JP4882523B2 - Document impression evaluation apparatus and document impression evaluation program - Google Patents

Description

  The present invention relates to a document impression evaluation apparatus and a document impression evaluation program that present information related to an impression that a document gives to a person based on visual information acquired from the document.

  With the recent increase in information processing speed and capacity of corporate marks, products, presentation materials, website pages, etc., many colored documents are used in presentation materials and various websites. It has become so. In these documents, the visual impression given to humans varies greatly depending on the selection, arrangement, area, and area ratio of colors used in the documents.

  However, much of the actual document creation is left to the general user. A general user who is unfamiliar with document creation often does not know how to color and color a document to obtain his intended atmosphere, aesthetics and visibility. In addition, a means for objectively evaluating an impression given to a user by a created document is also required for a document creation specialist.

  In order to solve this problem, research on processing and document evaluation based on human sensitivity has been actively conducted, and in particular, sensitivity evaluation means for evaluating human sensitivity, and the results and document characteristics are analyzed. Many sensuous document creation support means have been proposed that combine document analysis means.

  Patent Document 1 and Patent Document 2 disclose an interactive color arrangement support method and a color arrangement support apparatus using the same. In this color arrangement support method, when a user edits image data, the appropriateness of the color arrangement such as the background of the image, characters or graphs is presented as a specific numerical value, or a color suitable for the color arrangement is presented as a candidate. .

  In Patent Document 3, a feature amount is extracted from an image to be evaluated with reference to a database in which a condition for a visual feature amount related to an image is associated with expression information representing an impression felt by a person who views the image. An image impression evaluation apparatus is disclosed that extracts expression information associated with a condition that the extracted feature amount satisfies, and displays the expression information on an output screen as information about an impression that the image gives to a person.

JP-A-7-306935 JP-A-8-249441 JP 2004-30377 A

  However, in the above prior art, an impression given to a person by a document is not evaluated according to the purpose of expressing the document to be evaluated. For example, the impression that the document gives to people varies depending on the type of product included in the document such as the announcement, the content of the service, the type of business to which information such as the homepage is provided, the gender and age of the target person, etc. Evaluation according to the purpose of document expression (information content and information provider) could not be performed.

  SUMMARY OF THE INVENTION The present invention has been made in view of the above-described problems of the related art, and an object thereof is to provide a document impression evaluation apparatus and a document impression evaluation program for evaluating a document impression according to the purpose of expressing an evaluation target document.

The present invention relates to a document impression evaluation apparatus that presents information related to an impression given to a human by a document based on visual information extracted from the document. The document impression evaluation apparatus relates a keyword related to a document content to a classification category. A document discriminating unit that discriminates a classification category associated with a keyword included in an evaluated document to be evaluated with reference to a database; and a feature amount extracting unit that extracts at least one feature amount from the evaluated document; The document discriminating unit refers to an expression information database in which a condition for at least one visual feature amount related to a document is associated with expression information representing an impression felt by a person who views the document for each classification category. Of the document to be evaluated Impression class division and the conditions characteristic of the evaluation document satisfies the expression information extraction means for extracting the expression information associated with the representation of the extracted information, the the object to be evaluated document document has on the human And the output document is an advertisement of a product, and the classification is represented by at least one of a product type, service content, or viewer characteristics. It is characterized by that .

These document impression evaluation apparatuses according to the present invention can be realized by a program executed by a computer. Specifically, it is a document impression evaluation program for performing a document impression evaluation that presents information related to impressions given to human beings based on visual information extracted from a document, the computer including the contents of the document A document discriminating means for discriminating a classification category associated with a keyword included in the evaluated document to be evaluated with reference to a classification database in which keywords related to the classification category are associated, and at least one from the evaluated document An expression in which feature quantity extraction means for extracting one feature quantity, a condition for at least one visual feature quantity relating to a document, and expression information representing an impression felt by a person who sees the document are associated with each classification category Browse the information database before Conditions characteristic of classification categories and the evaluation target document obtained the evaluation target document meets the document discrimination means, and the expression information extraction means for extracting the expression information associated, the extracted expression information, the Functioning as an output means for outputting information about an impression given to a human by a document to be evaluated, wherein the evaluated document is an advertisement of a product, and the classification category includes a product type, a service content, Alternatively, it can be realized by a document impression evaluation program characterized by being represented by at least one of the viewer characteristics .

  Note that each means realized by the document impression evaluation program is equivalent to each step or combination of steps in the flowchart in the embodiment of the present invention. Each means realized by the document impression evaluation program need not be realized by one computer, and may be realized by a plurality of computers connected so as to be able to transmit information through a network or the like.

  Here, the output means divides the area of the output screen into a plurality of divided screen areas, assigns the condition for each of the divided screen areas, and the divided screen corresponding to the condition that the feature amount of the evaluated document satisfies. It is preferable to display the extracted expression information in a region. The output means divides an output screen area into a plurality of divided screen areas, assigns the condition to each of the divided screen areas, and the divided screen area corresponding to the condition satisfied by the feature amount of the evaluated document. It is also preferable to display information on the feature amount of the document to be evaluated.

  In this way, by dividing the output screen and displaying the information related to the expression information and the feature amount in the form of a chart, the impression of the evaluated document can be presented to the user in an easily understandable manner.

  According to the present invention, it is possible to evaluate the impression of an appropriate document according to the purpose of expressing the evaluation target document.

