JP2010231277A - Image processing apparatus, memory controller, and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To give a priority to the results of similarity determination based on a first feature value whose priority is high, and to adopt the results based on the feature value when a clear difference is generated between the image whose similarity is the highest and the other images. <P>SOLUTION: A similarity calculation unit 140 calculates similarity between a feature value extracted from a new stamp image and a feature value extracted from each of a plurality of registered stamp images for each type of feature value. When two or more similarities which are equal to or more than a threshold are calculated from the feature values whose calculated similarities are equal to or more than the threshold, and whose priorities are the highest, a decision unit 150 determines whether a difference between the highest similarity and the other similarities exceeds a predetermined size. When only one similarity which is equal to or more than the threshold is calculated from the feature values whose priorities are set or when the decision of the decision unit 150 is positive, an output unit 160 adopts the feature value, and when the decision is negative, the output unit adopts the feature value whose priority is lower than that, and displays registration stamp images in the order of similarities. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to an image processing device, a storage control device, and a program.

  There is a technique for extracting feature amounts from respective images that are targets of similarity determination, and determining the similarity of both images based on the similarity between the extracted feature amounts. The technique proposed in Patent Document 1 obtains similarity using a plurality of feature amounts, and searches for images having a high similarity overall by performing unique weighting for each feature amount.

Japanese Patent No. 3812161

  The object of the present invention is to give priority to the result of similarity determination based on the first feature amount having a high priority, but when there is a clear difference between the image having the highest similarity and the other images. The result by the feature amount is adopted.

  In order to solve the above-described problem, an image processing apparatus according to a first aspect of the present invention includes a first feature amount representing an image feature and a feature amount representing an image feature. Extraction means for extracting from the first image a plurality of types of feature quantities including at least a second feature quantity set with a priority lower than the set priority; and a plurality of second features different from the first image. The feature amount extracted from each of the two images is acquired, and the similarity between each acquired feature amount and the feature amount extracted by the extraction unit is obtained for each of the first and second feature amounts. When two or more similarities greater than or equal to the threshold are calculated by the similarity calculation means to be calculated and the similarity calculation means, the difference between the highest similarity and the other similarities among the similarities greater than or equal to the threshold Whether the feature size exceeds a predetermined size. A determination unit that determines each time, and a similarity that is equal to or greater than the threshold value of 1 is calculated by the similarity calculation unit for the first feature amount, or it is determined that the size exceeds the size by the determination unit If any one of the conditions is satisfied, an output unit that outputs information corresponding to the second image according to the similarity calculated for the first feature amount is provided.

  According to a second aspect of the present invention, in the configuration of the first aspect, a priority is set for each of the plurality of types of feature amounts, and the similarity calculation unit is configured to each of the plurality of types of feature amounts. Whether the output means calculates a similarity that is greater than or equal to the threshold value of 1 by the similarity calculation means, or determines that the size exceeds the size by the determination means The information is output for the feature quantity having the highest priority among the types of feature quantities satisfying any one of the conditions.

An image processing apparatus according to a third aspect of the present invention is the image processing apparatus according to the first or second aspect, wherein the output means is a feature quantity for which the similarity that is equal to or greater than the threshold value of 1 is calculated by the similarity calculation means, and When there is no feature quantity determined to exceed the size by the judgment means, the information is output for a type of feature quantity that satisfies a predetermined condition.
An image processing apparatus according to a fourth aspect of the present invention is the image processing apparatus according to the third aspect, wherein the output means has the highest similarity among the similarity calculated by the similarity calculation means equal to or greater than the threshold. It is characterized in that the information is output for a feature quantity of a kind that satisfies the condition that the magnitude of the difference from other similarities is maximized.

An image processing apparatus according to a fifth aspect of the present invention is the image processing apparatus according to any one of the first to fourth aspects, wherein a reception unit that receives an operation of designating priority for the plurality of types of feature amounts by a user; And priority setting means for setting the priority according to the specified operation received by each of the plurality of types of feature amounts.
An image processing apparatus according to a sixth aspect of the present invention comprises the semantic information obtaining means for obtaining the semantic information assigned to the meaning of the second image in the configuration of any one of the first to fifth aspects, The output means outputs control information for causing the display means to display the second image and information corresponding to the semantic information acquired by the semantic information acquisition means and assigned to the meaning of the second image. It is characterized by that.

A storage control device according to claim 7 of the present invention represents the image processing device having the configuration of claim 6, image information representing the first image, and the plurality of types of feature amounts extracted by the extraction means. Storage control means for associating information with semantic information representing the meaning assigned to the first image and storing the information in the storage means, in accordance with the control information output by the output means of the image processing apparatus Storage control means for storing semantic information designated by a user as a representation of the meaning assigned to the first image according to the display content displayed on the display means.
According to an eighth aspect of the present invention, there is provided a program for causing a computer to execute a first feature amount representing a feature of an image and a feature amount representing a feature of the image, and a priority set to the first feature amount. Extracting means for extracting a plurality of types of feature quantities including at least a second feature quantity set with a low priority from the first image, and extracting from each of a plurality of second images different from the first image A similarity calculation unit that acquires the obtained feature amount and calculates a similarity between each acquired feature amount and the feature amount extracted by the extraction unit for each of the first and second feature amounts; , When two or more similarities greater than or equal to the threshold are calculated by the similarity calculating means, a size in which the difference between the highest similarity and the other similarities is determined among the similarities equal to or greater than the threshold Judgment whether or not it exceeds each feature type A similarity that is greater than or equal to the threshold value of 1 is calculated by the similarity calculation unit for the first feature amount, or it is determined that the size exceeds the size by the determination unit If the above condition is satisfied, the information processing apparatus functions as an output unit that outputs information according to the second image in accordance with the similarity calculated for the first feature amount.

According to the first aspect of the invention, priority is given to the result of similarity determination based on the first feature amount having a high priority, but there is a clear difference between the image having the highest similarity and the other images. When it occurs, the result of the feature amount can be adopted.
According to the second aspect of the present invention, priority is given to the result of the similarity determination based on the feature amount having a higher priority according to the priority set for each type of feature amount, compared to the case where the present configuration is not provided. It is possible to adopt a result based on a feature amount in which a clear difference has occurred between an image having the highest similarity and other images.
According to the invention of claim 3, when there is no feature amount that has a clear difference between the image having the highest similarity and the other images, the feature amount that satisfies the predetermined condition is determined. The result of similarity determination can be adopted.
According to the invention according to claim 4, when there is no feature quantity that has a clear difference between the highest similarity and the others, the result of the feature quantity having the highest difference is adopted. can do.
According to the invention which concerns on Claim 5, the priority designated by the user can be set and the result of the similarity determination which changes according to the setting can be output.
According to the sixth aspect of the invention, information representing the meaning assigned to the second image can be output.
According to the invention of claim 7, the result of similarity determination can be displayed, and the meaning represented by the semantic information specified by the user according to the display content can be assigned to the first image.
According to the invention according to claim 8, priority is given to the result of similarity determination based on the first feature amount, but when there is a clear difference between the image having the highest similarity and the other images. The result by the feature amount can be adopted.

