JP2004102792A - Image retrieval device and image retrieval method, program, and medium - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image retrieval device and an image retrieval method capable of easily retrieving image even when key words fixed to image data are not recognized, a program, and a recording medium. <P>SOLUTION: This image retrieval device 11 comprises a ROM 13 storing various types of information tables, a digital camera 18, and a peripheral controller part 16 controlling the connection thereof to a CRT 15. The image data generated by the digital camera 18 is acquired together with the key words given by an operator and the criticality thereof, and a group into which key words are classified, the representative image data thereof, and the criticality level thereof are determined by the information tables stored in the ROM 13. In a retrieval screen displayed on a CRT 15, key words are arranged in a key word display part 63 based on the group and criticality level in the state of allowing to select registration in, adding to, and deletion from the conditions for retrieving specified image data, and the representative image data of the key words is displayed in a representative image data display part 64. <P>COPYRIGHT: (C)2004,JPO

Description

[0001]
TECHNICAL FIELD OF THE INVENTION
The present invention relates to an image search method and method, and a program and a medium, and more particularly, to an image search apparatus and method for displaying and searching a plurality of image data shot by an electronic still camera such as an electronic camera or a video camera, and a method thereof. Related to programs and media.
[0002]
[Prior art]
2. Description of the Related Art In recent years, electronic still cameras that store photographs as digital signals on a memory card or the like have been provided. Image data such as photos taken with these electronic still cameras requires more physical space by storing them in portable memory such as optical memory than photos taken with conventional silver halide film cameras. It is possible to save a large amount without doing.
[0003]
Conventionally, as a method of searching for desired image data from such a large amount of stored image data (hereinafter referred to as “image data search”), a keyword is assigned to each image data, and the assigned There is a method of performing a search by comparing a keyword with a search condition designated by an operator (hereinafter, referred to as “keyword search”). The specification of the search condition in the keyword search is performed by the operator inputting a plurality of keywords specified by a natural language sentence, a logical operator, or the like into an input field prepared in advance. In addition, the method of inputting a keyword into the input field requires that the operator select one of the keywords having the same first character as the first character input by the operator and searched and listed from the image database. Some are done by
[0004]
Further, as a keyword search method, there is a method of performing a search by providing a directory structure having a semantically hierarchized hierarchy in advance and using the hierarchized name as a keyword, presenting these hierarchies, and assigning or classifying the keyword ( For example, see Patent Document 1).
[0005]
[Patent Document 1]
JP-A-7-129621
[0006]
[Problems to be solved by the invention]
However, in the above-described image data search, in particular, in the keyword search, it is necessary to find an appropriate keyword and to input it into an input field in order to search for desired image data.
[0007]
For example, in the keyword search, if the keyword of the stored image data is recognized in advance, it is possible to easily perform the search. While doing the search.
[0008]
Also, for a method of performing a search by having a directory structure that is semantically hierarchized in advance and having the hierarchized name as a keyword, and presenting these hierarchies and assigning or classifying the keyword, the operator pre-hierarchizes. You need to understand. Further, it is troublesome to follow the hierarchical structure every time a keyword is assigned.
[0009]
In the method of searching for hierarchical data, only one keyword can be searched as a search condition. For example, when searching for image data of "children in the sea", the operator searches for "children in the sea" by either the "sea" or "children" in the manner described above, and the operator himself searches for "children in the sea". Needed to search.
[0010]
An object of the present invention is to provide an image search method and an image search method capable of easily performing a search without recognizing a keyword assigned to each image data, and a program and a medium.
[0011]
[Means for Solving the Problems]
In order to achieve the above object, the image search device according to claim 1, wherein a registration unit that registers a plurality of image data and a keyword to be assigned to the image data, wherein the registration is performed using the registered keyword as a search condition. In an image search device including a search unit that searches for desired image data, the registration unit obtains an image data obtaining unit that obtains the plurality of image data, and obtains a keyword assigned to the obtained image data. Storage means for storing keyword information, a first dictionary for managing semantic classification information, a second dictionary for managing part-of-speech information, and a keyword and an information table for managing the classification of the keyword. A keyword discriminator for discriminating whether or not the keyword thus determined is a keyword managed by the information table. Means for acquiring semantic classification information of the keyword from the first dictionary, part of speech acquiring means for acquiring the part of speech information of the keyword from the second dictionary, and a keyword for which the acquired keyword is managed. If not, a classification determining unit that determines the classification of the keyword based on the semantic classification information obtained for the keyword by the semantic classification obtaining unit and the part of speech information obtained for the keyword by the part of speech obtaining unit. The search means includes display means for displaying the classified image data and a keyword assigned to the image data.
[0012]
The image retrieval device according to claim 2 is the image retrieval device according to claim 1, wherein the semantic classification acquisition unit is configured to execute the method when the semantic classification information of the acquired keyword is not managed by the first dictionary. It is characterized in that the classification of the keyword is acquired based only on the acquired part-of-speech information.
[0013]
The image search device according to claim 3 is the image search device according to claim 1, wherein, when the keyword obtained by the keyword determination unit is present, the classification managed as the classification of the keyword in the information table is the keyword. It is characterized by including another classification determining means for determining as the classification of.
[0014]
According to a fourth aspect of the present invention, in the image retrieval apparatus according to the third aspect, the keyword display unit is displayed by a classification display unit that displays a classification managed by the information table and a classification display unit. For each classification, there is provided a keyword-in-class display means for displaying a keyword determined to the classification by the classification determination means.
[0015]
6. The image search device according to claim 5, wherein the search means adds a keyword displayed by the keyword-in-class display means to the keyword as the search condition. A search condition deleting unit that deletes the keyword displayed by the keyword-in-class display unit from the keyword that is the search condition; and a search condition setting unit that sets the keyword that is the search condition. .
[0016]
In order to achieve the above object, an image search method according to claim 6, wherein a registration step of registering a plurality of image data and a keyword to be assigned to the image data, wherein the registration is performed using the registered keyword as a search condition. In an image search method having a search step of searching for desired image data, the registration step obtains an image data obtaining step of obtaining the plurality of image data, and obtains a keyword assigned to the obtained image data. Acquiring a keyword, a first dictionary for managing semantic classification information, a second dictionary for managing part-of-speech information, and a step of storing a keyword and an information table for managing the classification of the keyword. Determine whether the keyword is in the keyword managed by the information table A keyword discriminating step, a semantic classification obtaining step of obtaining keyword semantic classification information from the first dictionary, a part of speech obtaining step of obtaining keyword part of speech information from the second dictionary, and When the keyword is not a keyword, the classification is determined based on the semantic classification information obtained for the keyword in the semantic classification obtaining step and the part of speech obtained for the keyword in the part of speech obtaining step. And a display step of displaying the classified image data and a keyword assigned to the image data.