  Embodiments of the present invention will be described below in detail with reference to the drawings. As shown in FIG. 1, an impression evaluation apparatus according to an embodiment of the present invention basically includes a control unit 10, an internal storage unit 12, a large-capacity storage unit 14, an output unit 16, a standard input unit 18, and a document input unit 20. Composed. The control unit 10, the internal storage unit 12, the mass storage unit 14, the output unit 16, the standard input unit 18, and the document input unit 20 are connected via a bus 24 so that information (data) can be transmitted.

  It is preferable that the impression evaluation apparatus further includes an interface unit 22. The interface unit 22 is connected to other components via a bus 24 so that information (data) can be transmitted. The interface unit 22 can be used for reading document data such as a web page from a server external to the impression evaluation apparatus via the network 26.

  The control unit 10 executes the document impression evaluation program held in the internal storage unit 12 and appropriately reads the document data stored in the large-capacity storage unit 14 or the document data via the network 26, and the document data Evaluate the impression of the document expressed by. The result of impression evaluation is output to the output unit 16 or the internal storage unit 12. As the control unit 10, a CPU of a general computer can be used.

  The internal storage unit 12 stores and holds an impression evaluation program executed by the control unit 10, each parameter used for evaluation processing, an impression evaluation value that is a processing result, and the like. The data held in the internal storage unit 12 can be referred to as appropriate from the control unit 10 via the bus 24. As the internal storage unit 12, a general semiconductor memory, a hard disk device, a magneto-optical disk device, or the like can be appropriately selected and used.

  The large-capacity storage unit 14 stores and holds document data of a document (hereinafter referred to as an “evaluated document”) subjected to impression evaluation by the control unit 10. Document data is held in various data formats such as a bitmap format, a jpeg format, a tiff format, and an HTML format. The document data held in the large-capacity storage unit 14 can be referred to as appropriate from the control unit 10 via the bus 24. The large-capacity storage unit 14 can select and use a large-capacity storage device such as a general hard disk device or a magneto-optical disk device.

  The output unit 16 outputs the processing result output from the control unit 10 as information that can be confirmed by the user. It is also used as a user interface when the user inputs control commands and parameters necessary for impression evaluation processing. As the output unit 16, a display device, a printer, a touch panel, or the like can be appropriately selected and used.

  The standard input unit 18 is used by a user to input control commands and parameters for document evaluation. For example, it is used when a control command for instructing the control unit 10 to start executing an impression evaluation program held in the internal storage unit 12 is used. The standard input unit 18 can be used by appropriately selecting a keyboard, a mouse, a touch panel, or the like.

  The document input unit 20 inputs document data to be subjected to impression evaluation. Document data input from the document input unit 20 is stored and held in the mass storage unit 14 or the internal storage unit 12. The document input unit 20 can appropriately select and use a general document input device such as a scanner, a digital camera, or a digital video.

  The impression evaluation apparatus according to the present embodiment can be basically configured by a general computer. Moreover, it is not limited to the said structure, It is suitable to add, delete, or change a component suitably.

<Document impression evaluation method>
A document impression evaluation processing method using the impression evaluation apparatus will be described below with reference to the drawings. FIG. 2 shows a flowchart of the first document impression evaluation method.

  As shown in FIG. 2, the present impression evaluation method inputs an evaluated document (step S1), determines the purpose of expressing the evaluated document (step S2), divides the evaluated document (step S3), and physical features. Including a subroutine (step S4) for acquiring the image, extraction of impression evaluation (step S5), and output of the evaluation result (step S6), which are executed by sequentially performing each step. Specifically, each step is stored and held in the internal storage unit 12 as an impression evaluation program that can be executed by a computer, and the impression evaluation program is executed by the control unit 10.

  When the user inputs a control command for starting execution of the impression evaluation program from the standard input unit 18, the control unit 10 executes the impression evaluation program held in the internal storage unit 12. As a result, the process immediately proceeds to step S1.

  In step S <b> 1, using the document input unit 20, document data of a document that is an impression evaluation target is acquired. For example, document data displayed on a display such as a web page is acquired via the network 26. In addition, document data displayed on a print medium such as paper is acquired using the document input unit 20. The document data is stored and held in the large-capacity storage unit 14 or the internal storage unit 12, and is appropriately read out by the control unit 10 for processing.

  In step S2, the purpose of expressing the evaluated document is determined. In the present embodiment, character data and image data included in the document to be evaluated are analyzed, the purpose of expression of the document to be evaluated is determined, and the document is classified for each type of expression purpose.

  When it is intended to present specific information such as an advertisement of the evaluated document to the viewer (user), classification is performed based on the content of the content of the evaluated document. Information representing a product or service included in the content of the evaluated document is extracted, and the evaluated document is classified based on the extracted information.

  When classifying based on the contents of the evaluated document, one or more keywords are set for each document classification category, and the classification category is set with a keyword that matches the character data extracted from the evaluated document. And the evaluated document can be classified into classification categories for which matching keywords are set.

  For example, keywords such as “meat”, “vegetables”, “fruits”, etc. are associated with the “food” category, and “refrigerator”, “washing machine”, “cleaning” are associated with the “electrical appliance” category. A keyword such as “machine” is registered in the storage unit 12 as an associated document classification database. Then, a keyword is extracted from the content of the evaluated document using an existing word extraction algorithm or the like, and a classification category associated with the keyword extracted in the document classification database is set as the content of the content of the evaluated document. For example, when the keyword “refrigerator” is included in the evaluated document, it is determined that the content of the content of the evaluated document is “electric appliance”.

