JP2009031996A - Image collation device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image collation device which collates two deformed images, and outputs a collation result. <P>SOLUTION: This image collation device is provided with: a change amount database for storing the change amount of the position of the featured point of an object shape included in a first image which has changed in the case of deforming the first image by matching the position of the featured point of the object shape included in the first image with the position of the featured point of a reference shape; an image deformation part for deforming a second image by moving the position of the featured point of the object shape included in the second image by using the change amount; and a collation part for collating the deformed first image with the deformed second image, and for outputting the collation result. <P>COPYRIGHT: (C)2009,JPO&INPIT

Description

  The present invention relates to an image collation device that collates objects included in an image.

  As a conventional image matching device, a face image transformed so that the feature point of the face image matches the feature point of the standard face image, and the feature point of the newly input face image are used as the standard face image. There is one that performs personal identification based on a correlation value with a face image deformed so as to coincide with the feature point (see, for example, Patent Document 1). FIG. 11 shows a conventional image matching apparatus described in Patent Document 1.

In FIG. 11, an image input unit 1111 captures a face image of a person to be identified. The primary face normalization unit 1112 normalizes the face image of the person to be identified so that the positions of the eyes and the mouth are in a predesignated position. The feature point extraction unit 1113 extracts feature point position information from the normalized face image. The secondary face normalization unit 1115 normalizes the feature points detected by the feature point extraction unit 1113 so as to come to the standard human face feature point positions registered in the standard face image feature point database unit 1114. Deform the face image. The image correlation calculation unit 1117 includes a reference face image (database unit 1116) for each person deformed so as to match the feature points of a standard face image, and the face image deformed by the secondary face normalization unit 1115. Find the correlation. Further, the displacement amount calculation unit 1121 calculates a displacement amount between the position information extracted by the feature point extraction unit 1113 and the feature point coordinates of the standard face. Then, the displacement amount correlation calculation unit 1123 obtains a correlation between the displacement amount of the feature point of the face image for each person with respect to the feature point of the standard face image and the displacement amount of the feature point calculated by the displacement amount calculation unit 1121. Based on the two correlation values calculated by the image correlation calculation unit 1117 and the displacement amount correlation calculation unit 1123, the determination unit 1118 specified the face person to whom the image was input from the image input unit 1111.
Japanese Patent Laid-Open No. 11-161791

  However, in the conventional image matching device described above, the correlation between the face image calculated by the image correlation calculation unit 1117 and the correlation between the displacement amount of the feature point position calculated by the displacement amount correlation calculation unit 1123 tend to be different. It has a correlation. Here, there was a problem that two different correlations had to be integrated.

  The present invention has been made in view of this point, and an object of the present invention is to provide an image collation apparatus that collates two deformed images and outputs a collation result.

  The image collating apparatus according to the present invention changes the first image that is changed when the first image is deformed by matching the position of the feature point of the object shape included in the first image with the position of the feature point of the reference shape. A change amount database for storing a change amount of the position of the feature point of the object shape included in the image, and the position of the feature point of the object shape included in the second image by using the change amount An image deforming unit that deforms the image of the image, and a collation unit that collates the deformed first image with the deformed second image and outputs a collation result.

  The image matching apparatus according to the present invention is configured such that the position of the feature point of the object shape changed when the registered image is deformed by matching the position of the feature point of the object shape included in the registered image with the position of the feature point of the reference shape A change amount database that stores a change amount of the image, a deformation registration image database that stores the deformed registered image as a deformation registration image, a collation image input unit that inputs an image as a collation image, and an object included in the collation image A matching feature point detection unit that detects a feature point of a shape as a matching feature point, and a matching image that generates a modified matching image by deforming the matching image by moving the position of the matching feature point using the change amount The configuration includes a deformation unit, and a collation unit that collates the deformation registration image and the deformation collation image and outputs a collation result.

  The image collating apparatus of the present invention includes a registered image, a registered image database that stores the position of the feature point of the object shape included in the registered image as the position of the registered feature point, and a collated image input unit that inputs the image as a collated image A matching image feature point detection unit that detects a feature point of the matching image as a matching feature point, a reference feature point database that stores a position of a feature point of a reference shape as a position of a reference feature point, and A collation image deformation unit that deforms the collation image by matching a position with the position of the reference feature point, generates a deformation collation image, and outputs a change amount of the position of the collation feature point and the deformation collation image; A change amount database for storing the output change amount, and the registered image by moving the position of the registered feature point using the change amount stored in the change amount database. Deformed and a registered image deformation unit configured to generate a modified reference image, the collated deformation collation image and said modified registration image, employs a configuration having a, a collating unit for outputting a verification result.