[0017]
The image search method according to claim 7, wherein, in the image search method according to claim 6, when the semantic classification information of the obtained keyword is not managed by the first dictionary, It is characterized in that the classification of the keyword is acquired based only on the acquired part-of-speech information.
[0018]
The image search method according to claim 8 is the image search method according to claim 6, wherein when the keyword obtained in the keyword determination step is present, the classification managed as the classification of the keyword in the information table is changed to the keyword. Characterized by having another classification determining step of determining as the classification of (1).
[0019]
In the image search method according to the ninth aspect, in the image search method according to the eighth aspect, the keyword display step is performed by a classification display step of displaying a classification managed by the information table and a classification display step. A keyword display step of displaying, for each classification, a keyword determined in the classification by the classification determination step;
[0020]
The image search method according to claim 10, wherein the search step is a search condition adding step of adding a keyword displayed by the keyword-in-class display step to a keyword that is the search condition. And a search condition deleting step of deleting a search condition for deleting and specifying the keyword displayed in the classified keyword display step from the keyword as the search condition, and a search condition setting step of setting the keyword as the search condition. It is characterized by the following.
[0021]
In order to achieve the above object, a program according to claim 11, further comprising a registration step of registering a plurality of image data and a keyword to be assigned to the image data, and the registered image data being registered using the registered keyword as a search condition. A program for causing a computer to execute a search step of searching for a desired one of the plurality of image data, wherein the registering step obtains an image data obtaining step of obtaining the plurality of image data and a keyword assigned to the obtained image data. A keyword obtaining step, a first dictionary for managing semantic classification information, a second dictionary for managing part-of-speech information, and a storing step of storing a keyword and an information table for managing the classification of the keyword, Keywords managed by the information table for keywords A keyword determining step of determining whether or not the keyword is present, a semantic classification obtaining step of obtaining keyword semantic classification information from the first dictionary, a part of speech obtaining step of obtaining keyword part of speech information from the second dictionary, When the acquired keyword is not the managed keyword, the keyword is determined based on the semantic classification information acquired for the keyword in the semantic classification acquisition step and the part of speech information acquired for the keyword in the part of speech acquisition step. And a step of displaying the classified image data and a keyword assigned to the image data.
[0022]
In order to achieve the above object, the medium according to claim 12, wherein a registration step of registering a plurality of image data and a keyword to be assigned to the image data, and the registered image data using the registered keyword as a search condition. A computer-readable medium storing a program for causing a computer to execute a search step of searching for a desired one of the plurality of image data, wherein the registration step includes: an image data acquisition step of acquiring the plurality of image data; A keyword acquiring step for acquiring a keyword to be added to data, a first dictionary for managing semantic classification information, a second dictionary for managing part-of-speech information, and an information table for managing keywords and the classification of the keyword are stored. Storage means for storing the information keyword in response to the acquired keyword. A keyword discriminating step of discriminating whether or not the keyword is managed by the keyword, a semantic classification acquiring step of acquiring semantic classification information of the keyword from the first dictionary, and a part of speech information of the keyword from the second dictionary. The part of speech acquisition step of acquiring, and when the acquired keyword is not the managed keyword, the semantic classification information acquired for the keyword by the semantic classification acquisition step, and the semantic classification information acquired by the part of speech acquisition step A classification determining step of determining the classification of the keyword based on the part of speech information; and an image classification step of classifying the acquired image data based on the acquired keyword. Image data and keys assigned to the image data. Characterized in that it has a display step of displaying the word.
[0023]
BEST MODE FOR CARRYING OUT THE INVENTION
Hereinafter, an image retrieval apparatus according to an embodiment of the present invention will be described in detail with reference to the drawings.
[0024]
FIG. 1 is a block diagram of an image search device according to an embodiment of the present invention.
[0025]
In FIG. 1, an image search device 11 includes a ROM 13 for storing a program for performing an image search, an information table shown in FIGS. 2 and 3 described later, a thesaurus dictionary shown in FIG. 4 described later, and a program stored in the ROM 13. The image processing apparatus includes a RAM 14 to be executed, a peripheral controller 16 for controlling connection of the image search device 11 to external devices, and a CPU 12 connected to and controlling the ROM 13, the RAM 14, and the peripheral controller 16.
[0026]
The external devices controlled by the peripheral controller 16 for connection control include a CRT 15 for displaying the processing status of the image search device 11 under the control of the CPU 12, an external storage device 17 for storing image data, and a digital device for generating image data. There are a camera 18 and a digital video camera 19, and an input device 20 such as a mouse and a keyboard.
[0027]
Image data obtained by the operator with the digital camera 18, digital video camera 19, or the like is stored in the external storage device 17 or the ROM 13.
[0028]
The ROM 13 includes an image information table shown in FIG. 2, a keyword information table shown in FIG. 3, a thesaurus dictionary shown in FIG. 4, a keyword classification information table shown in FIG. 5, a synonym dictionary shown in FIG. Save each concept word dictionary. These tables and dictionaries may be stored in the external storage device 17.
[0029]
FIG. 2 is a diagram showing an image information table stored in the ROM 13 in FIG.
[0030]
In FIG. 2, the image information table includes a file name of image data obtained by the operator with the digital camera 18 or the digital video camera 19 and an image ID for identifying the image data. For example, the image information table manages a file name “flower01.jpg” and an image ID “001” of one image data.
[0031]
In the present embodiment, an arbitrary numerical value (serial number) is used as the image ID. However, when the file name of the image data is unique, the file name of the image data may be used as the image ID.
[0032]
FIG. 3 is a diagram showing a keyword information table stored in the ROM 13 in FIG.
[0033]
3, the keyword information table includes an image ID of the image data managed in the image information table of FIG. 2, a keyword assigned to the image data, and the importance of the keyword. In addition, a value is assigned to the importance of the keyword of each image data based on a criterion arbitrarily determined by the operator. For example, "3" when the keyword assigned to the image data is the subject of the image data, "2" when the image data can be searched by the keyword, "2", the keyword of the image data If it is not important but is appropriate, “1” is assigned as the importance of the keyword to the image data. The importance of the keyword may be assigned based on a predetermined criterion, and is not limited to the criterion of the present embodiment.