  In addition, by changing the association between the keywords registered in the document classification database and the classification categories, the documents to be evaluated are classified into categories of “person”, “landscape”, “object”, etc. based on the contents of the document to be evaluated. It can also be classified.

  In addition, when the evaluated document is intended to present information to a viewer having specific characteristics, such as a company web page, the classification may be performed based on the characteristics of the target audience of the evaluated document. Is preferred. Extract the gender, age (age), residential area, physical characteristics (height, weight, etc.), economic characteristics (assets, income, income, etc.) of the users who will be browsing the evaluated document, Classify the evaluated document based on the extracted information.

  Specifically, a keyword is associated with each classification category of a viewer (user) who is a destination of information included in the content of the evaluated document, and is registered in the storage unit 12 as a document classification database. A keyword included in the evaluated document is extracted using an extraction algorithm or the like, and a classification category associated with the extracted keyword in the document classification database is set as a user characteristic of the evaluated document.

  For example, keywords such as “automobile”, “computer”,... Are associated with the category of “male / 20s”, and “mobile phone”, “cosmetics”,. The keywords such as are associated and registered in the storage unit 12 as a document classification database. Similarly, other genders and other age combinations are associated with keywords and registered in the document classification database. Then, a keyword included in the evaluated document is extracted, and if the extracted keyword is “automobile”, the classification category “male / 20s” from the document classification database is a characteristic of the viewer of the evaluated document. It is determined.

  The purpose of expressing the evaluated document may be manually set by the user inputting the purpose of expressing the evaluated document using the standard input unit 18.

  In step S3, a process for dividing the document to be evaluated is performed. The document data acquired in step S1 may be an entire document such as a website as shown in FIG. 3, or a header part 30, a title part 31, a heading part 32, a link button part 33, and an image image part 34. Also, the document may be divided into regions such as a text portion 35. It is also preferable to automatically divide the document into regions by applying an image region dividing algorithm typified by clustering based on the K-average algorithm, and to make each a target document.

  In step S4, a physical feature amount is calculated for each document area. Step S4 is processed as a subroutine as shown in FIG. Hereinafter, a case where a document such as a website is divided into several document areas and each area is processed as an evaluated document will be described.

Here, the physical feature amounts are the area of the document area, the width w and height h of the document area, the upper left positions x and y, the color average of the entire document area, the average value of the L ^* component, and the a ^* component. mean values of, b ^* average component, color difference color average and the document area overall color average of the entire document ΔL * ^a * ^{b *,} color differences in color average of the document area to be close to the document region Δ _n L ^* a ^* B ^* , the color difference of the average color between the background and the document area [Delta _] bL ^* a ^* b ^* , the color dispersion in the document area, the number of the same color, and the like.

  Hereinafter, although the calculation method of each feature-value is demonstrated, it is not necessary to calculate about all the feature-values in this Embodiment, What is necessary is just to calculate only about the feature-value used for document evaluation in step S4 and subsequent steps.

In step S41, the color space of the document data of each divided document area is converted from the RGB space to the L ^* a ^* b ^* space. In general, document data used in a computer is often expressed in an RGB color space, which is a non-uniform color space, and cannot be analyzed close to the human sense. Therefore, by converting the RGB color space, which is a non-uniform color space, to the L ^* a ^* b ^* color space, which is a uniform color space, the document can be analyzed and evaluated as an impression closer to human sensitivity. The L ^* a ^* b ^* space is also preferable because it can express the dynamic range of the document. Of course, if the document data is already expressed in the L ^* a ^* b ^* color space, no conversion is necessary.

  FIG. 5 shows an example in which the uniform color space is expressed as a general Munsell hue ring (different colors are expressed by different hatching in the drawing). Usually, the Munsell hue ring performs multi-color classification, but a part of them is shown in FIG. In the following, description will be made using a Munsell hue ring in which the color space is divided into 10 for both hue and brightness. Of course, it is preferable to use a color palette in which the color space is further subdivided.

  In step S42, the document data of each document area subjected to color space conversion is stored and held in the large-capacity storage unit 14 or the internal storage unit 12.

  In step S43, the feature quantity of position, area, color average, color dispersion, and number of colors is extracted for each document area. The document data of each document area held in the large-capacity storage unit 14 or the internal storage unit 12 is sequentially selected, and each feature amount for each document area is obtained. At this time, a document area having the maximum area among all document areas is defined as a background area.

For example, the position is determined by obtaining the position x, y of the upper left pixel for each document area. The area is determined by extracting the width w and height h of each document area. The color average and color variance are determined by calculating the arithmetic mean and variance value for the L ^* component, a ^* component, and b ^* component of all pixels in each document area, respectively. The number of colors is determined by accumulating the number of colors in which all of the L ^* component, a ^* component, and b ^* component are different for all pixels in each document area. Similarly, the color average, color dispersion, and number of colors of the entire document and the background area can be obtained.