It is the figure which showed the structure of the stamp registration apparatus which concerns on embodiment of this invention. It is the figure which showed an example of the stamp image of the embodiment. It is the figure which showed an example of the structure of the feature-value management table. It is the figure which showed an example of the structure of a priority management table. It is the figure which showed an example of the structure of a confidentiality management table. It is a block diagram which shows an example of a functional structure of the stamp registration apparatus of the embodiment. It is a flowchart which shows the flow of the process which the control part of the embodiment performs. It is a flowchart which shows the flow of the process which the control part of the embodiment performs. It is the figure which showed an example of the new stamp image of the embodiment. It is a figure explaining the feature-value extracted from the new stamp image of the embodiment. It is the figure which showed an example of the display content displayed on a UI part. It is the figure which showed an example of the display content displayed on a UI part.

(A) Configuration An embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram showing a configuration of a stamp registration apparatus according to an embodiment of the present invention. The stamp registration apparatus 10 according to the present embodiment is an example of a storage control apparatus according to the present invention, and is an apparatus that registers an image obtained by reading (that is, imprinting) a stamp (a seal) such as a seal stamp. is there. The image obtained by reading the stamp imprint is hereinafter referred to as a “stamp image”. The stamp image registered in the stamp registration apparatus 10 is used, for example, when a document is digitized and managed, and the attribute of the digitized information is determined from the stamp impressed on the document, and the digitized information is determined for each determined attribute. There are uses such as classification and management.

The stamp registration apparatus 10 includes a control unit 100, a storage unit 200, a UI (User Interface) unit 300, and a reading unit 400. Each component of the stamp registration apparatus 10 is connected via a bus.
The control unit 100 includes a calculation device and a memory including a CPU (Central Processing Unit) and an ASIC (Application Specific Integrated Circuit), and controls each unit of the stamp registration device 10 according to a calculation program stored in the memory or the storage unit 200. . The storage unit 200 is an example of a storage unit of the present invention, and includes a large-capacity storage device such as an HDD (Hard Disk Drive). The storage unit 200 includes image information representing a control program and a stamp image, a feature amount management table 210, a priority level, and the like. A management table 220 and a confidentiality management table 230 are stored. The UI unit 300 is an example of a display unit according to the present invention. The UI unit 300 notifies information by an image or sound based on control information from the control unit 100, and includes a touch panel and various buttons as operation units. The UI unit 300 supplies operation information representing the operation content to the control unit 100 in accordance with a user operation. The reading unit 400 is a reading unit that optically reads the surface of a recording medium such as paper, and generates image information representing the read image and supplies the image information to the control unit 100. The reading unit 400 is a scanner device built in or externally attached to the stamp registration device 10, for example, and generates image information representing an image obtained by reading the surface of the recording medium and supplies the image information to the control unit 100.

  FIG. 2 is a diagram illustrating an example of a stamp image according to the present embodiment. The stamp image ST1 shown on the left side of the figure is an image marked "important" in a substantially rectangular frame, and the stamp image ST2 shown on the right side of the figure is an image marked "secret" inside the circled circle. It is. These stamp images ST1 and ST2 are registered in the stamp registration apparatus 10, and identification information (hereinafter referred to as “ID”) for identifying each stamp image is assigned to each stamp image. As shown in FIG. 2, an ID “stamp # 1” is assigned to the stamp image ST1, and an ID “stamp # 2” is assigned to the stamp image ST2.

Next, configurations of the feature amount management table 210, the priority management table 220, and the confidentiality management table 230 stored in the storage unit 200 will be described. Each of these tables describes information related to the stamp image.
FIG. 3 is a diagram illustrating an example of the configuration of the feature amount management table 210.
As shown in the figure, in the feature quantity management table 210, “ID” and “feature quantity” are described in association with each other. “ID” is identification information for identifying the above-described stamp image. “Feature amount” is information representing the feature of the stamp image, and the feature amount management table 210 describes a plurality of types of feature amounts. There are four types of feature amounts in this embodiment: “complexity”, “color”, “shape”, and “dimension”. The “complexity” is information indicating the degree of complexity of the stamp image, and has a value corresponding to the amount of harmonic components included in the spatial frequency of the stamp image, for example. Here, relative values “1” to “15” are used as the complexity, and the larger the value, the more complex the image. “Color” is the color of the stamp image itself. “Shape” is information representing the outer shape of the stamp image. “Dimension” is information indicating the size of the outer shape of the stamp image. For example, a rectangular stamp image is represented by vertical and horizontal lengths, and a circular stamp image is represented by a diameter length. Is done.
The feature quantity management table 210 shown in FIG. 3 describes the feature quantities of the stamp images ST1 and ST2 shown in FIG. As shown in FIG. 3, the feature amount of the stamp image ST1 has a complexity of “12”, a color of “red”, a shape of “rectangular (horizontally long)”, and a dimension of “vertical × horizontal: 13 mm × 27 mm ”. The feature amount of the stamp image ST2 has a complexity of “10”, a color of “blue”, a shape of “circle”, and a dimension of “diameter: 16 mm”.

FIG. 4 is a diagram showing an example of the configuration of the priority management table 220.
As shown in the figure, in the priority management table 220, “feature” and “priority” are described in association with each other. The “feature amount” is the same as described above. The “priority” is a priority set for each of a plurality of types of feature amounts, and the smaller the value, the higher the priority. In the example of the figure, the priority of the complexity is “1”, the priority of the color is “2”, the priority of the shape is “3”, and the priority of the dimension is “4”. The former feature amount has a higher priority.