[0034]
Therefore, as shown in FIG. 3, the keywords “flower”, “red”, and “park” are assigned to the image data of the image ID “001”, and “(flower = ) 3 "," (red =) 1 ", and" (park =) 1 ".
[0035]
FIG. 4 is a diagram showing a thesaurus dictionary stored in the ROM 13 in FIG.
[0036]
In FIG. 4, the thesaurus dictionary includes headwords, semantic classification of headwords, and part of speech of headwords. The headword refers to a keyword acquired in a keyword registration process of FIG. 9 described later, or a different notation / synonym / lower concept word acquired in a keyword search process of FIG. 11 described later. This is the same for the different notation synonym dictionary in FIG. 6 described later and the lower concept word dictionary in FIG. 7 described later.
[0037]
In this thesaurus dictionary, for example, the headword “park” indicates the part of speech “noun” and the semantic classification “specific object: product: abstract object: place”. At this time, it indicates that it has two meanings, “specific object (first level), manufactured product (second level)” and “abstract object (first level), location (second level)”. The headword “red” has a part of speech “adjective” and no semantic classification. In the present embodiment, the part of speech is a component of the thesaurus dictionary. However, the part of speech may not be a component of the thesaurus dictionary, but a word dictionary may be separately configured and used as the component.
[0038]
When the headword matches the keyword in the keyword information table of FIG. 3 and there is a semantic classification, the name of the second hierarchy is assumed to be the name of the classification to which the keyword belongs, and FIG. It is reflected on the keywords in the keyword classification information table.
[0039]
On the other hand, even when the headword matches the keyword in the keyword information table of FIG. 3, a word whose part of speech does not have a meaning represented by a particle, an auxiliary verb, a conjunction, a pronoun, etc. is classified as an “abstract object”. Then, it is reflected on a keyword in a keyword classification information table of FIG. 5 described later. Thereby, the classification names of all the keywords can be reliably obtained.
[0040]
Further, in the present embodiment, words that have no meaning represented by particles, auxiliary verbs, conjunctions, pronouns, etc. are classified as “abstract objects”, but the semantic classification set in other parts of speech It is not limited to this as long as it can be distinguished. In the present embodiment, the classification items are classified as “human”, “plant”, “animal”, “place”, “manufactured product”, and “abstract object”, but are not limited to this classification. The thesaurus dictionary of FIG. 4 constitutes a first dictionary and a second dictionary.
[0041]
FIG. 5 is a diagram showing a keyword classification information table stored in the ROM 13 in FIG.
[0042]
In FIG. 5, the keyword classification information table includes a keyword, a classification to which the keyword belongs, image ID of representative image data indicating the highest score value in FIG. 10 described later among image data to which the keyword is assigned, and An importance level is a value obtained by adding the importance of the keyword assigned to the image data. This importance level increases when the importance assigned to the keyword is high and the keyword is assigned to many pieces of image data. If there are a plurality of image data with the highest score values, the image data with the smaller image ID number is selected as the representative image data.
[0043]
When a plurality of keywords are managed in one classification by the keyword classification information table in FIG. 5, the keywords are managed in the keyword classification information table in FIG. 5 so that the keyword having a higher importance level is ranked higher. For example, since the importance levels of “elephant” and “dog” belonging to the same classification “animal” as the keyword “fish” are “10” and “2”, respectively, the keyword classification information table of FIG. “Fish”, “Elephant”, and “Dog” are managed in the descending order as keywords of the category “Animal”.
[0044]
FIG. 6 is a diagram showing a synonym dictionary for different notations stored in the ROM 13 in FIG.
[0045]
6. In FIG. 6, the synonym dictionary with a different notation includes a headword and a different notation or a synonym applicable to the headword. For example, “flower”, “flower”, “flower”, and the like are registered as different notations or synonyms that apply to the headword “flower”.
[0046]
FIG. 7 is a diagram showing a lower concept word dictionary stored in the ROM 13 in FIG.
[0047]
In FIG. 7, the lower concept word dictionary includes a headword, a word indicating a lower concept of the headword, and a hierarchy level indicating the hierarchy. For example, “Porsche”, “motorcycle”, and “wheel” are registered as words that fall under the subordinate concept of the headword “car”. It is given as "bike (0.64)" or "wheel (0.55)".
[0048]
FIG. 8 is a diagram showing a search screen displayed on the CRT 15 in FIG.
[0049]
The search screen displayed on the CRT 15 is controlled by the CPU 12, and includes a classification display section 61 for displaying the name of the classification, and a keyword number display section for displaying the number of keywords belonging to the classification displayed on the classification display section 61. 62, a keyword display section 63 for displaying keywords belonging to the classification, a representative image data display section 64 for displaying representative image data of the selected keyword, and a search condition input section 65 for inputting a keyword serving as a search condition. , A search execution unit 66 that transmits a search start signal to the CPU 12 when an input is made from the input device 20, a check box 67, and a scroll bar 68.
[0050]
The search condition input unit 65 includes a plurality of input units, and each input unit is configured to be able to input one keyword.
[0051]
A check box 67 is provided for each search condition input unit 65, and a check box control is selected by the operator using the input device 20 and input to the search condition input unit 65 for the check box 67 turned ON. The determined keyword is determined to be a search condition, and the keyword input to the search condition input unit 65 for the check box 67 turned off by the operator not selecting or canceling the selection using the input device 20 is the search condition. It is determined that it is not.
[0052]
As long as it is possible to determine which of the keywords input to the search condition input section 65 has been selected as the search condition, the present invention is not limited to the check box control as in the present embodiment. And so on.
[0053]
The scroll bar 68 is provided for browsing all of the keywords belonging to the same category and only the representative image data displayed only in one portion at both ends of the representative image data display section 64, and the button control is provided by the operator at the left end. When the scroll bar 68 is selected by the input device 20, representative image data having a higher importance level than the representative image data displayed on the representative image data display section 64 is displayed, and the operator presses the rightmost scroll bar 68 on the input device 20. When selected, representative image data having a lower importance level than the representative image data displayed on the representative image data display section 64 is displayed so as to be controlled.