In step S44, the color difference between each document area, the entire document, and the background area is obtained. First, differences ΔL _i ^* , Δa _i ^*, and Δb _i ^* of the respective color components with respect to the entire document are obtained for each document area. Differences ΔL _i ^* , Δa _i ^*, and Δb _i ^* between the color components are expressed by Equation (1).
(Equation 1)
[Delta] L _i ^* = color average of ^{the L *} component of the document area i - color average of the entire document ^{L *} component .DELTA.a _i ^* = color average of ^{a *} component of the document area i - entire document ^{a *} component of the color average [Delta] b _i ^* = B of the document area i ^* color average of the component-b of the entire document ^* color average of the component where i is an identification code of the document area (1)

Similarly, differences ΔL _j ^* , Δa _j ^*, and Δb _j ^* between the color components of the document area and the background area are obtained. Differences ΔL _j ^* , Δa _j ^*, and Δb _j ^* between the color components are expressed by Equation (2).
(Equation 2)
ΔL _j ^* = color average of ^{the L *} component of the document area j - color average of ^{the L *} component of the background area .DELTA.a _j ^* = color average of ^{a *} component of the document area j - the background area ^{a *} component of the color average [Delta] b _j ^* = B of the document area j ^* color average of the component-b of the background area ^* color average of the component where j is an identification code of the document area (2)

The color difference Δ _i L ^* a ^* b ^* between each document area and the entire document and the color difference Δ _bj L ^* a ^* b ^* between each document area and the background area can be obtained using Equation (3).
(Equation 3)
Δ _i L ^* a ^* b ^* = {ΔL _i ^{* 2} + Δa _i ^{* 2} + Δb _i ^{* 2} } ^1/2
Δ _bj L ^* a ^* b ^* = {ΔL _j ^{* 2} + Δa _j ^{* 2} + Δb _j ^{* 2} } ^1/2
Here, i and j are document area identifiers (3).

In step S45, the color difference between each document area and the document area adjacent to that area is obtained. When the document area i is adjacent to the document area j, the differences ΔL _ij ^* , Δa _ij ^*, and Δb _ij ^* of the respective color components are obtained using Equation (4).
(Equation 4)
[Delta] L _ij ^* = color average of ^{the L *} component of the document area i - color average of ^{the L *} component of the document area j .DELTA.a _ij ^* = color average of ^{a *} component of the document area i - of the ^{a *} component of the document area j color average [Delta] b _ij ^* = average color of ^{the b *} component of the document area i - of the document area j ^{b *} component of the color average ........................... (4)

The color difference Δ _ni L ^* a ^* b ^* between the document area i and the document area j can be obtained using Equation (5).
(Equation 5)
_Δni L ^* a ^* b ^* = {ΔL _ij ^{* 2} + Δa _ij ^{* 2} + Δb _ij ^{* 2} } ^1/2
・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ ・ （5）

If the document area is adjacent to a plurality of document areas, the difference ΔL _ij ^* , Δa _ij ^* and Δb _ij ^* of each color component with respect to all adjacent document areas is obtained, and the color difference is obtained by performing an arithmetic average thereof. Is preferably obtained.

  FIG. 6 illustrates the result of obtaining each feature amount for each document area. These feature amounts are stored and held in the mass storage unit 14 or the internal storage unit 12 in association with each document area. When the extraction of each feature amount is completed, the process proceeds to step S5 of the main routine.

  In step S5, an evaluation for the impression of the document is obtained based on each feature amount extracted in step S4. Specifically, an expression information database in which document feature value values and expression information expressing the impression of the document are associated is prepared in advance, and expression information corresponding to the feature values extracted in step S4 is extracted. . The expression information is preferably not only text information but also image information and audio information.

  In the expression information database, a condition relating to at least one feature amount and expression information relating to an impression felt by a person who views the document are associated and registered. At this time, expression information corresponding to the difference in the purpose of expression of the evaluated document is associated even under the same feature amount condition. The expression information database is stored and held in the internal storage unit 12 or the mass storage unit 14.

Here, as an example, description will be made using the expression information database shown in FIG. In this database, expression information is registered for each product type representing the content of information on the condition that the combination of the L ^* component, a ^* component, and b ^* component of the average color in the document. That is, the purpose of the expression here is the product type included in the content shown to the user who browses the document.

For example, if the L ^* component of the background region (the document region having the largest area) is 6, the a ^* component is 20 and the b ^* component is 10, the product represented by the evaluated document in step S2 is “electric appliance”. If it is determined that the expression information “urban” is extracted under the condition A2, and the document to be evaluated is determined as “food”, the condition B2 is satisfied. Thus, the expression information “gorgeous” is extracted. When the acquired feature amount matches other conditions, expression information associated with the conditions is extracted.

Further, as described above, when processing is performed based on the characteristics of the viewer of the evaluated document, the expression information database shown in FIG. 8 can be used. In this database, expression information is registered for each characteristic of the user who becomes the viewer of information on the condition that the combination of the L ^* component, a ^* component, and b ^* component of the average color in the document. That is, the purpose of the expression here is the gender and age of the user viewing the document.

For example, if the L ^* component of the background area (the document area having the largest area) is 6, the a ^* component is 20 and the b ^* component is 10, the user who becomes the viewer of the evaluated document in step S2 is “male. If it is determined that it is “20's”, the expression information “gorgeous” corresponding to the condition A2 is extracted, and the document to be evaluated is determined to be “female / 20's” The expression information “cute” corresponding to the condition B2 is extracted. When the acquired feature amount matches other conditions, expression information associated with the conditions is extracted.