FIG. 5 is a diagram showing an example of the configuration of the confidentiality management table 230.
As shown in the figure, in the confidentiality management table 230, “ID” and “sensitiveness” are described in association with each other. “ID” is identification information for identifying the above-described stamp image. The “confidentiality” is an example of the semantic information of the present invention and represents the meaning assigned to the stamp image. In the example shown in the figure, the confidentiality “MIDDLE” is assigned to the ID “stamp # 1”, the confidentiality “HIGH” is assigned to the ID “stamp # 2”, and the confidentiality “LOW” is assigned to the ID “stamp # 3”. Is assigned, and the confidentiality “MIDDLE” is assigned to the ID “stamp # 4”. “HIGH”, “MIDDLE”, and “LOW” in descending order of confidentiality means that the confidentiality is high, medium, and low in order from the former. For example, the confidentiality of the document on which the stamp that is the basis of the stamp image ST1 is pressed is medium, and the confidentiality of the document that is pressed on the basis of the stamp image ST2 is high.
In this way, the storage unit 200 stores the image information representing the stamp image, the information representing the feature amount extracted from the stamp image, and the information representing the confidentiality assigned to the stamp image in association with each other. .
Next, functions related to registration of the stamp image of the stamp registration apparatus 10 will be described.

FIG. 6 is a block diagram illustrating an example of a functional configuration of the stamp registration apparatus 10.
As shown in the figure, the control unit 100 includes an extraction unit 110, a reception unit 120, a priority setting unit 130, a similarity calculation unit 140, a determination unit 150, an output unit 160, a confidentiality information acquisition unit 170, and a registration unit. 180 functions are realized. Each of these functions may be realized by one or a plurality of hardware circuits included in the control unit 10, may be realized by executing one or a plurality of programs by an arithmetic device, or a combination thereof. May be.
In the following description, a stamp image to be newly registered in the stamp registration apparatus 10 is referred to as a “new stamp image”. This new stamp image is an example of the first image of the present invention. A stamp image that has been registered in the stamp registration apparatus 10 and that is a target of similarity determination with a new stamp image is referred to as a “registered stamp image”. This registered stamp image is an example of the second image of the present invention. Here, a plurality of stamp images are already registered in the stamp registration apparatus 10.

The reading unit 400 reads an imprint on the recording medium, represents the read image, and generates image information including a new stamp image. The reading unit 400 supplies the generated image information to the extraction unit 110.
The extraction unit 110 acquires image information from the reading unit 400, extracts a feature amount from a new stamp image included in the acquired image information, and supplies the feature amount to the similarity calculation unit 140. Here, the extraction unit 110 performs various types of image processing on the image information, and extracts four types of feature amounts of “complexity”, “color”, “shape”, and “dimension”.

The accepting unit 120 is an example of accepting means of the present invention, and accepts an operation for designating priority for a plurality of types of feature amounts by the user. When the UI unit 300 is operated by the user and the operation of designating the priority for each feature amount of “complexity”, “color”, “shape”, and “dimension” is performed, the reception unit 120 Is obtained from the UI unit 300 and supplied to the priority setting unit 130.
The priority setting unit 130 is an example of a priority setting unit of the present invention, and sets a priority for each type of feature amount based on the operation information supplied from the reception unit 120. Specifically, the priority setting unit 130 rewrites the contents of the priority management table 220 so that the priority specified by the user is obtained.

  The similarity calculation unit 140 is an example of the similarity calculation unit according to the present invention. The similarity calculation unit 140 acquires feature amounts extracted from each of a plurality of registered stamp images, and the extracted feature amounts are extracted by the extraction unit 110. The similarity with the feature amount is calculated for each type of feature amount. The similarity calculation unit 140 supplies the calculated similarity to the determination unit 150.

  The determination unit 150 is an example of a determination unit according to the present invention. For the feature amount for which two or more similarities greater than or equal to the threshold are calculated by the similarity calculation unit 140, among the similarities, the highest similarity and the others It is determined whether or not the difference with the degree of similarity exceeds a predetermined size. The determination unit 150 supplies the determination result to the output unit 160. Note that if the registered stamp image is not specified for all feature amounts by the similarity calculating unit 140, the determining unit 150 supplies information indicating that the registered stamp image has not been specified to the output unit 160.

The output unit 160 is an example of the output unit of the present invention, and the feature amount set with a certain priority (an example of the first feature amount of the present invention) is equal to or greater than a threshold of 1 by the similarity calculation unit 140. If either the degree of similarity is calculated or if it is determined that the difference exceeds the size determined by the determination unit 150, the degree of similarity calculated for the type of feature amount is used. Information corresponding to the registered stamp image is output, and if the condition is not satisfied, information is output for the feature amount (an example of the second feature amount of the present invention) for which a lower priority is set. The output unit 160 outputs control information for controlling display of the UI unit 300 as information corresponding to the registered stamp image.
The confidentiality information acquisition unit 170 is an example of the semantic information acquisition means of the present invention, and is the confidentiality associated with the registered stamp image in the confidentiality management table 230, and the confidentiality assigned to each registered stamp image. Is obtained and supplied to the output unit 160.
The registration unit 180 is an example of a storage control unit according to the present invention, and registers a new stamp image in its own apparatus according to the content specified by the user according to the content displayed on the UI unit 300.

(B) Operation | movement of embodiment Next, operation | movement of embodiment is demonstrated along the content of each table illustrated in FIGS. 7 and 8 are flowcharts showing the flow of processing executed by the control unit 100 of the stamp registration apparatus 10.
Prior to the flows shown in FIGS. 7 and 8, the user operates the UI unit 300 at a certain time to perform an operation of specifying the priority for each feature amount. The control unit 100 sets the priority according to the operation content to each type of feature amount. At this time, for example, the user designates the priority so that the priority is higher for the feature amount that the result of similarity determination is to be emphasized. Here, as shown in the priority management table 220 of FIG. 4, it is assumed that the priority is designated by the user.
When a user newly registers a stamp image in the stamp registration apparatus 10, the user operates the UI unit 300 to switch the operation mode of the stamp registration apparatus 10 to an operation mode for registering a stamp image. Next, the user seals the stamp image on the sheet using the stamp to be registered, sets the sheet in the reading unit 400, and operates the UI unit 300 to start reading.
The control unit 100 reads the imprint by the reading unit 400 in accordance with operation information from the UI unit 300, generates image information representing the read image, and acquires image information including a new stamp image (step S1). Here, the control unit 100 acquires image information including the new stamp image STn shown in FIG. As shown in the figure, the new stamp image STn is an image in which “important” is marked inside the circle.