[0054]
A keyword serving as a search condition is input directly by the operator using the input device 20 to the search condition input unit 65 or a keyword displayed on the keyword display unit 63 is selected by the operator using the input device 20 and the search condition input unit By dragging and dropping the search condition into the search condition input portion 65, the search condition is input to the search condition input portion 65.
[0055]
In FIG. 8, the image data is roughly classified into “animals”, “humans”, “plants”, “places”, “manufactured products”, and “abstract objects” as displayed on the classification display unit 61. You. Also, as shown in the keyword number display section 62, the number of keywords applicable to each classification is "50" for "animal", "100" for "human", "40" for "plant", and "place". Is "99", "product" is "12", and "abstract" is "50". Further, the keywords applicable to the classification “animal” include “fish”, “elephant”, and “dog” as displayed in the keyword display section 63. The representative image data for each keyword is displayed on the representative image data display 64 next to each keyword display 63.
[0056]
Next, an image search method executed by the image search device 11 of FIG. 1 will be described in detail with reference to the drawings. This image search method is performed after image keywords to be displayed on each keyword display unit 63 and image data to be displayed on each representative image data display unit 64 of the search screen in FIG. In a later-described image data search process shown in FIG. 12, an image data is searched using a keyword input by the operator using the search screen shown in FIG. 8 as a search condition.
[0057]
FIG. 9 is a flowchart showing the image classification processing executed by the image data search device 11 in FIG.
[0058]
In FIG. 9, first, it is determined whether or not the registration of the image data to which the keyword is to be added has been completed (step S101). If the registration has not been completed (NO in step S101), the input device 20 has already started. Image data existing in the external storage device 17 or the ROM 13 or one of image data directly obtained from the digital camera 18 or the digital video camera 19 via the peripheral controller unit 16 is obtained as registered image data (step). S102) (Image data acquisition means). At this time, the file name of the acquired registered image data and its image ID are acquired and registered in the image information table of FIG.
[0059]
Next, a keyword to be added to the registered image data obtained in step S102 is obtained from the input device 20 (step S103) (keyword obtaining means), and the importance of the obtained keyword is obtained (step S104). In the present embodiment, the importance is obtained as data input by the input device 20, but the importance may be automatically given according to the input order of the keyword.
[0060]
Thereafter, the processing of steps S103 to S104 is repeated until the acquisition of all keywords to be added to the registered image data acquired in step S102 is completed (NO in step S105).
[0061]
In step S105, when the acquisition of all the keywords to be assigned to the registered image data acquired in step S102 and the importance thereof is completed, these keywords and importance are registered in the keyword information table of FIG. 3 (step S106). , And the processing after step S101 is repeated. The registration of the keyword and the importance in the keyword information table of FIG. 3 is performed for each image ID assigned to the registered image data acquired in step S102. This makes it possible to create an index of the data registered in each information table to speed up the search.
[0062]
As a result of the determination in step S101, when all the registration of the image data to which the keyword is added is completed (YES in step S101), the keyword registered in step S106 is acquired as the registered keyword for each registered image data (step S101). S107).
[0063]
Next, for one of the registered keywords acquired in step S107, it is determined whether or not the keyword is managed by the keyword classification information table of FIG. 5 (step S108) (keyword determination means). (YES in step S108), the classification managed as the classification of this keyword in the information table is determined as the classification of this keyword, and the process proceeds to step S115. As a result, the classification of the registered keywords acquired in step S107 can be quickly determined. The determined classification is hereinafter referred to as “decision classification”.
[0064]
On the other hand, as a result of the determination in step S108, when the registered keyword acquired in step S107 is not managed, the semantic classification for the registered keyword is acquired using the thesaurus dictionary of FIG. 4 (step S109) (semantic classification acquisition). Means), the part of speech for the registered keyword is acquired (step S110) (part of speech acquisition means), and the determined classification of the registered keyword is determined with reference to the semantic classification acquired in step S109 (step S111) (classification determining means). .
[0065]
In the following step S112, the registered keyword acquired in step S107 and the determined classification determined in step S111 are registered as the keyword and the classification to which the keyword belongs in the keyword classification information table of FIG. 5, and then the registered keyword is used as a search condition sentence. The keyword search process of FIG. 10 described later is set (step S113), and the image data having the highest score value described later obtained by the keyword search process is acquired as the representative image data of the registered keyword (step S114). Until the acquisition of all registered keywords is completed (YES in step S115), the processing of steps S107 to S115 is repeated. The representative image data acquired in step S114 is registered as representative image data in the keyword classification information table of FIG.
[0066]
As a result of the determination in step S115, when the acquisition of all the registered keywords is completed (YES in step S115), the semantic classification order processing of FIG. 11 described below is performed only for the classification in which the keywords have been increased (step S116), and this processing ends. I do. The importance level of the keyword obtained by this semantic classification order processing is reflected on the importance level of the keyword classification information table of FIG. Steps S101 to S105 in FIG. 9 constitute a registration unit.
[0067]
According to the processing of FIG. 9, when the registered keyword is not stored in the keyword classification information table of FIG. 5 (NO in step S108), the registered keyword is based on the semantic classification and the part of speech acquired from the thesaurus dictionary of FIG. Thus, the classification of the registered keyword is determined (steps S109 to S111), so that the classification of the image data can be performed reliably.
[0068]
FIG. 10 is a flowchart showing the keyword search process executed in step S113 of FIG.
[0069]
In FIG. 10, first, the registered keyword acquired in step S107 of FIG. 9 is acquired as a search condition sentence (step S301), and the search condition sentence is divided into words by natural language processing (step S302). For this natural language processing, the existing natural language processing is used, and the description is omitted. Further, of the divided words, words classified as “abstract objects” in step S109 in FIG. 9 are deleted from the divided words.
[0070]
Next, in each of the words divided in step S302, a different notation and a synonym are acquired from the different notation synonym dictionary of FIG. 6 (step S303), and the lower concept word and its hierarchical level are converted to the lower concept word of FIG. The information is acquired using a dictionary (step S304).
[0071]
Until the notation, synonym, lower concept word and its hierarchical level are obtained for all the divided words (NO in step S305), the processing of steps S303 to S304 is repeated, and all the words are obtained. (YES in step S305), an image is searched using all the words acquired by the processing in steps S301 to S304 as search condition sentences (step S306).