  According to the present invention, two deformed images can be collated and a collation result can be output.

  Hereinafter, the present embodiment will be described with reference to the drawings. Hereinafter, in this embodiment, the object to be collated is described as a human face. However, this embodiment can apply not only an object such as an animal or the whole body of a person but also an object such as an animal or a part of an object such as a human hand or an animal's face as an object to be collated.

(Embodiment 1)
FIG. 1 is a block diagram showing a configuration of an image collating apparatus according to Embodiment 1 of the present invention.

  In FIG. 1, reference numeral 100 denotes a collation image input unit. The collation image input unit 100 captures an image including a collation face image (object shape) as a collation image using a camera or the like, and outputs the collation image.

  Reference numeral 101 denotes a collation image feature point detection unit that detects a face image (face area) from an image photographed by the collation image input unit 100 and outputs a face image and a collation feature point. The matching image feature point detection unit 101 detects feature points such as eyebrows, eyebrows, eyes, corners of eyes, both ends of the mouth, points on the contour of the face, and the like as matching feature points. The position is estimated, and the face image and matching feature points are output.

  102, the face image output from the matching image feature point detection unit 101 and the position of the matching feature point are input, and the face image is deformed by matching each matching feature point with a predesignated position. It is a collation image deformation | transformation part which outputs an image as a deformation | transformation collation image.

  Reference numeral 103 denotes a collation unit that collates the deformation collation image output from the collation image deformation unit 102 and the deformation registration image registered in advance in the deformation registration image database 104.

  104 is modified by matching the position of the feature point of the face image included in the registered image, which is an image photographed with a camera or the like, or the photographed image, with the position of the feature point of the standard face image (reference shape). 10 is a modified registered image database for storing registered images.

  Reference numeral 105 denotes a registration that has changed when the registered image is deformed by matching the position of the feature point of the face image included in the registered image (first image) with the position of the feature point of the reference face image (reference shape). It is a change amount database that stores a change amount (movement amount) of a feature point position of a face image included in an image.

  In such a configuration, the collation image input unit 100 captures an image including a face image as a collation image. The collation image feature point detection unit 101 detects a face area included in the collation image, and the eyebrows, eyebrows, eyes, corners, nose, left and right ends of the mouth, upper and lower ends of the mouth, and face Matching feature points such as points on the contour are detected.

  Here, in order to facilitate the detection of the face area at the time of shooting, as shown in FIG. 3, a rectangular area is displayed at the time of shooting, and the user is cautioned at the time of shooting so that the face fits in the rectangular area. Also good. Further, a face region may be automatically detected using a method such as template matching or boosting.

  When detecting the coordinates of feature points in the image area from the face area, you may detect eyebrow area, eye area, mouth area, etc. based on template matching and edges, Active Shape Models, Active Appearance Models, etc. A method of creating a face model in advance and detecting the feature point position by matching with the model may be used.

  When detecting the coordinates of feature points in the image plane and the position in the depth direction representing the unevenness of the face from the face area, it is preferable to use a method such as Active Shape Models, Active Appearance Models, or Bundle adjustment.

  FIG. 4 is a diagram illustrating the amount of change in the position of each feature point when a registered image of a face facing the front is transformed into a standard face shape such as an average face. The change amount database 105 stores a change amount for each person to be registered in advance. The matching image deformation unit 102 uses the amount of change (a set of changes) to move the position of the detected matching feature point, thereby deforming the entire matching image for matching, and deforming matching. Generate an image.

  In FIG. 4, x and y coordinates are coordinates on a plane parallel to the image plane, and z coordinates are coordinates perpendicular to the image plane, that is, coordinates representing the unevenness of the face in the depth direction of the face. In FIG. 4, the coordinates are three-dimensional coordinates, but two-dimensional coordinates including only x and y coordinates may be used. In the case of two-dimensional coordinates, the matching image deformation unit 102 moves the position of the matching feature point in the image plane to deform the matching image. In the case of three-dimensional coordinates, the collation image deformation unit 102 moves the position of the collation feature point in consideration of the three-dimensional shape of the face, and deforms the collation image. In the case of two-dimensional coordinates, the amount of processing is relatively small compared to three-dimensional coordinates, and high-speed alignment is possible. In the case of three-dimensional coordinates, it is possible to perform accurate positioning considering the unevenness of the face.