Note that the feature amount condition that can be used to extract information related to the impression of the document is not limited to the combination of the L ^* component, a ^* component, and b ^* component of the average color in the document. Area color average (L ^* component, a ^* component and b ^* component), feature amount related to color components of document area (main area) with color difference from background area greater than or equal to a predetermined value, and area less than predetermined value The feature amount relating to the color component of the document area (accent area) having the largest color difference from the background area may be used. The feature amount includes the area of the document area, the width w and height h of the document area, the upper left positions x and y, the color average of the entire document area, the average value of the L ^* component, the average value of the a ^* component, b ^* Average value of components, average value of R components, average value of G components, average value of B components, color difference ΔL ^* a ^* b ^* between the color average of the entire document and the color average of the entire document area, and close to the document area chrominance color average of the document region _{^{Δ n L * a * b *}} , the background and the color difference of the color average of the document region _{^{Δ b L * a * b *}} , the color of the dispersion in the document area, even with the same color number Good.

  Further, a multidimensional expression information database in which conditions combining these feature quantities and expression information related to the impression of a document are associated may be used. In this way, by combining as many feature quantities as possible, the impression that the document gives to humans can be made more detailed. In particular, by using conditions that include differences in feature quantities between different document areas, such as color differences between document areas, to extract expression information that represents impressions received from places with strong contrast in the evaluated document Is also possible.

Moreover, you may combine a some attribute also about the objective of the expression combined with the conditions of a feature-value. A plurality of document content types may be combined, and the expression information may be registered in association with the feature amount condition for each combination. For example, for the combination of “electric appliance” and “automobile”, the combination of “food” and “clothing”, etc., the L ^* component, the a ^* component, and the b ^* component of the feature amount are combined. The associated expression information may be registered in the expression information database.

  In addition, when the document to be evaluated is classified into “person”, “landscape”, “object”, etc., the relationship between the feature amount condition and the expression information is registered in the expression information database for each content type. May be used.

Further, the expression information may be registered in association with the feature amount condition for each of the attributes of the user who will receive the document, alone or in combination. For example, for the combination of “gender”, “age”, and “residential area”, the expression information associated with the combination of the L ^* component, the a ^* component, and the b ^* component of the feature amount is stored in the expression information database. You may register.

  In step S <b> 6, the output unit 16 displays information about the acquired feature amount and the extracted expression information. The output is preferably performed in such a manner that the user can easily grasp information about the document.

  For example, as shown in FIG. 9, a document 90 to be evaluated, a document color information display unit 91 indicating the distribution and average color distribution of each document area included in the document to be evaluated, and expression information extracted based on the feature amount It is preferred to include a color image 92 to be shown and a display 93 to show the average color of the main document area.

  As shown in FIG. 10, the document color information display unit 91 obtains, for each document area included in the evaluated document, a color on the color palette that most closely approximates the average color of each document area, and for each color of the color palette. Then, the total area of the document areas having an average color approximate to the color is obtained, and the area proportional to the area is filled with the color of the color palette and displayed.

Specifically, as shown in FIG. 11, a database in which the screen position of the document color information display unit 91 and the condition of the color component range included in the color palette are associated is stored in the internal storage unit 12 or the large capacity storage unit 14. Each color is referred to by referring to each color component (L ^* component, a ^* component and b ^* component) and area of the average color of each document area held in the internal storage unit 12 or the large-capacity storage unit 14 A document in which the total area of a document area having an average color that matches the component range condition is obtained, and the area proportional to the total area is filled with the color palette color at the screen position associated with the color component range condition. Display information.

  The document color information display unit 91 allows the user to confirm at a glance the distribution of colors (average colors) included in the document to be evaluated.

  As shown in FIG. 12, the color image 92 divides the area of the output screen into a plurality of divided screen areas 95, and assigns a condition for the feature amount included in the database in step S4 to each divided screen area 95 in advance. In step S4, the expression information extracted from the condition is displayed in the divided screen area 95 to which the condition used when extracting the information related to the impression of the document is assigned.

  Specifically, as shown in FIG. 13, a display database in which the position information of each divided screen area 95 and the condition for the feature amount included in the database shown in FIG. The expression information is displayed at the position of the screen of the matching condition with reference to the database by storing in the large-capacity storage unit 14. At this time, when assigning conditions to the divided screen area 95, it is preferable to assign conditions closer to the impression that humans feel from the document to the divided screen areas 95 that are close to each other.

  For example, in the color image 92 of FIG. 14, conditions that give a “warm” impression are assigned to the left side of the color image 92, and conversely, conditions that give a “cold” impression are assigned to the right side of the color image 92. In addition, a condition that gives a “soft” impression is assigned to the upper side of the color image 92, and conversely, a condition that gives a “hard” impression is assigned to the lower side of the color image 92. However, the assignment is not limited to these, and the assignment may be changed according to information required by the user.

For example, by assigning a condition in which the a ^* component increases from the left side to the right side of the screen area and the b ^* component increases from the lower side to the upper side of the screen area, , It can be assigned to a divided screen area 95 that is closer. The same applies to other feature amounts. In this way, by assigning conditions with a tendency, it is possible to know the tendency for the impression of the document to be evaluated from the position and concentration of the information displayed on the color image 92.

  Further, a display database is registered for each purpose of document expression, and processing is performed by changing the display database to be referred to according to the purpose of expression of the evaluated document determined in step S2. In the present embodiment, as shown in FIG. 13, a display database is prepared for documents whose information content is “electric appliances”, “food”,.