[0072]
In step S306, the registered keyword acquired in step S301, the word divided in step S302, the notation and synonym acquired in step S303, the lower concept word acquired in step S304, and the keyword in the keyword information table in FIG. When there is a matching keyword (hereinafter referred to as a “matching keyword”) in the keyword information table of FIG. 3, the image data of the matching keyword is used as the image data for score calculation as a result of the image search. get.
[0073]
A score is calculated for the acquired image data (step S307). The score calculation is to calculate a score value that is the sum of the total value of the matching levels of the registered keywords acquired in step S301 and the matching keywords obtained in step S306 and the importance assigned to each matching keyword. Say. Further, in the present embodiment, the matching level of the registered keyword is 3 when the keyword matches the keyword in the keyword information table of FIG. 3 and 2 when the word divided from the keyword is in the synonym dictionary in FIG. When a word divided from a keyword exists in the lower concept word dictionary in FIG. 7, it is assumed that one is added to the hierarchy level. The method of obtaining the score value is not limited to the method according to the embodiment of the present invention as long as it is based on a certain standard.
[0074]
Until the calculation of the scores of all the registered keywords acquired in step S301 is completed (NO in step S308), the process of step S307 is repeated. When the calculation of the scores of all the registered keywords is completed (YES in step S308), This processing ends.
[0075]
According to the process of FIG. 10, an image search is performed using a registered keyword, a word obtained by dividing the word, a different notation, a synonym, and a lower concept word for all the divided words as search conditions (steps S301 to S306). Since the score calculation for obtaining the sum of the matching level and the importance of the obtained matching keyword is performed for each registered keyword (step S307), the image data having the highest score value among the image data for score calculation is used as the representative image of the registered keyword. By displaying the data on the representative image data display unit 64 of the search screen in FIG. 8, the search can be easily performed without recognizing the keyword assigned to the image data.
[0076]
FIG. 11 is a flowchart showing the semantic classification order processing executed in step S116 in FIG.
[0077]
This processing is performed only for the classification in which the number of keywords is increased, and the initial value of the importance level is set to 0.
[0078]
11, first, one of the decision categories determined in step S111 is obtained from the keyword classification information table of FIG. 5 (step S401), and one of the keywords belonging to the decision category is also obtained from the keyword classification information table of FIG. Obtain (step S402).
[0079]
Next, the importance of the keyword acquired in step S402 is acquired from the keyword information table of FIG. 3 (step S403), and is added to the importance level (step S404).
[0080]
Thereafter, the processing of steps S402 to S403 is repeated until the acquisition of step S402 is completed for all keywords belonging to the determined category in step S401 (YES in step S404). The importance level of each keyword obtained by this processing is registered in the keyword classification information table of FIG.
[0081]
Next, it is determined whether or not the processing of steps S402 to S404 has been completed for all of the determination categories determined in step S111 of FIG. 9, and if completed (YES in step S405), this processing ends.
[0082]
According to the semantic classification ranking process of FIG. 11, the importance level is calculated for each keyword belonging to the determined classification (steps S401 to S404), and the keywords displayed on the keyword display unit 63 of the search screen of FIG. By arranging based on the degree level, a search can be easily performed without recognizing the keyword assigned to the image data.
[0083]
FIG. 12 is a flowchart showing an image data search process executed by the image data search device 11 in FIG.
[0084]
12, first, a classification, a keyword belonging to the classification, and representative image data corresponding to the keyword are obtained in descending order of importance level from the keyword classification information table in FIG. 5, and further applied to each classification. The number of keywords is also acquired and stored in the RAM 14 (step S201). Thereafter, the display is displayed on the classification display unit 61 of the search screen in FIG. 8 in a state where duplicates of the determined classifications acquired in step S201 are removed (step S202), and the number of keywords for the displayed determined classification is displayed on the search screen in FIG. It is displayed on the keyword number display section 62 (step S203).
[0085]
Next, the keywords and their representative image data are displayed on the keyword display unit 63 and the representative image data display unit 64 of the search screen in FIG. 8 in the order acquired in step S201 (steps S204 to S205), and are displayed in the classification in step S202. Until the keyword ends (YES in step S206), the processing in steps S204 and S205 is repeated. Until the classification to be displayed is completed (YES in step S207), the processing in steps S202 to S206 is repeated. The processing in steps S201 to S207 in FIG. 12 constitutes a display unit.
[0086]
By the processing of steps S201 to S207 described above, at the time of retrieval, the categories managed by the information table and the keywords belonging to each category are displayed (step S202). The search can be easily performed without recognizing the keyword assigned to each image data.
[0087]
Further, since the keywords belonging to each category are displayed in the order of the importance level (step S204), the search can be easily performed even if the keyword assigned to each image data registered in advance in the image search device 11 is not recognized. It can be carried out.
[0088]
The display processing of the search condition input unit 65, the search execution unit 66, and the check box 67 of the search screen of FIG. 8 is not particularly described, but is similarly displayed on the search screen of FIG.
[0089]
Thereafter, when a keyword is input to at least one search condition input unit 65 and the check box 67 corresponding thereto is turned on (YES in step S208), the keyword input to search condition input unit 65 is performed in step S217 described later. The search condition sentence is obtained in the keyword search of FIG.
[0090]
The keyword set as a search condition sentence in the search condition input unit 65 can also be obtained by the operator selecting a keyword displayed on the keyword display unit 63 using the input device 20 and dragging and dropping the keyword to the search condition input unit 65. Is entered.
[0091]
Next, it is determined whether or not a keyword has been additionally specified in the search condition input unit 65 (step S209). Specifically, it is determined whether or not there is a check box 67 newly turned ON after the determination in step S208. The method of additionally specifying a keyword is not limited to the present embodiment, and a different method may be used depending on the search screen.
[0092]
As a result of the determination in step S209, when a keyword is additionally specified, the keyword is also set as a search condition sentence acquired in the keyword search in FIG. 10 performed in step S217 described later (step S210). On the other hand, if the result of determination in step S209 is that no keyword has been designated for addition, it is determined whether or not the keyword has been designated for deletion (step S211). Specifically, it is determined whether or not any of the check boxes 67 that were ON at the time of the determination in step S208 has been turned OFF. Also in the method of designating a keyword to be deleted, similarly to the method of designating an additional keyword, the present invention is not limited to the present embodiment, and a different method may be used depending on a search screen.