  The image deformation performed by the collation image deformation unit 102 will be described more specifically. First, the collation image deformation unit 102 adds each change amount stored in the change amount database 105 to the coordinates of each collation feature point detected by the collation image feature point detection unit 101, and determines the position of the feature point after deformation. The entire collation image is deformed. When the coordinates of the feature points are three-dimensional coordinates, the collation image deformation unit 102 projects the three-dimensional coordinates onto the image surface and obtains the two-dimensional coordinates in the image surface. The collation image deformation unit 102 gives the pixel value of the position of the feature point before deformation to the coordinate position obtained here.

<Comparison of light and shade patterns of image>
Next, an example in which face image matching is performed by comparing the shading pattern of the deformation registered image and the shading pattern of the deformation matching image will be described.

  The shading pattern of the face image is obtained by obtaining the pixel value of the area surrounded by the three feature points over the entire face area.

For example, as shown in FIG. 5, the collation image deformation unit 102 obtains the sum of pixel values for each triangular region that connects three feature points. FIG. 5 shows the shape of the face after deformation by the collation image deformation unit 102. Specifically, the triangle abc in FIG. 5 will be described. If the coordinates of the vertices a, b, and c are (x _a , y _a ), (x _b , y _b ), and (x _c , y _c ), the arbitrary coordinates in the triangle abc are (x, y) , (X _a , y _a ), (x _b , y _b ), and (x _c , y _c ) can be expressed as the following two equations.
(X, y) = α ( x a, y a) + β (x b, y b) + γ (x c, y c) ( Equation 1)
α + β + γ = 1 (Formula 2)
The coordinates of the feature points after deformation are obtained using α, β, and γ that satisfy these two expressions.

Next, the coordinates of the feature points before deformation are obtained using α, β, and γ. The coordinates of the vertices a, b, and c before transformation are (x ′ _a , y ′ _a ), (x ′ _b , y ′ _b ), and (x ′ _c , y ′ _c ), respectively, and inside the triangle abc before transformation If any coordinate of is (x ′, y ′), it can be expressed as follows.
(X ′, y ′) = α (x ′ _a , y ′ _a ) + β (x ′ _b , y ′ _b ) + γ (x ′ _c , y ′ _c ) (Formula 3)
α + β + γ = 1 (Formula 2)
At this time, the pixel value of the coordinate (x, y) after deformation is set as the pixel value of (x ′, y ′) before deformation.

  When the modified collation image for collation is generated in the collation image deformation unit 102 as described above, the collation unit 103 collates the deformation registration image stored in the deformation registration image database 104 with the deformation collation image. . As a result, the two images can be verified by applying a deformation reflecting the geometric shape characteristics of the object to be verified to the verification image. Here, the matching method may be any method in pattern matching. For example, the collation unit 103 collates the similarity between the light and shade patterns of two images, and determines whether the face images are the same person. Specifically, the similarity of the gray value at the same pixel position is collated between two images. Here, as a method for calculating the similarity, one vector is formed by arranging the grayscale values of one image in a line in the raster scan order, and the similarity between the two images is determined by the correlation value of the two vectors. There are methods.

  Here, the deformation registration image is an image obtained by deforming the registration image into a standard face shape by the above-described method. As the standard face shape, an arbitrary face shape may be used, but an average face shape may be used. As described above, the amount of change (Δ) required when the face image of the registered image is converted into the reference shape is applied to the collated image, so that the collated image can be the face of the same person as the registered image as shown in FIG. For example, a deformation matching image similar to the face shape of the reference shape is obtained. Also, as shown in FIG. 6, if the collation image is a face of a person (Mr. A) different from the registered image, that is, the collation image is a face different in shape from the face of the registration image (Mr. B), the amount of deformation When the collation image is deformed by Δ2, a face collation image having a shape different from the reference shape is obtained. Thus, the matching unit 103 performs matching of the shading pattern feature of the image and matching of the shape feature of the face with respect to the deformation registered image and the deformation matching image. In the example illustrated in FIG. 6, the collation unit 103 collates whether the collation image is Mr. A or Mr. B. And the collation part 103 outputs the signal which shows that a collation image is Mr. A as a result.