  For example, the information content of the document 2 which is one of the evaluated documents is “electrical appliance”, and the feature amount acquired from the document 2 matches the condition A2 in FIG. When the information is extracted, referring to the display database in FIG. 13, as shown in FIGS. 9 and 14, the expression information is displayed in the divided screen area 96 to which the condition A <b> 1 in the color image 92 is assigned. "Is displayed.

  Furthermore, it is preferable to simultaneously display information indicating feature amounts that match the conditions. For example, as shown in FIG. 9 and FIG. 14, an area 96 proportional to the area of the document to be evaluated that matches the condition A1 is displayed with the average color of the document to be evaluated. Thus, by simultaneously displaying text information indicating the impression of the document and information indicating the feature amount on the color image 92, the user can determine what feature amount is used to evaluate the impression of the document. It can be easily recognized in association with the feature amount.

Of course, when processing is performed by combining a plurality of expressions in step S5, the processing in step S6 can also be performed using a display database that combines a plurality of expressions. For example, in step S5, as shown in FIG. 15, a combination of "Electronics" and "car", "food" and for the combination ... of "clothing", each characteristic quantity ^{L *} component, a Expression information for the document to be evaluated can be extracted using an expression information database in which expression information is associated with a condition combining the ^* component and the b ^* component. In this case, in step S6, as shown in FIG. 16, a combination of "Electronics" and "car", the combination ... for "food" and "clothing", the feature of ^{the L *} component, a By performing processing using the display database registered by associating the display position of the display information with the combination of the ^* component and the b ^* component, the feature amount condition used for the expression information extraction processing in step S5 is obtained. The expression information extracted in step S5 can be presented at the associated display position.

The same processing can be performed when the document to be evaluated is classified into “person”, “landscape”, “object”,... For example, in step S5, as shown in FIG. 17, a condition that combines the L ^* component, a ^* component, and b ^* component of the feature quantity with respect to “person”, “scenery”, “object”,. Expression information for the document to be evaluated can be extracted using an expression information database in which expression information is associated with. In this case, in step S6, as shown in FIG. 18, a condition in which the L ^* component, a ^* component, and b ^* component of the feature quantity are combined with “person”, “landscape”, “object”. By performing processing using the display database registered in association with the display position of the display information, the display position associated with the feature amount condition used in the expression information extraction process in step S5 is set in step S5. The extracted expression information can be presented.

  The same processing can also be performed when impression evaluation is performed by classifying evaluated documents according to information providing destinations. That is, as shown in FIG. 19, it is preferable to use a display database in which the display position of the expression information is associated with the feature amount condition for each information providing destination. In this case, the expression information extracted in step S5 is presented at the display position associated with the feature amount condition used for extraction of expression information in step S5.

Further, the expression information may be registered by associating the attributes of the viewer (user) serving as a document recipient alone or in combination with each other and associating them with the feature amount condition. For example, for the combination of “gender”, “age”, and “residential area”, the expression information associated with the combination of the L ^* component, the a ^* component, and the b ^* component of the feature amount is stored in the expression information database. You may register.

  Further, in the present embodiment, different expression information databases and display databases are used for each purpose of document expression, but processing is performed using the same expression information database and display database regardless of the purpose of document expression. It can also be done. For example, the feature amount extracted in step S4 is corrected according to the purpose of document expression, the expression information associated with the condition that the corrected feature amount satisfies is extracted with reference to the expression information database, Expression information or the like is displayed in the divided screen area on the color image associated with the condition that the feature amount corrected with reference to the display database satisfies.

<Second document impression evaluation method>
The second document impression evaluation method will be described below. In the second document impression evaluation method, a color palette including a plurality of colors is prepared in advance, a database is prepared in which expression information representing the impression of the document is assigned to a combination of the colors included in the color palette. A feature amount relating to the color is extracted from the evaluation document, and the expression information assigned to the color on the color palette corresponding to the feature amount is selected and displayed.

  A document impression evaluation processing method using a color palette will be described below with reference to the drawings. As in the first document impression evaluation method, as shown in FIG. 2, input of the evaluated document (step S1), determination of the purpose of expressing the evaluated document (step S2), division of the evaluated document (step S3), including a subroutine for obtaining physical features (step S4), impression evaluation extraction (step S5), and evaluation result output (step S6), and is executed by sequentially performing each step.

  In steps S1 to S3, processing is performed in the same manner as in the first document impression evaluation method, and a description thereof will be omitted.

  In step S4, a feature amount is calculated for each document area, and the feature amount relating to the color in the document region is associated with the color on the color palette. Step S4 is processed as a subroutine as shown in FIG.

In step S41, the color space of the document data in each document area is converted from the RGB space to the L ^* a ^* b ^* space. In step S42, the document data of each document area subjected to color space conversion is stored and held in the large-capacity storage unit 14 or the internal storage unit 12. Since these processes are the same as those in the first document impression evaluation method, description thereof will be omitted.

  In step S47, the feature quantity of the position, area, color dispersion, and number of colors is extracted for each document area, and the color on the color palette that approximates the feature quantity of the color average of each document area is further characterized. Calculate as a quantity.

  Since the calculation method of the feature amount regarding the position, area, color dispersion, and number of colors of each document area is the same as the first document impression evaluation method, description thereof is omitted.

  The feature amount of the color on the color palette is obtained by performing pattern matching based on the color average of each document area calculated in the same manner as the first document impression evaluation method, so that the color palette is closest to the color average. Find the color.