[0093]
If the result of determination in step S211 indicates that a keyword has been designated for deletion, this keyword is deleted from the search condition sentence acquired in the keyword search of FIG. 10 performed in step S217 described below (step S212). On the other hand, if the result of determination in step S211 is that the keyword has not been designated for deletion, it is determined whether a search execution state has been established (step S215). This determination is specifically performed by determining whether the search execution unit 66 has been selected.
[0094]
As a result of step S215, when the search is in the search execution state, the keyword in the search condition input unit 65 whose check box 67 is currently ON is acquired as a search condition sentence (step S216). On the other hand, if the result of determination in step S215 is that the state is not a search execution state, processing from step S208 is performed.
[0095]
Thereafter, the above-described keyword search process of FIG. 10 is performed using the keyword acquired in step S216 as a search condition sentence (step S217), and the image data file names are acquired from the image IDs in the order acquired in this process. Is displayed (step S218), and the process ends. The processing of steps S208 to S218 in FIG. 12 constitutes a search unit.
[0096]
By the processing of steps S208 to S218 described above, at the time of search, the keyword displayed as a keyword in the category is added to the keyword as the search condition (step S210), and the keyword is displayed as a keyword in the category from the keyword as the search condition. The deleted keyword is deleted (step S212), and a keyword serving as a search condition is set (step S216), so that a plurality of search conditions can be specified, thereby improving the operability of image search.
[0097]
Hereinafter, each process described above will be described using a specific example.
[0098]
First, a specific example of the image classification processing in FIG. 9 will be described.
[0099]
In the present embodiment, this image classification processing is performed on a red flower blooming in a park captured by the digital camera 18.
[0100]
First, in step S102 of FIG. 9, the image data captured by the digital camera 18 is obtained as registered image data with a data file name “flower01.jpg”. Specifically, for the acquired registered image data, the image data file name “flower01.jpg” and the image ID “001” are registered in the image information table of FIG.
[0101]
Next, in steps S103 to S104, when the keyword “flower” and its importance “3” are input by the input device 20 by the operator, the input keyword and importance are added to the registered image data in step S102. It is acquired as a keyword and importance to be performed.
[0102]
Until the operator inputs the keyword and its importance (NO in step S106), the processing of steps S103 to S104 is repeated, and the keyword "red" and its importance "1", the keyword "park" and its importance ""1" is also acquired as the keyword and importance to be added to the registered image data in step S102.
[0103]
Next, the process proceeds to step S106, where the image ID “001”, the keyword “flower” and the importance “3”, and the image ID “001”, the keyword “red” and the importance “1” are stored in the keyword information table of FIG. , The image ID “001”, the keyword “park” and the importance “1” are registered.
[0104]
Since registration of image data is performed only for one image data in the present embodiment, after the above processing is performed, it is determined in step S101 that registration of image data has been completed, and the process proceeds to step S107.
[0105]
Next, classification processing is performed on the registered keywords.
[0106]
First, in step S107, the keywords “flower”, “red”, and “park” registered in the keyword information table of FIG. 3 are obtained. In step S108, it is determined whether or not the keyword “flower” is managed by the keyword classification information table in FIG. At this time, it is assumed that no keyword or image data is managed in the keyword classification information table of FIG. Therefore, it is determined that the keyword “flower” is not registered in the keyword classification information table (NO in step S108), the process proceeds to step S109, and the classification for the keyword “flower” is obtained from the thesaurus dictionary of FIG. , The semantic classification “specific object: plant:” is obtained.
[0107]
Next, the part of speech “noun” is obtained from step S110. In the next step S111, the classification to which "flower" belongs is determined to be "plant" which is the second hierarchical classification of the semantic classification.
[0108]
In step S112, as the classification in the keyword classification information table of FIG. 5, after the decision classification “plant” and “flower” are registered as keywords belonging to the decision classification, in order to acquire representative image data of “flower”, In step S113, the keyword search processing of FIG. 9 is performed with the search condition set to “flower”. In step S114, the image data having the highest score value among the image data searched in step S113 is represented by the representative image data of “flower”. And
[0109]
The processing of steps S108 to S114 is also performed for the remaining registered keywords “red” and “park” (NO in step S115). At this time, as described above, since there is no data managed by the keyword classification information table of FIG. 5, the processes of steps S109 to S114 are performed for the keywords “red” and “park”.
[0110]
In step S109, since there is no semantic classification for “red” and the part of speech is “adjective”, “abstract object” is determined as the determination classification of “red”, and the semantic classification for “park” is “concrete”. Since there are two second-level classifications of “product: product: abstract: location”, two “products” and “locations” are determined as the determination categories of “park”.
[0111]
As a result of the processing up to this point, the classification “plant” and its keyword “flower”, the classification “abstract” and its keyword “red”, and the classification “manufactured” as the classification managed by the keyword classification information table of FIG. "And its keyword" park ", the classification" place "and its keyword" park "are newly registered.
[0112]
When the processing of steps S108 to S114 is completed for all registered keywords (YES in step S115), semantic classification order processing is performed in step S116.
[0113]
First, in step S401 in FIG. 11, “plant”, which is one of the determined classifications, is obtained from the keyword classification information table in FIG.
[0114]
Next, in step S402, "flower" which is a keyword belonging to the acquired decision classification "plant" is acquired.
[0115]
In step S403, the importance level of the keyword “flower” is registered in the keyword classification information table of FIG.
[0116]
In step S404, it is determined whether all the importance levels in the keyword classification information table of FIG. 5 have been registered for the keywords in the determined classification “plant”. At this time, since only the keyword “flower” is registered in the determination category “plant”, if the importance level of the keyword “flower” is registered, the processing of steps S402 to S404 is then performed for all remaining keywords. The determination is performed for the “abstract object”, “place”, and “manufactured” (YES in step S405), and the process ends.
[0117]
Through the above processing, the importance level of each keyword to which each of the determined classifications “plant”, “abstract object”, “place”, and “manufactured product” belongs is registered in the keyword classification information table of FIG.
[0118]
When the semantic classification order processing ends, the image classification processing of FIG. 9 ends with respect to the image data of the red flower blooming in the park, which is captured by the digital camera 18.
[0119]
Next, the image data search processing of FIG. 12 will be described using a specific example.