  Generally, when you smile, your mouth tends to go up. The comparison image shown in FIG. 6 shows a difference (change) in the shape of the mouth compared to the registered image of Mr. A's normal face. Also, when there is no change in facial expression, there is no difference in the mouth. Therefore, when collating the deformation registration image and the deformation collation image, the collation accuracy is improved by comparing other shapes excluding the shape of the mouth and collating the deformation registration image and the deformation collation image.

  Here, the face shape and the reference shape included in the deformation registration image are the same, but in the present embodiment, the shading pattern matching between the deformation registration image and the deformation collation image is performed. The image database 104 needs to store a plurality of modified registration images for a plurality of registration subjects.

  Then, the collation image deformation unit 102 deforms one collation image with respect to a plurality of registration images using the change amount corresponding to each deformation registration image. The collation unit 103 collates the deformation registration image with one deformation collation image deformed using the change amount corresponding to the deformation registration image, and outputs a face image collation result. The output result may be displayed on the display, the output result may be notified by voice, or the output result may be used for other collation processing.

  FIG. 7 is a diagram for explaining an example of collating whether or not the collation image is a specific person. In the example shown in FIG. 7, it is collated whether or not the collation image is Mr. B. The collation image deformation unit 102 deforms the collation image with the change amount Δ2 of Mr. B. The collation unit 103 collates the shade pattern feature of the image and the facial shape feature of the deformation registration image of Mr. B and the image of the deformation collation image. The collation unit 103 outputs a signal indicating that it is not Mr. B as the collation result.

  As described above, in the present embodiment, by comparing the deformation registered image and the deformation matching image, which integrate the correlations of the light and shade patterns related to the two object images to be compared with the correlations having different tendencies of the feature point positions. The collation result can be output.

(Modification)
In the present embodiment, the registered image is transformed into a reference shape, and the collation image is transformed using the deformation amount (Δ) of the registered image. However, the present invention is not limited to this, and the registered image may be deformed and collated using the amount of change (Δ ′) obtained when the collated image is transformed into the reference shape. Further, in the present embodiment, the case of collation has been described, but the same modification may be performed at the time of learning.

(Embodiment 2)
FIG. 8 is a block diagram of an image collating apparatus according to Embodiment 2 of the present invention. In the present embodiment, a registered image input unit 110, a registered image feature point detection unit 111, a registered image deformation unit 112, and a reference feature point database 113 are added as new configurations to the configuration of the first embodiment. Other configurations are the same as those of the first embodiment, and thus the description thereof is omitted.

  The registered image input unit 110 captures an image including a registration face image (object shape) as a registered image using a camera or the like, and outputs the registered image.

  The registered image feature point detection unit 111 detects a face image (face region) from the image captured by the registered image input unit 110, and outputs a face image as a registered image and a registered feature point. The registered image feature point detection unit 111 detects, as registered feature points, feature points such as the eyebrows, the eyebrows, the eyes, the corners of the eyes, the ends of the mouth, and the points on the contour of the face from the face image. The position is estimated, and the registered image and registered feature points are output to the registered image transformation unit 112.

  The registered image transformation unit 112 receives the registered image output from the registered image feature point detection unit 111 and the position of the registered feature point, and matches each registered feature point with the position of the reference feature point specified in advance. Is used to generate a deformation registration image. The registered image deformation unit 112 stores the change amount of the position of the registered feature point in the change amount database 105, and stores the deformation registered image in the deformation registered image database 104.

  The reference feature point database 113 stores the position of the feature point of the reference shape as the position of the reference feature point.

  According to the present embodiment, the change amount of the position of the registered feature point of the person to be collated can be stored in the change amount database 105, and the deformation registration image can be newly registered in the deformation registration image database 104.

(Embodiment 3)
FIG. 9 is a block diagram of an image collating apparatus according to Embodiment 3 of the present invention. In the first embodiment, the example in which the collation image is deformed by moving the position of the collation feature point using the change amount of the position of the registered feature point has been described. In the present embodiment, an example will be described in which the collation image is deformed by moving the position of the registered feature point using the amount of change in the position of the collation feature point.