  Next, the control unit 100 performs various types of image processing on the image information including the new stamp image, and extracts a feature amount from the new stamp image STn (step S2). Here, the control unit 100 extracts four types of feature amounts shown in FIG. 10 and stores them in the memory. Specifically, the feature amount of the new stamp image STn has a complexity of “12”, a color of “red”, a shape of “circle”, and a dimension of “diameter: 16 mm”.

Next, the control unit 100 reads the feature amount management table 210 stored in the storage unit 200, acquires the feature amount of the registered stamp image, and stores the content in the memory (step S3). Next, the control unit 100 reads the priority management table 220 stored in the storage unit 200, acquires the priority of each feature amount, and stores the content in the memory (step S4). The contents of the feature quantity management table 210 and the priority management table 220 read in steps S3 and S4 are as shown in FIGS.
Next, the control unit 100 reads and acquires the threshold value determined for each feature amount from the storage unit 200 (step S5). This threshold value is a reference value for determining whether or not the new stamp image and the registered stamp image are similar. If the degree of similarity exceeds the threshold value, it is determined that the two images are similar. The threshold value is determined at the design stage for each feature amount and stored in the storage unit 200. However, the stamp registration device 10 has a configuration in which the threshold value is changed according to the setting contents of the user or service person. In this case, it is more preferable that the threshold value is set to a value suitable for the intended use and the characteristics of the components of the apparatus.

  The control unit 100 determines the priority based on the feature quantity having the highest priority among the feature quantities that have not yet been collated for the collation process for collating the feature quantity of the new stamp image with the feature quantity of the registered stamp image. Is calculated and collation processing is executed (step S6). Here, since the matching process has not been executed yet, the control unit 100 executes the matching process based on the “complexity” that is the feature quantity having the highest priority. In this collation processing, the control unit 100 calculates the similarity between the feature amount extracted from the new stamp image STn in step S2 and the feature amount acquired from each registered stamp image in step S5 for each type of feature amount. To do. Here, the control unit 100 calculates the similarity for the feature amount “complexity”. For the feature amount “complexity”, the control unit 100 calculates a difference in complexity between the new stamp image STn and each registered stamp image. For example, if the difference is “4” or less, the similarity is equal to or greater than a threshold value. It is judged that.

  Then, the control unit 100 identifies a registered stamp image in which the similarity between the new stamp image and the registered stamp image is greater than or equal to a threshold value according to the result of the collation process executed for the feature amount “complexity” (step S7). Here, it is assumed that the control unit 100 identifies three registered stamp images ST1, ST2, and ST3. Further, the control unit 100 has the highest similarity with the registered stamp image ST2, and thereafter, the similarity is higher in the order of the registered stamp images ST1 and ST3.

Next, the control unit 100 determines whether or not a registered stamp having a similarity equal to or greater than a threshold value is specified in step S7 (step S8). Here, since “YES” is determined in step S8, the control unit 100 proceeds to step S9.
Next, the control unit 100 determines whether or not the number of registered stamp images specified in step S7 is “2” or more (plural) (step S9). As described above, since there are three registered stamp images identified for the feature amount “complexity”, ST1, ST2, and ST3, the control unit 100 determines “YES” and proceeds to step 10.

The control unit 100 calculates the highest similarity corresponding to two or more registered stamp images identified as having high similarity to the new stamp image, the similarity calculated in step S6 being equal to or higher than the threshold, and others. It is determined whether there is a significant difference between the similarities (step S10). In this step S10, the control unit 10 is not a subtle difference that may be an error range between the registered stamp image determined to be most similar to the new stamp and the registered stamp image determined to be similar to the other, It is determined whether or not the magnitude of the difference between these similarities exceeds a predetermined condition. For the feature amount “complexity”, the control unit 100 calculates the Euclidean distance between predetermined vectors of the new stamp image STn and the registered stamp image. Then, the control unit 100 compares the Euclidean distance between the new stamp image STn and the registered stamp image ST2 having the highest similarity and the Euclidean distance between the new stamp image STn and the registered stamp image ST1 having the second highest similarity. (Distance ratio) is calculated. Then, the control unit 100 determines that there is a significant difference if the distance ratio exceeds “2”, and determines “YES” in step S10. On the other hand, if the distance ratio is “2” or less, it is determined that there is no significant difference, and the control unit 100 determines “NO” in step S10. Note that the distance calculated by the control unit 100 is not limited to the Euclidean distance, and for example, a known metric space such as the Mahalanobis distance may be used.
In the following description, when the control unit 100 identifies two or more registered stamp images in step S7, the size of the difference between the highest similarity corresponding to them and the other similarities is determined. Exceeding a given condition may be explained using the expression that a significant difference has occurred.

(B-1) When there is a significant difference When the calculated distance ratio is “2” or less, the control unit 100 determines the similarity between the registered stamp image ST2 having the highest similarity and the other registered stamp images. When it is determined that the magnitude of the difference from the similarity between ST1 and ST3 exceeds the determined condition and there is a significant difference between them (step S10; YES), the process proceeds to step S11. Next, the control unit 100 determines registered image candidate ranks in descending order of similarity according to the similarity order calculated in step S7 (step S11). Here, the control unit 11 determines the candidate rank by setting the stamp image ST2 as “first place”, the stamp image ST1 as “second place”, and the stamp image ST3 as “third place”.

  Then, the control unit 100 displays the registered stamp images in descending order of the determined candidate order, and displays information indicating the confidentiality assigned to each of the stamp images on the UI unit 300 (step S12). In step S12, the control unit 100 outputs control information for controlling the display of the UI unit 300 so that the registration stamp images are arranged in the descending order of the candidate order determined in step S11 and displayed on the UI unit 300. Thereby, the image shown in FIG. 11 is displayed.

  FIG. 11 is a diagram illustrating an example of display contents displayed on the UI unit 300 by executing step S12. The outermost rectangle shown in the figure represents the range of the display area of the UI unit 300. As shown in the figure, on the right side of the display area of the UI unit 300, a registered stamp image whose similarity is determined to exceed the threshold by the control unit 100 and the confidentiality assigned to the registered stamp image are attached. Soft buttons are arranged. Specifically, the upper row includes an image DST2 obtained by reducing the stamp image ST2 having the first candidate rank, and a soft button SBb having “Level: HIGH” indicating that the sensitivity is “HIGH”. In the lower row, an image DST1 obtained by reducing the stamp image ST1 having the second candidate rank, and a soft button SBc with “Level: MIDDLE” indicating that the sensitivity is “MIDDLE” are arranged. ing. Below these, a soft button SBd is arranged, and when this button is selected by the user, information on the above contents is displayed for the stamp image ST3 having the third candidate rank under the control of the control unit 100. On the left side of the display area of the UI unit 300, an image DSTn, which is labeled “input stamp” and represents a new stamp image STn, and a soft button SBa labeled “candidate level HIGH” are arranged. In this way, the control unit 100 arranges and displays the registered stamp images whose similarity exceeds the threshold value in descending order of the calculated similarity.