[0120]
From step S201, a classification is obtained from the keyword classification information table in FIG. 5, and a classification to be displayed after eliminating duplication is obtained. In this case, “animal”, “human”, “plant”, “place”, “product”, and “abstract object” are acquired. Also, the keyword and the representative image data are acquired in descending order of the importance level for each classification. At this time, the number of keywords corresponding to the classification is also acquired.
[0121]
Next, from step S202, the classification "animal" is displayed on the classification display section 61 on the search screen of FIG. Also, in step S203, the number of keywords “50” is displayed on the keyword number display section 62. Then, in steps S204 to S206, keywords such as "fish", "elephant", and "dog" existing in the classification of "animal" and their representative image data are displayed in the order of acquisition.
[0122]
If it is determined in step S207 that the processes in steps S202 to S206 have been performed for all the classifications, the search screen in FIG. 8 is displayed.
[0123]
Next, a specific example of the processing of steps S208 to S218, that is, the processing performed on the search screen of FIG. 8 will be described.
[0124]
When the keyword "run in the city" is input to the keyword display section 63 and the corresponding check box 67 is turned on (YES in step S208), the keyword "run in the city" is registered as a search condition sentence.
[0125]
The method of registering the keyword in the search condition sentence is not limited to the above method, but can also be performed by the operator selecting one of the image data displayed on the representative image data display unit 64 using the input device 20. . For example, when image data of a flower displayed on one of the representative image data display sections 64 is selected by the operator using the input device 20, “flower” is displayed on the keyword display section 63, and the keyword display section 63 is displayed. Is turned on, and the keyword “flower” is registered in the search condition sentence.
[0126]
Next, when "child" is added to the search condition sentence, the operator turns on the check box 67 corresponding to the keyword display section 63 after inputting the keyword "child" in another keyword display section 63. .
[0127]
When this operation is performed by the operator, it is determined in step S209 that the additional designation of the keyword has been performed, and in step S210, the keyword “child” is also added to the search condition sentence.
[0128]
The method of adding a keyword to the search condition sentence is not limited to the above-described method, and may be performed by the operator selecting one of the image data displayed on the representative image data display unit 64 using the input device 20. .
[0129]
Next, when deleting the keyword “child” from the search condition sentence, the operator turns off the check box 67 corresponding to the keyword display unit 63 in which the keyword “child” is input.
[0130]
When this operation is performed by the operator, it is determined in step S211 that the keyword has been specified to be deleted, and in step S212, the keyword “child” is deleted from the search condition sentence.
[0131]
The method of deleting the keyword from the search condition sentence is not limited to the above-described method, and may be performed, for example, by deselecting the representative image data display unit 64 already selected by the operator. For example, the keyword “flower” displayed on the keyword display unit 63 is deleted from the keyword display unit 63 by reselecting the representative image data of the keyword “flower”, and the corresponding check box 67 is also turned off. And the keyword “flower” is deleted from the search condition sentence.
[0132]
When the operator starts the search, the operator selects the search execution unit 66.
[0133]
If this operation is performed by the operator, it is determined in step S215 that the search execution state has been established, and in step S216, a keyword currently registered as a search condition sentence is set as a search condition, and in step S217, Perform keyword search processing.
[0134]
When the keywords set as the search condition sentences are “running in the city” and “car”, the search condition is searched for “car running in the city”. In the present embodiment, a plurality of keywords specified in the search condition sentence are processed as character strings combined with spaces, but the way of combining the keywords set as the search condition sentence is not limited to this. .
[0135]
Further, a difference is set in the keyword set as the search condition sentence between the case where the operator directly inputs / deletes the keyword in the search condition display unit 65 and the case where the operator selects / reselects the representative image data designation unit 64. Although not shown, the image data may be displayed separately, or keyword search processing may be separately performed as different search conditions, and the searched image data may be displayed collectively at the end.
[0136]
The present invention is achieved by supplying a software program (flow charts in FIGS. 9 to 12) for realizing the functions of the above-described embodiments to a CPU, and the CPU reads and executes the supplied program. be able to.
[0137]
In this case, the program is supplied directly from a storage medium on which the program is recorded, or is downloaded from another computer or database (not shown) connected to the Internet, a commercial network, a local area network, or the like. Supplied.
[0138]
The form of the program may be in the form of object code, program code executed by an interpreter, script data supplied to an OS (Operating System), or the like.
[0139]
The present invention can also be achieved by supplying a storage medium storing a software program for realizing the functions of the above-described embodiments to a CPU, and reading and executing the program stored in the storage medium by the CPU. can do.
[0140]
In this case, the program code itself read from the storage medium implements the functions of the above-described embodiments, and the storage medium storing the program code constitutes the present invention.
[0141]
Examples of storage media for storing program codes include ROM, RAM, NV-RAM, floppy (registered trademark) disk, hard disk, optical disk, magneto-optical disk, CD-ROM, MO, CD-R, CD-RW, and DVD. -ROM, DVD-RAM, DVD-RW, DVD + RW, magnetic tape, nonvolatile memory card, and the like.
[0142]
The functions of the above-described embodiments are not only executed by executing the program code read from the computer, but also executed by the OS or the like running on the computer in part or all of the actual processing based on the instruction of the program code. This can also be realized by:
[0143]
【The invention's effect】
As described above in detail, according to the image search device according to claim 1, the image search method according to claim 6, the program according to claim 11, and the medium according to claim 12, the image data and the image When registering a keyword to be assigned to data, acquire a plurality of image data, acquire a keyword to be assigned to the acquired image data, classify the acquired image data based on the acquired keyword, and register the registered keyword. When searching for desired ones of the image data registered as search conditions, the classified image data and the keyword assigned to this image data are displayed, so that the keyword assigned to each registered image data is displayed. A search can be easily performed without being recognized, and a first dictionary for managing semantic classification information and a second dictionary for managing part of speech information And a keyword and an information table for managing the classification of the keyword. When the acquired keyword is not in the information table, semantic classification information and part-of-speech information are acquired for the keyword. Since the classification of the acquired keyword is determined based on, the classification of the image data can be performed reliably.
[0144]
According to the image search device of the second aspect and the image search method of the seventh aspect, when the semantic classification information of the acquired keyword is not managed by the first dictionary, the part of speech information acquired for the keyword Since the keyword classification is acquired based only on the keyword, the acquired keyword can be reliably classified.
[0145]
According to the image search device of the third aspect and the image search method of the eighth aspect, when the acquired keyword is in the information table, the classification managed as the classification of the keyword in the information table is changed to the classification of the keyword. Thus, the obtained keywords can be quickly classified.