  In step S48, the base color, sub color, and accent color are calculated based on the color palette color obtained for each document area and the area of the document area. Here, the base color is a color used in the maximum area in the document to be evaluated. The sub color is a color in which the color difference from the base color is not less than a predetermined value and two colors used in a wider area are selected in order. The accent color is a color selected from two colors in order from the one in which the area used is equal to or less than a predetermined ratio to the area of the document to be evaluated and the color difference from the base color is large. FIG. 21 illustrates one selected base color, two sub-colors, and two accent colors. However, the color selection conditions are not limited to these, and any color that clearly shows the characteristics of the document to be evaluated may be used.

  In step S5, based on the selected color, that is, the base color, the sub color, and the accent color, information on the impression given to the human by the evaluated document is extracted. Specifically, an expression information database is created in advance by associating conditions that combine base colors, sub-colors, or accent colors with expression information related to impressions received by humans from the combinations, and select the selected base color and sub-color. Expression information associated with a condition that matches the combination of color and accent color is extracted.

  At this time, also in the present embodiment, the evaluation is performed using the expression information database in which different expression information is associated with the condition combining the base color, the sub color, or the accent color according to the type of the document to be evaluated.

  An example will be described using the database shown in FIG. This database includes a database that uses a combination of base colors and two sub-colors as a condition, and a database that uses a combination of two accent colors. However, the combination conditions of the base color, the sub color, and the accent color are not limited to these, and other combinations may be used as the conditions.

  Also in the present embodiment, the evaluation is performed using an expression information database in which different expression information is associated with the same feature value depending on the purpose of expression of the evaluated document. Referring to these databases, for example, it is determined in step S2 that the product represented by the evaluated document is an “electric appliance”, and the base color, sub color, and accent selected from the evaluated document as shown in FIG. When the color is extracted, the text information “classic” is extracted.

  In step S6, the output unit 16 displays the selected base color, sub color, accent color, and extracted expression information. The output is preferably performed in such a manner that the user can easily grasp information about the document.

  For example, as shown in FIG. 23, a document 160 to be evaluated 160, a document color information display unit 161 indicating the color area on the color palette included in the document to be evaluated, and a color image that displays expression information extracted from the selected color. It is preferable to include a display unit 163 that displays 162 and the selected color.

  Similarly to the first document impression evaluation method, the document color information display unit 161 displays the color area on the color palette included in the document to be evaluated and the screen area proportional to the area on the color palette. It is displayed filled with.

  The color image 162 divides an output screen area into a plurality of divided screen areas, assigns a condition composed of a combination of a base color, a sub color, and an accent color to each divided screen area, and is used when extracting expression information The expression information extracted from the condition is displayed in the area to which the condition that matches is assigned. In addition, it is also preferable to display the base color, the sub color, and the accent color used for extracting the expression information.

  At this time, as in the first embodiment, a display database is prepared for each purpose of document expression, and the display database corresponding to the purpose of expression of the evaluated document determined in step S2 is referred to. To determine the screen position for displaying expression information and color. Since the specific method is the same as the first document impression evaluation method, the description thereof is omitted here.

  As described above, according to the first and second document impression evaluation methods, information on the impression that the document gives to the human from the feature amount suitable for evaluating the impression that the document gives to the human and the condition that the combination satisfies. Can be presented automatically. Therefore, even a user who has little experience in document creation can obtain information regarding an accurate impression only by inputting document data. In addition to the information related to the impression that the document gives to humans, the feature quantities used in the evaluation can be presented in an easy-to-understand manner for the user, and the user can easily grasp general information about the document. be able to.

  In addition, by using a database that is suitable for evaluation according to the purpose of document representation, or by correcting features according to the purpose of document representation, the impression of the document can be made appropriate for the purpose of document representation. Can be evaluated.

<Third document impression evaluation method>
In the case of images and documents in which content such as hypertext format is structured, document segmentation and feature extraction can be automatically performed by analyzing the syntax of the text information that composes the content. it can.

  Hereinafter, a processing method for impression evaluation of a structured document will be described with reference to the drawings. Similar to the first document impression evaluation method, input of the evaluated document (step S1), determination of the purpose of expression of the evaluated document (step S2), division of the evaluated document (step S3), physical feature amount Including a subroutine (step S4) for acquiring the image, extraction of impression evaluation (step S5), and output of the evaluation result (step S6), which are executed by sequentially performing each step.

  In step S1, an evaluated document is acquired in the same manner as in the first document impression evaluation method. In step S2, the purpose of expressing the evaluated document is determined. Since the process here is the same as that of the first embodiment, a description thereof will be omitted.

In step S3, a process for dividing the document to be evaluated is performed. Here, when the document to be evaluated is divided into a plurality of regions, the structured document data is analyzed and divided into divided document regions for each structured content. For example, when a document described in the HTML format read from the network 26 is an evaluation target, the content enclosed by each tag included in the document data is set as one divided document area. For example, a portion between a <head> tag and a </ head> tag can be divided as a header portion.

  In step S4, a feature amount is extracted from each document area divided in step S2. That is, information such as color, size, and arrangement for each content divided as each document area is extracted as a feature amount.

  In step S5 and step S6, since the process is performed in the same manner as the first document impression evaluation method, the description is omitted.