  Here, the candidate level means a confidentiality that is a candidate assigned to the new stamp image STn. In the example of FIG. 11, the candidate level is “HIGH”, and the confidentiality assigned to the registered stamp image that the control unit 100 has ranked first in the candidate ranking is assigned. Since there is often a correlation between the feature amount extracted from the stamp image and the confidentiality to be assigned to the stamp image, the control unit 100 is considered to have a high confidentiality that is likely to be set by the user. Are assigned to candidate levels. As a result, when the user thinks that the sensitivity “HIGH” is set for the new stamp image STn, the sensitivity desired by the user is assigned to the new stamp image without performing an operation for setting the sensitivity. .

  When the user performs an operation for instructing registration of the new stamp image STn via the UI unit 300, the control unit 100 registers the new stamp image STn in the stamp registration apparatus 10 (step S13). Specifically, in step S13, the control unit 100 assigns an ID (eg, “#stamp n”) to the new stamp image STn and causes the storage unit 200 to store image information representing the new stamp image STn. Then, the control unit 100 describes the feature amount of the new stamp image STn having the contents shown in FIG. 10 in the feature amount management table 210, and describes the confidentiality assigned to the new stamp image STn in the confidentiality management table 230. Update these.

  Incidentally, as shown in FIG. 11, “HIGH” is assigned as a candidate level to the new stamp image STn, but the candidate level does not necessarily match the confidentiality desired by the user. In this case, the control unit 100 causes the user to specify the desired confidentiality and assigns the specified confidentiality to the new stamp image STn. For example, when the user wishes to set the sensitivity “MIDDLE”, the user selects the soft button SBc attached with “LEVEL1: MIDDLE”. By this selection, the control unit 100 changes the candidate level from “HIGH” to “MIDDLE” and changes the display content of the soft button SBa in FIG. 11 to “candidate level MIDDLE”. In addition, when the soft button SBb is selected by the user, the control unit 100 changes the candidate level to “HIGH”. When the user operates the soft button SBa, the control unit 100 displays each sensitivity in a drop-down menu format, for example, and changes the sensitivity assigned to the new stamp image STn according to the user's operation content. May be. Accordingly, the control unit 100 assigns the confidentiality designated by the user according to the display content displayed on the UI unit 300 to the new stamp image.

(B-2) When there is no significant difference In step S10, the control unit 100 determines the difference between the similarity of the registered stamp image ST2 having the highest similarity and the similarity of the other registered stamp images ST1 and ST3. If it is determined that there is no significant difference between the determined values (step S10; NO), the process proceeds to step S14.
Next, the control unit 100 determines whether or not there is a feature amount that has not yet been subjected to collation processing (step S14). Here, since the control unit 100 has only performed the matching process for the feature amount “complexity”, it determines “YES” and returns to step S6. That is, for a certain feature amount, when no significant difference is found between the similarity to the registered stamp image most similar to the new stamp image and the similarity to other registered stamp images, the other feature amount Therefore, the registered stamp image similar to the new stamp image is to be specified.

  In step S <b> 6, the control unit 100 executes collation processing for calculating the similarity between the new stamp image STn and each registered stamp image for the feature amount “color” having the second highest priority after “complexity”. Here, the control unit 100 calculates the degree of similarity according to the degree of closeness of color visually. For example, the control unit 100 calculates the similarity so that the similarity between red and orange is increased to “10” and the similarity between yellow and black is decreased to “1”. Regarding the closeness of colors, for example, the degree of similarity may be determined according to the color difference of the CIELAB color system. Then, the control unit 100 executes Step S7 for the feature amount “color” (for example, the threshold value is “7”), and as a result, the color is “red”. , And thereafter, it is determined that the similarity is high in the order of registered stamp images ST2 and ST3 in the order of similarity. Next, if the control part 10 judges "YES" at Step S8, it will progress to Step S9. If the control unit 100 identifies “2” or more registered stamp images and determines “YES” in step S9, the process proceeds to step S10. Then, the control unit 100 exceeds the condition where the difference between the highest similarity and the other similarities for the feature amount “color” is determined, and there is a significant difference between them. Whether or not (step S10).

  Here, the control unit 100 determines the difference between the similarity corresponding to the registered stamp image ST1, the similarity corresponding to the registered stamp image ST2 that is “blue”, and the similarity to the registered stamp image ST3. Is determined to exceed a predetermined condition (step S10; YES). For example, if there is a difference of “5” or more between the similarity of the registered stamp image with the highest similarity and the similarity of the registered stamp image with the second highest similarity, the control unit 100 Judge that the size exceeds the determined condition. Then, the control unit 100 determines candidate ranks in descending order of similarity calculated for the feature amount “color” in step S7 (step S11). In step 11, the control unit 11 determines a candidate rank in which the stamp image ST1 is “1st”, the stamp image ST2 is “2nd”, and the stamp image ST3 is “3rd” (step S11). Then, the control unit 100 outputs control information to the UI unit 300 so that the registered stamp images are displayed in order of the determined candidate rank, and displays information indicating the registered stamp image and sensitivity (step S12). As a result, the control unit 100 displays the registered stamp image and the information indicating the confidentiality in association with each other in accordance with the similarity order of the feature amount “color” set with a lower priority than the feature amount “complexity”. Will do.

  FIG. 12 is a diagram showing an example of display contents displayed on the UI unit 300 by executing step S12. In addition, about the same structure shown in FIG. 11, the same code | symbol is attached | subjected and the description is abbreviate | omitted. Here, since the display is performed according to the candidate rank based on the feature amount “color”, the information related to the stamp image ST1 ranked first in the candidate rank is arranged in the upper part of the display area, and the stamp image ST2 ranked second in the candidate rank. Information is arranged in the lower part of the display area. Here, the candidate level is “HIGH” according to the stamp image ST1 ranked first in the candidate ranking. Thus, the display content (particularly the candidate level) displayed in step S12 differs depending on the priority set for the feature amount and the type of feature amount that causes a significant difference. The operation in step S13 is the same as described above.