  The collation image input unit 100 captures an image including a collation face image (object shape) as a collation image using a camera or the like, and outputs the collation image.

  The collation image feature point detection unit 101 detects a face image (face region) from the image captured by the collation image input unit 100, and outputs a face image as a collation image and a collation feature point. The matching image feature point detection unit 101 detects feature points such as eyebrows, eyebrows, eyes, corners of eyes, both ends of the mouth, points on the contour of the face, and the like as matching feature points. A position is estimated and a collation image and a collation feature point are output.

  The collation image deformation unit 102 receives the collation image output from the collation image feature point detection unit 101 and the position of the collation feature point, and matches each collation feature point to the position of the reference feature point designated in advance. Is deformed to generate a modified collation image. The collation image deformation unit 102 stores the amount of change in the position of the collation feature point in the change amount database 105 and outputs the deformation collation image to the collation unit 103.

  The reference feature point database 113 stores the position of the feature point of the reference shape as the position of the reference feature point.

  The registered image database 114 stores registered images and registered image feature points in advance.

  The registered image transformation unit 112 inputs the registered image output from the registered image database 114 and the position of the registered feature point, and moves the position of each registered feature point using the change amount stored in the change amount database 105. Then, the registered image is deformed, a deformed registered image is generated, and the deformed registered image is output to the matching unit 103.

  The collation unit 103 collates the deformation collation image and the deformation registration image, and outputs a collation result.

  According to the present embodiment, it is possible to collate a plurality of deformation registered images and deformation collation images only by storing the amount of change of the collation feature point position of the collation target person in the change amount database 105.

(Embodiment 4)
FIG. 10 is a block diagram of an image collating apparatus according to Embodiment 4 of the present invention. In the present embodiment, a registered image input unit 110 and a registered image feature point detection unit 111 are added to the configuration of the third embodiment as new configurations. Other configurations are the same as the configurations of the third embodiment, and thus the description thereof is omitted.

  The registered image input unit 110 captures an image including a registration face image (object shape) as a registered image using a camera or the like, and outputs the registered image.

  The registered image feature point detection unit 111 detects a face image (face region) from the image captured by the registered image input unit 110, and outputs a face image as a registered image and a registered feature point. The registered image feature point detection unit 111 detects, as registered feature points, feature points such as the eyebrows, the eyebrows, the eyes, the corners of the eyes, the ends of the mouth, and the points on the contour of the face from the face image. The position is estimated, and the registered image and registered feature points are output to the registered image database 114.

  The registered image database 114 stores registered images and registered image feature points output from the registered image feature point detection unit 111.

  According to the present embodiment, the registered image and registered image feature points of the person to be collated can be newly registered in the registered image database 114.

  The image matching apparatus according to the present invention can be applied to matching any object with little change in shape and shading pattern, and can be applied to any object matching.

1 is a block diagram of an image matching device according to Embodiment 1 of the present invention. The figure showing an example of the feature point detected with the image collation apparatus in Embodiment 1 of this invention. The figure showing an example of the display at the time of imaging | photography with the image collation apparatus in Embodiment 1 of this invention. The figure showing an example of the variation | change_quantity of the feature point of a registration image with the image collation apparatus in Embodiment 1 of this invention. The figure showing an example of the area | region division at the time of deform | transforming a face with the image collation apparatus in Embodiment 1 of this invention. The figure showing the collation example 1 of the image collation apparatus in Embodiment 1 of this invention. The figure showing the collation example 2 of the image collation apparatus in Embodiment 1 of this invention. The block diagram of the image collation apparatus in Embodiment 2 of this invention The block diagram of the image collation apparatus in Embodiment 3 of this invention The block diagram of the image collation apparatus in Embodiment 4 of this invention Block diagram of a conventional image matching device

Explanation of symbols

DESCRIPTION OF SYMBOLS 100 Collation image input part 101 Collation image feature point detection part 102 Collation image deformation | transformation part 103 Collation part 104 Deformation registration image database 105 Change amount database 110 Registration image input part 111 Registration image feature point detection part 112 Registration image deformation | transformation part 113 Reference | standard feature point Database 114 Registered image database

Claims (17)

