrects the rotational shift of both images by moving the ρ-θ input image or the ρ-θ template image in parallel. The method of claim 15, And said circular object is cured. In the image contrast method of contrasting an image by comparing characteristics of an image between an input image of a circular object and a plurality of template images registered in advance, A polar coordinate conversion image generation step of generating a ρ-θ input image and a ρ-θ template image correcting rotational shifts of both images after polar coordinate conversion of the input image and the template image; A correlation value image is generated in which the correlation value for each corresponding pixel of the ρ-θ input image and the ρ-θ template image is a pixel value, and the correlation value image is positive depending on whether or not the pixel value is equal to or larger than a threshold value. A correlation value image separation step of separating the correlation value image and the negative correlation value image; A template image separation step of separating the ρ-θ template image into a positive template image and a negative template image according to whether or not the pixel value is equal to or larger than a threshold value; A positive / negative separation correlation image generation step of generating a plurality of positive / negative separation correlation images by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image; And a collation determination step of performing collation determination using the positive / negative separation correlation image. A computer-readable recording medium having recorded thereon an image control program for contrasting an image by comparing characteristics of an image between an input image of a circular object and a plurality of template images registered in advance, A polar coordinate conversion image generation step of generating a ρ-θ input image and a ρ-θ template image correcting rotational shifts of both images after polar coordinate conversion of the input image and the template image; A correlation value image is generated in which the correlation value for each corresponding pixel of the ρ-θ input image and the ρ-θ template image is a pixel value, and the correlation value image is positive depending on whether or not the pixel value is equal to or larger than a threshold value. A correlation value image separation step of separating the correlation value image and the negative correlation value image; A template image separation step of separating the ρ-θ template image into a positive template image and a negative template image according to whether or not the pixel value is equal to or larger than a threshold value; A positive / negative separation correlation image generation step of generating a plurality of positive / negative separation correlation images by combining the positive correlation value image and the negative correlation value image with the positive template image and the negative template image; A computer-readable recording medium having recorded thereon an image control program for causing a computer to perform a contrast determination process of performing a contrast determination using the positive / negative separated correlation image.

Images