  In step S10, when the control unit 100 determines that the difference amount does not exceed a predetermined condition for the feature amount “color”, “shape” or “dimension” is used as another feature amount. The processing steps are executed in the same procedure as above. Note that the feature amount “shape” is calculated such that the similarity between the squares and the circles is high, and the similarities between the shapes of the control unit 100 are higher. For the feature amount “dimension”, the control unit 100 calculates the similarity so as to increase the similarity even though the sizes are similar. Therefore, by specifying the priority in advance that the user wants to prioritize in consideration of the application of the stamp and the tendency of the relationship between the sensitivity and the feature amount, the operation related to the setting of the sensitivity when registering the stamp image. The burden is reduced.

  As described above, the display result in S12 differs depending on which feature quantity is determined for which feature level. For example, if the priority of the feature quantity “color” is set to “1” and the priority of the feature quantity “complexity” is set to “2”, for example, the control unit 100 is significant for the feature quantity “complexity”. Even if there is a difference, the same result as in FIG. Further, the display contents displayed in step S12 are different depending on how to set a plurality of types of priorities including the priorities of the feature amounts “shape” and “dimension”.

  By the way, when the control unit 100 determines in step S9 that the number of registered stamp images specified in step S7 is “1” (step S9; NO), the process proceeds to step S11 and one specified registered stamp is determined. The candidate ranking “first place” is assigned to the image, and only information corresponding to the registered stamp image is displayed in step S13. In other words, the control unit 100 determines the feature amount of one registered stamp image and the feature amount determined that the difference between the new stamp image and the registered stamp image exceeds a predetermined condition. The display control in step S12 is executed for the feature amount having the highest priority set.

  The control unit 100 performs collation processing for all types of feature quantities of “complexity”, “color”, “shape”, and “dimension”, and there is a feature quantity that identifies one registered stamp image and a significant difference. If the determined feature amount does not exist, “NO” is determined in step S14. In this case, the control unit 100 proceeds to step S15. Then, the control unit 100 determines that the difference between the highest similarity and the other similarities is the largest among those whose similarity exceeds the threshold based on the matching results for all the feature values subjected to the matching process. Specify the feature amount. Then, the control unit 100 determines the candidate order of registered stamp images in descending order of similarity among the identified feature quantities whose similarity exceeds the threshold, and proceeds to step S12. As a result, even when there is no feature amount with high reliability of the result, the control unit 100 adopts a result with a feature amount having high priority and relatively high reliability of the result. Can be assigned the desired candidate level.

If the control unit 100 determines in step S8 that there is no registered stamp image whose similarity to the new stamp image is greater than or equal to the threshold based on the collation process in step S7 (step S8; NO), step Proceed to S16. Then, the control unit 100 determines whether there is a feature amount that has not been subjected to the matching process. Here, if the control unit 100 determines that there is a feature quantity that has not been subjected to collation processing (step S16; YES), the control unit 100 returns to step S6 and executes the above processing steps.
On the other hand, when the control unit 100 determines that there is no feature amount for which the matching process has not been performed (step S16; NO), the control unit 100 proceeds to step 17. If the registered stamp image whose similarity is equal to or greater than the threshold in the matching process in step S6 has already been specified for all feature quantities, the control unit 100 is based on the similarity of the feature quantity having the highest priority. The candidate ranking is determined (step S17). If no registered stamp image similar to the new stamp image is specified for all feature amounts, the control unit 100 determines that there is no similarity. And the control part 100 progresses to step S12, and performs the said process. However, if the control unit 100 determines in step S16 that there is no candidate, for example, a message indicating that there is no similar registered stamp image is displayed, and a predetermined default sensitivity (for example, “MIDDLE”) is set as the candidate level. Instead of the default sensitivity, the highest confidentiality among the registered stamps registered in the stamp registration apparatus 10 may be set as the candidate level.

  According to the embodiment described above, the stamp registration apparatus 10 sets a priority for each of a plurality of feature amounts extracted from the stamp image. When registering a new stamp image, the stamp registration device 10 specifies a registered stamp image whose similarity with the new stamp image exceeds a threshold value for each type of feature amount. When the stamp registration apparatus 10 identifies two or more registered stamp images with respect to the feature amount having the highest priority among the types of feature amounts in which there exists a registered stamp image whose similarity is equal to or greater than the threshold value, It is determined whether or not there is a significant difference between the highest degree of similarity and the other degree of similarity. Then, the stamp registration apparatus 10 determines a candidate rank based on the feature amount determined to have a significant difference, and displays information corresponding to the registered stamp image on the UI unit 300 according to the determined candidate rank. Further, at this time, the control unit 100 assigns the confidentiality assigned to the registered stamp image having the candidate rank “1” as the candidate level of the new stamp image. On the other hand, if the stamp registration apparatus 10 determines that there is no significant difference, the stamp registration apparatus 10 specifies only one registered stamp image whose similarity exceeds the threshold value, or a significant difference occurs with respect to the similarity of the lower-order registered stamp images. The matching process is repeated in order of priority until the registered stamp image is specified.

  With the above configuration, there is an effect that a result with high reliability is adopted while emphasizing the result of specifying an image with high similarity based on a feature quantity with high priority. Even if it is judged that the similarity is high for a feature with high priority, if there is no clear difference between the registered stamp image with the highest similarity and the other stamp images, the candidate rank is affected by the error. This is because the actual similarity may not be correctly reflected. Therefore, in this case, the control unit 100 adopts a result with higher reliability based on other feature values. In addition, the font and size of the character image included in the stamp are important in determining the attribute of the stamp, but the configuration that extracts the feature amount of the stamp by character recognition cannot fully cope with it. High feature quantity can be obtained. In addition, for the purpose of registering a new stamp image, a situation such as a loss or loss of a stamp may occur in practice. However, when trying to register a stamp image of a substitute for this, a user related to the assignment of confidentiality Is reduced.