The feature of the object shape included in the first image changed when the first image is deformed by matching the position of the feature point of the object shape included in the first image with the position of the feature point of the reference shape A change amount database for storing the change amount of the point position;
An image deformation unit that deforms the second image by moving the position of the feature point of the object shape included in the second image using the change amount;
A collation unit that collates the deformed first image and the deformed second image and outputs a collation result;
An image collating apparatus comprising: The position of the feature point is a two-dimensional coordinate in the image.
The image collating apparatus according to claim 1. The position of the feature point is a three-dimensional coordinate on the object,
The image collating apparatus according to claim 1. The collation unit collates the object shape of the deformed first image with the object shape of the deformed second image, and outputs a collation result.
The image collating apparatus according to claim 1. The collation unit collates the shade pattern of the deformed first image with the shade pattern of the deformed second image, and outputs a collation result.
The image collation apparatus in any one of Claims 1-4. The first image is one or more registered images;
The second image is one collation image,
The change amount database stores a plurality of change amounts corresponding to a plurality of registered images.
The image collation apparatus according to claim 1. The image deformation unit deforms one collation image using each change amount corresponding to each deformed registered image,
The collation unit collates the deformed registered image with one collation image deformed using a change amount corresponding to the deformed registered image, and outputs a collation result.
The image collating apparatus according to claim 6. A specifying means for specifying a set of change amounts corresponding to a specific registered image among the plurality of change amounts;
The image deformation unit deforms one collation image using a change amount corresponding to the specific registered image,
The collation unit collates the deformed specific registration image with one collation image deformed using a change amount corresponding to the deformed specific registration image, and outputs a collation result.
The image collating apparatus according to claim 6. The first image is one collation image;
The second image is one or more registered images;
The change amount database stores a change amount corresponding to one collation image,
The image deformation unit deforms each registered image using a change amount corresponding to the one collation image,
The collation unit collates each deformed registered image and the one collated image, and outputs a collation result.
The image collating apparatus according to claim 1. The amount of change for storing the amount of change in the position of the feature point of the object shape changed when the registered image is deformed by matching the position of the feature point of the object shape included in the registered image with the position of the feature point of the reference shape A database,
A modified registered image database for storing the modified registered image as a modified registered image;
A collation image input unit for inputting an image as a collation image;
A matching feature point detection unit that detects a feature point of an object shape included in the matching image as a matching feature point;
A collation image deformation unit that deforms the collation image by moving the position of the collation feature point using the change amount, and generates a deformation collation image;
A collation unit that collates the deformation registration image and the deformation collation image and outputs a collation result;
An image collating apparatus comprising: The position of the feature point is a two-dimensional coordinate in the image.
The image collating apparatus according to claim 10. The position of the feature point is a three-dimensional coordinate on the object,
The image collating apparatus according to claim 10. A registered image input unit for inputting the registered image;
A registered image feature point detection unit that detects a feature point of an object shape included in the registered image as a registered feature point;
The registered image is deformed by matching the position of the registered feature point with the position of the feature point of the reference shape to generate the deformed registered image, and the change amount of the position of the registered feature point is stored in the change amount database. A registered image deformation unit for storing the deformation registered image in the deformed image database;
A reference feature point database for storing the position of the feature point of the reference shape as the position of the reference feature point;
The image collating apparatus according to claim 10, further comprising: A registered image, and a registered image database that stores the position of the feature point of the object shape included in the registered image as the position of the registered feature point;
A collation image input unit for inputting an image as a collation image;
A matching image feature point detection unit for detecting a feature point of the matching image as a matching feature point;
A reference feature point database for storing the position of the feature point of the reference shape as the position of the reference feature point;
A matching image that deforms the matching image by matching the position of the matching feature point with the position of the reference feature point, generates a modified matching image, and outputs the amount of change in the position of the matching feature point and the modified matching image A deformation part;
A change amount database for storing the output change amount;
A registered image deformation unit that deforms the registered image by moving the position of the registered feature point using a change amount stored in the change amount database, and generates a deformed registered image;
A collation unit that collates the deformation collation image and the deformation registration image and outputs a collation result;
An image collating apparatus comprising: The position of the feature point is a two-dimensional coordinate in the image.
The image collating apparatus according to claim 14. The position of the feature point is a three-dimensional coordinate on the object,
The image collating apparatus according to claim 14. A registered image input unit for inputting the registered image;
A registered image feature point detection unit that detects a position of a registered feature point from the registered image and stores it in the registered image database together with the registered image;
The image collating apparatus according to claim 14, further comprising:

Images