  As described above, according to the third document impression evaluation method, it is possible to extract feature amounts from structured document data such as a hypertext format and present information related to impressions given to humans by the document. Furthermore, by combining the document data in the bitmap format of the document to be evaluated and the above parsing, more accurate area division can be performed.

It is a block diagram of the structure of the document impression evaluation apparatus in this Embodiment. It is a figure which shows the flowchart of the document impression evaluation method in this Embodiment. It is a figure which shows the example of the document used as evaluation object. It is a figure which shows the flowchart of step S4 of the 1st document impression evaluation method in embodiment of this invention. It is a figure which shows the Munsell hue ring in uniform color space. It is a figure which shows the example of the feature-value acquired with the document impression evaluation method in embodiment of this invention. It is a figure which shows the example of the expression information database used with the document impression evaluation method in embodiment of this invention. It is a figure which shows another example of the expression information database used with the document impression evaluation method in embodiment of this invention. It is a figure which shows the example of an output of the evaluation result of the document impression evaluation method in embodiment of this invention. It is explanatory drawing of the method of displaying the document color information display part of the document impression evaluation method in embodiment of this invention. It is explanatory drawing of the method of displaying the document color information display part of the document impression evaluation method in embodiment of this invention. It is explanatory drawing of the method of displaying the color image of the document impression evaluation method in embodiment of this invention. It is an example of the database for a display for displaying the color image of the document impression evaluation method in embodiment of this invention. It is a figure which shows the example of a display of the color image of the document impression evaluation method in embodiment of this invention. It is a figure which shows another example of the expression information database used with the document impression evaluation method in embodiment of this invention. It is another example of the database for a display for displaying the color image of the document impression evaluation method in embodiment of this invention. It is a figure which shows another example of the expression information database used with the document impression evaluation method in embodiment of this invention. It is another example of the database for a display for displaying the color image of the document impression evaluation method in embodiment of this invention. It is another example of the database for a display for displaying the color image of the document impression evaluation method in embodiment of this invention. It is a figure which shows the flowchart of step S4 of the 2nd document impression evaluation method in embodiment of this invention. It is a figure which shows the example of the base color, sub color, and accent color which were acquired with the 2nd document impression evaluation method in embodiment of this invention. It is a figure which shows the example of the expression information database used with the 2nd document impression evaluation method in embodiment of this invention. It is a figure which shows the example of an output of the evaluation result of the 2nd document impression evaluation method in embodiment of this invention.

Explanation of symbols

  10 control section, 12 internal storage section, 14 large capacity storage section, 16 output section, 18 standard input section, 20 document input section, 22 interface section, 24 bus, 26 network, 30 header section, 31 title section, 32 header section , 33 Link button part, 34 Image image part, 35 Text part, 90 Document to be evaluated, 91 Document color information display part, 92 Color image, 93 Average color display part, 95 Split screen area, 96 Display area, 160 Document to be evaluated 161 Document color information display section, 162 color image, 163 color display section.

Claims (4)

A document impression evaluation device that presents information about an impression given to a human by the document based on visual information extracted from the document,
A document discriminating means for discriminating a classification category associated with a keyword included in an evaluated document to be evaluated with reference to a classification database in which keywords related to document contents are associated with classification categories ;
Feature quantity extraction means for extracting at least one feature quantity from the evaluated document;
For each classification category , the document discrimination means refers to an expression information database that associates a condition for at least one visual feature amount related to a document with expression information representing an impression felt by a person who views the document. Expression information extraction means for extracting expression information associated with the classification category of the obtained document to be evaluated and the condition that the feature amount of the document to be evaluated satisfies,
An output means for outputting the extracted expression information as information on an impression given to a human by the document to be evaluated;
Equipped with a,
The evaluated document is an advertisement of a product,
The classification is represented by at least one of a product type, service content, or viewer characteristics,
Document impression evaluation device. The document impression evaluation device according to claim 1 ,
The output means divides the area of the output screen into a plurality of divided screen areas, assigns the condition for each of the divided screen areas, and assigns the condition to the divided screen area corresponding to the condition satisfied by the feature amount of the evaluated document. A document impression evaluation apparatus characterized by displaying extracted expression information. The document impression evaluation device according to claim 1 or 2 ,
The output means divides the area of the output screen into a plurality of divided screen areas, assigns the condition for each of the divided screen areas, and assigns the condition to the divided screen area corresponding to the condition satisfied by the feature amount of the evaluated document. A document impression evaluation apparatus characterized by displaying information on a feature amount of a document to be evaluated. A document impression evaluation program for performing a document impression evaluation based on visual information extracted from a document and presenting information on an impression given to a human by the document,
Computer
A document discriminating means for discriminating a classification category associated with a keyword included in an evaluated document to be evaluated with reference to a classification database in which keywords related to document contents are associated with classification categories ;
Feature quantity extraction means for extracting at least one feature quantity from the evaluated document;
For each classification category , the document discrimination means refers to an expression information database that associates a condition for at least one visual feature amount related to a document with expression information representing an impression felt by a person who views the document. Expression information extraction means for extracting expression information associated with the classification category of the obtained document to be evaluated and the condition that the feature amount of the document to be evaluated satisfies,
An output means for outputting the extracted expression information as information on an impression given to a human by the document to be evaluated;
To function as,
The evaluated document is an advertisement of a product,
The classification is represented by at least one of a product type, service content, or viewer characteristics.
Document impression evaluation program characterized by that.

Images