In addition, according to the configuration of the present embodiment, when the user uses a stamp in consideration of the relationship between the confidentiality and the feature amount, the user himself / herself specifies the priority of the feature amount based on the relationship. This increases the probability that the confidentiality desired by the user is assigned as a candidate level for the new stamp image. As a result, even when the sensitivity level becomes enormous, the operation burden in registering the stamp image is reduced. In particular, since the correlation between the sensitivity and the feature amount often varies greatly depending on the user's tendency to use the stamp, etc., it is difficult to fix the correlation and determine the priority. 10, the result of similarity determination is presented in response to a request from the user.
In addition, the contents of the stamp mark are slightly different depending on the method of stamping by the user, the material of the paper, and the like. Thereby, the feature amount extracted from the new stamp image by the registered stamp image 10 varies depending on the state of the mark. On the other hand, the registered stamp image 10 performs a collation process on a plurality of types of feature amounts, and places importance on the result of similarity determination based on the feature amount having a high priority, and the image having the highest similarity and the others. Since a result based on a feature amount in which there is a clear difference from the image is adopted, a stable result is presented to the user regardless of the state of the mark.

(C) Modified Example The present invention can be implemented in a form different from the above-described embodiment. Moreover, you may combine each of the modification shown below.
(C-1) Modification 1
In the above-described embodiment, the example of the control unit 100 built in the stamp registration apparatus 10 has been described. However, the control unit 100 is not limited to that built in the stamp registration apparatus 10, and for example, the stamp registration apparatus 10. Alternatively, it may be realized by a computer device (image processing device) connected via a communication means such as a USB (Universal Serial Bus) cable or a LAN (Local Area Network). In this case, the image processing apparatus includes a communication interface, and realizes each function of the control unit 100 illustrated in FIG. 6 based on the new stamp image included in the image information received and acquired from the external apparatus. Output the result.
Further, a part of the configuration of the stamp registration image 10 of the above-described embodiment may be provided in the external device, that is, each configuration may be dispersed.

(C-2) Modification 2
In the above-described embodiment, in step S14, the control unit 100 determines the highest if the feature amount for which one registered stamp image is specified and the feature amount determined to have a significant difference by the determination unit 150 do not exist. The candidate ranking is determined on the basis of the feature quantity that maximizes the difference between the similarities of and other similarities. Instead, the control unit 100 may determine the candidate rank based on the feature amount having the highest priority among the feature amounts in which the registered stamp image having the similarity equal to or greater than the threshold exists. Moreover, you may employ | adopt the feature-value specified beforehand by the user. In short, the control unit 100 outputs information of a registered stamp image having a high similarity with respect to a feature amount that satisfies a condition for which the calculated similarity is determined. Further, as a reference for determining whether or not the control unit 100 has a significant difference, in addition to the determination based on the difference between the highest similarity and the second highest similarity, for example, the second highest It may also be included as a condition that there is a difference larger than the determined size between the similarity and the third highest similarity. That is, if there are a plurality of second highest similarities and multiple similarities, the reliability of the feature quantity extracted by the control unit 100 may be low.

(C-3) Modification 3
Further, when different confidentiality is assigned to a plurality of registered stamp images whose feature amounts are remarkably approximate, there may be a problem in use. Therefore, when there is a registered stamp image that is remarkably similar to the feature amount extracted from the new stamp image, the control unit 100 only sets the sensitivity of the new stamp image to the same sensitivity as the similar registered stamp image. May be allowed and other confidentiality may be prohibited. For determining whether or not the feature amounts are remarkably similar, for example, when the similarity exceeds a threshold value for a predetermined number (for example, three types) of feature amounts, the control unit 100 determines that they are remarkably similar. .

(C-4) Modification 4
In the embodiment described above, the control unit 100 obtains the similarity for each type of feature quantity, but the similarity degree calculated by the similarity calculation unit 140 of the control unit 100 for each type of feature quantity is more than that. You may correct | amend according to the similarity calculated | required about the feature-value with high priority. Specifically, the similarity calculation unit 140 includes a correction unit that performs correction, and when it is determined that there is no significant difference between the high-priority feature amounts, the similarity calculation unit 140 is obtained from the higher-priority feature amounts. A correction amount corresponding to the magnitude of the similarity difference is applied to the similarity obtained from the lower-level feature amount to correct it. At this time, the correction unit performs correction by increasing the correction amount for relatively increasing the similarity as the similarity of the higher priority feature amount is larger. As a result, the similarity obtained for the feature quantity having a higher priority is reflected in the result of the similarity for the feature quantity having a lower priority. At this time, weighting may be performed so that the correction amount by the feature amount having a high priority acts more greatly.

(C-5) Modification 5
In the embodiment described above, in step S12, the control unit 100 arranges the registered stamp images in order of the obtained similarity as shown in FIGS. 11 and 12, but in determining the arrangement order, You may use the similarity of a feature-value. For example, if the difference in similarity between the registered stamp images of the second and third candidate ranks is less than or equal to the threshold value, the control unit 100 calculates the similarity calculated using a feature amount having a higher or lower priority. Pay attention to. Then, if there is a clear difference between the similarities of the noted feature quantities, the control unit 100 arranges the features so that the one with the higher similarity is higher. Thus, when the control unit 100 displays a registered stamp image whose similarity is equal to or higher than a threshold value, the registered stamp image may be arranged using a specific result based on another feature amount. In addition, the control unit 100 may display a registered image whose similarity is equal to or less than a threshold according to the user's operation of the soft button SBd. In this configuration, the control unit 100 determines candidate ranks in order of similarity for registered stamp images whose similarity is below the threshold for the feature quantity with the highest priority, and those higher than the threshold are described in the embodiment. As described above, the candidate rank may be determined using the lower-level feature amount. In addition, the control unit 100 may display information corresponding to registered stamp images in descending order of candidate rank by increasing the candidate rank in ascending order of similarity.

(C-6) Modification 6
In the embodiment described above, priority is assigned to all of the plurality of types of feature amounts. However, it is only necessary that at least one feature amount has a higher priority than other feature amounts.
For example, the priority “1” is assigned to the feature amount “complexity”, and the same priority “2” is assigned to the other “color”, “shape”, and “dimension”. . In this configuration, when the control unit 100 does not identify one registered stamp image and finds a significant difference with respect to the feature amount “complexity” having a high priority, the control unit 100 performs collation using other feature amounts. Based on the result, information representing the registered stamp image is output. In this configuration, the control unit 100 uses a technique disclosed in Japanese Patent No. 3812161 for a feature amount other than a feature amount “complexity” having a high priority, so that a registration stamp having a high similarity level overall. An image may be specified. In this case, the first feature value of the present invention corresponds to “complexity”, and the second feature value of the present invention corresponds to “color”, “shape”, and “dimension”.

(C-7) Modification 7
In the above-described embodiment, the semantic information assigned to the meaning of the stamp image represents the confidentiality, but the content may be anything, for example, the type of stamp, the attribute of the document on which the stamp is pressed, etc. It only has to represent some meaning of. Note that, regardless of the content of the semantic information, the operation of the stamp registration apparatus 10 is executed in the same procedure as in the above-described embodiment. In addition, the semantic information itself may not be used. In this case, the control unit 100 outputs control information for displaying the registered stamp image without outputting information indicating the confidentiality as information corresponding to the registered stamp image.
In step S12, the control unit 100 outputs control information for displaying information representing the registration stamp image. However, the control unit 100 is not limited to the display output, and the control unit 100 relates to the registration stamp image in other output formats such as communication output. Information may be output.
In addition, the control unit 100 is not limited to setting the priority according to the user's specification, and may acquire information related to the priority setting from a device connected via the communication unit. Is not limited to the aspect of the embodiment. Further, the present invention does not prevent the configuration in which the priority for each feature amount is fixed.

(C-8) Modification 8
In the above-described embodiment, a stamp image that is an image corresponding to the stamp imprint is registered, but the content of the image to be registered may be any. For example, some printer apparatuses generally have a function of forming an image similar to the stamp image shown in FIG. 2, and this image may be registered as the new stamp image. In short, any image may be used as long as the similarity is determined for the feature amount extracted from the image.

(C-9) Modification 9
Further, the present invention may be applied to a collation apparatus that performs collation processing for a plurality of feature amounts extracted from an image as described in the above-described embodiment. This collation apparatus performs collation processing for feature amounts with high priority, and when it is determined that there is no significant difference, collation processing is performed for other feature amounts with lower priority. Thus, even if the collation device does not have a function related to image registration, the same effect as the configuration of the embodiment can be obtained as long as it has a configuration related to similarity determination between images. That is, the present invention is not limited to being applied only to an apparatus (system) for registering an image, but may be applied to an apparatus having a function of determining similarity of an image using a feature amount.

  In addition, the feature amounts described in the above-described embodiments are merely examples, and a plurality (at least two types) of feature amounts smaller than this may be used, or more feature amounts may be used. Further, the feature amount is not limited to the type described in the embodiment, as long as it is information representing the feature of the image. For example, a color histogram, a density distribution, or the like may be used as the feature amount.

DESCRIPTION OF SYMBOLS 10 ... Stamp registration apparatus, 100 ... Control part, 110 ... Extraction part, 120 ... Reception part, 130 ... Priority setting part, 140 ... Similarity calculation part, 150 ... Judgment part, 160 ... Output part, 170 ... Sensitivity information Acquisition unit, 180 ... registration unit, 200 ... storage unit, 210 ... feature management table, 220 ... priority management table, 230 ... confidentiality management table, 300 ... UI unit, 400 ... reading unit.

Claims (8)

A first feature amount that represents a feature of the image, and a second feature amount that is a feature amount that represents the feature of the image and has a priority lower than the priority set for the first feature amount. Extraction means for extracting at least a plurality of types of feature quantities from the first image;
The feature amount extracted from each of a plurality of second images different from the first image is acquired, and the similarity between each acquired feature amount and the feature amount extracted by the extraction unit is expressed as the first amount. Similarity calculation means for calculating each of the first and second feature amounts;
When two or more similarities greater than or equal to the threshold are calculated by the similarity calculating means, the difference between the highest similarity and the other similarities exceeds a predetermined size. Determining means for determining whether or not for each type of feature amount;
For the first feature amount, the similarity calculation unit calculates a similarity that is equal to or greater than the threshold value of 1, or the determination unit determines that the size exceeds the size. In this case, an image processing apparatus comprising: output means for outputting information corresponding to the second image according to the similarity calculated for the first feature amount. A priority is set for each of the plurality of types of feature values,
The similarity calculation means calculates the similarity for each of the plurality of types of feature amounts,
The output means is of a type that satisfies a condition that either the similarity that is equal to or greater than the threshold value of 1 is calculated by the similarity calculation means or that the size is determined to be greater than the size by the determination means. The image processing apparatus according to claim 1, wherein the information is output for a feature amount having the highest priority. In the case where there is no feature quantity for which the similarity that is equal to or greater than the threshold value of 1 is calculated by the similarity calculation means and a feature quantity that is determined to exceed the size by the determination means, The image processing apparatus according to claim 1, wherein the information is output for a feature quantity of a type that satisfies a predetermined condition. The output means is a type that satisfies a condition that a degree of difference between the highest similarity and the other similarities is maximized among those whose similarity calculated by the similarity calculation means is equal to or greater than the threshold. The image processing apparatus according to claim 3, wherein the information is output with respect to the feature amount. Accepting means for accepting an operation of designating priority for the plurality of types of feature amounts by the user;
5. The apparatus according to claim 1, further comprising: a priority setting unit configured to set a priority according to the specified operation received by the receiving unit for each of the plurality of types of feature amounts. The image processing apparatus described. Semantic information acquisition means for acquiring semantic information assigned to the meaning of the second image;
The output means outputs control information for causing the display means to display the second image and information corresponding to the semantic information acquired by the semantic information acquisition means and assigned to the meaning of the second image. The image processing apparatus according to claim 1, wherein the image processing apparatus is an image processing apparatus. An image processing apparatus according to claim 6;
Storage means for associating image information representing the first image, information representing the plurality of types of feature amounts extracted by the extraction means, and semantic information representing the meaning assigned to the first image Storage control means for storing in
The meaning information designated by the user according to the display content displayed on the display means in accordance with the control information output by the output means of the image processing device represents the meaning assigned to the first image. And a storage control means for storing the storage control device. Computer
A first feature amount that represents a feature of the image, and a second feature amount that is a feature amount that represents the feature of the image and has a priority lower than the priority set for the first feature amount. Extraction means for extracting at least a plurality of types of feature quantities from the first image;
The feature amount extracted from each of a plurality of second images different from the first image is acquired, and the similarity between each acquired feature amount and the feature amount extracted by the extraction unit is expressed as the first amount. Similarity calculation means for calculating each of the first and second feature amounts;
When two or more similarities greater than or equal to the threshold are calculated by the similarity calculating means, the difference between the highest similarity and the other similarities exceeds a predetermined size. Determining means for determining whether or not for each type of feature amount;
For the first feature amount, the similarity calculation unit calculates a similarity that is equal to or greater than the threshold value of 1, or the determination unit determines that the size exceeds the size. In the case, a program for functioning as an output unit that outputs information according to the second image according to the similarity calculated for the first feature amount.

Images