[0146]
According to the image search device of the fourth aspect and the image search method of the ninth aspect, when a keyword is displayed at the time of search, the classification managed by the information table is displayed, and for each displayed classification, the classification is performed. Is displayed, the search can be performed more easily.
[0147]
According to the image search device of the fifth aspect and the image search method of the tenth aspect, at the time of search, a keyword displayed as a keyword in a category is added to a keyword as a search condition, and the keyword as a search condition is added. Since the keywords displayed as the keywords in the classification are deleted and the keywords serving as the search conditions are set, a plurality of search conditions can be specified, and the operability of the image search can be improved.
[Brief description of the drawings]
FIG. 1 is a block diagram of an image search device according to an embodiment of the present invention.
FIG. 2 is a diagram showing an image information table stored in a ROM 13 in FIG.
FIG. 3 is a view showing a keyword information table stored in a ROM 13 in FIG. 1;
FIG. 4 is a view showing a thesaurus dictionary stored in a ROM 13 in FIG. 1;
FIG. 5 is a diagram showing a keyword classification information table stored in a ROM 13 in FIG. 1;
FIG. 6 is a diagram showing a synonym dictionary of different notations stored in a ROM 13 in FIG. 1;
FIG. 7 is a diagram showing a lower concept word dictionary stored in a ROM 13 in FIG. 1;
8 is a diagram showing a search screen displayed on the CRT 15 in FIG.
FIG. 9 is a flowchart illustrating an image classification process performed by the image search device 11 in FIG. 1;
FIG. 10 is a flowchart showing a keyword search process executed in step S113 of FIG.
FIG. 11 is a flowchart showing a semantic classification order process executed in step S116 of FIG. 9;
12 is a flowchart showing an image data search process executed by the image search device 11 in FIG.
[Explanation of symbols]
11 Image retrieval device
12 CPU
13 ROM
14 RAM
15 CRT
16 Peripheral controller
17 External storage device
18 Digital Camera
19 Digital Video Camera
20 Input device
61 Classification display section
62 Keyword number display
63 Keyword display
64 Representative image data display section
65 Search condition input section
66 Search execution part
67 Check Box

Claims (12)

An image search apparatus comprising: a registration unit that registers a plurality of image data and a keyword to be assigned to the image data; and a search unit that searches for a desired one of the registered image data using the registered keyword as a search condition. At
The registration means,
Image data acquisition means for acquiring the plurality of image data,
Keyword acquisition means for acquiring a keyword to be assigned to the acquired image data,
Storage means for storing a first dictionary for managing semantic classification information, a second dictionary for managing part of speech information, and a keyword and an information table for managing the classification of the keyword;
Keyword determining means for determining whether or not the acquired keyword is a keyword managed by the information table,
Means for acquiring meaning classification information of the keyword from the first dictionary;
A part-of-speech obtaining means for obtaining part-of-speech information of a keyword from the second dictionary;
When the acquired keyword is not the managed keyword, the keyword is determined based on the semantic classification information acquired for the keyword by the semantic classification acquisition unit and the part of speech information acquired for the keyword by the part of speech acquisition unit. Classification determination means for determining the classification of
The image search apparatus, wherein the search means includes a display means for displaying the classified image data and a keyword assigned to the image data. When the semantic classification information of the obtained keyword is not managed by the first dictionary, the semantic classification obtaining means obtains the classification of the keyword based only on the obtained part of speech information. The image search device according to claim 1, wherein 2. The apparatus according to claim 1, further comprising another classification determining unit configured to determine a classification managed as a classification of the keyword in the information table when the keyword obtained by the keyword determination unit is present. The image search device described in the above. The keyword display means,
Classification display means for displaying a classification managed by the information table,
4. The image retrieval apparatus according to claim 3, further comprising: a keyword-in-class display means for displaying a keyword determined by the classification determination means for each classification displayed by the classification display means. The search means,
A search condition adding unit that adds a keyword displayed by the in-category keyword display unit to the keyword that is the search condition;
Search condition deleting means for deleting a keyword displayed by the keyword-in-class display means from the keyword serving as the search condition;
The image search apparatus according to claim 4, further comprising: a search condition setting unit configured to set a keyword as the search condition. An image search method, comprising: a registration step of registering a plurality of image data and a keyword to be assigned to the image data; and a search step of searching a desired one of the registered image data using the registered keyword as a search condition. At
The registration step includes:
Image data obtaining step of obtaining the plurality of image data,
A keyword obtaining step of obtaining a keyword to be assigned to the obtained image data,
Storing a first dictionary that manages semantic classification information, a second dictionary that manages part-of-speech information, and an information table that manages keywords and classification of the keywords;
A keyword determining step of determining whether or not the acquired keyword is a keyword managed by the information table;
A semantic classification obtaining step of obtaining semantic classification information of the keyword from the first dictionary;
A part-of-speech obtaining step of obtaining part-of-speech information of a keyword from the second dictionary;
When the acquired keyword is not the managed keyword, the keyword is determined based on the semantic classification information acquired for the keyword in the semantic classification acquisition step and the part of speech information acquired for the keyword in the part of speech acquisition step. And a classification determining step of determining the classification of
The image search method, wherein the search step includes a display step of displaying the classified image data and a keyword assigned to the image data. The semantic classification obtaining step is characterized in that when the semantic classification information of the obtained keyword is not managed by the first dictionary, the classification of the keyword is obtained based only on the obtained part of speech information. The image search method according to claim 6, wherein 7. The method according to claim 6, further comprising, when there is the keyword obtained in the keyword determining step, another category determining step of determining a category managed as the category of the keyword in the information table as the category of the keyword. Image search method described. The keyword displaying step includes:
A classification display step of displaying a classification managed by the information table, and a classification-in-class keyword display step of displaying, for each classification displayed in the classification display step, a keyword determined in the classification by the classification determination step. 9. The image search method according to claim 8, comprising: The search step includes:
A search condition adding step of adding the keyword displayed by the keyword-in-class display step to the keyword that is the search condition;
A search condition deleting step of deleting a keyword displayed by the in-category keyword display step from the keyword being the search condition;
10. The image search method according to claim 9, further comprising a search condition setting step of setting a keyword as the search condition. Causing a computer to execute a registration step of registering a plurality of image data and a keyword to be assigned to the image data, and a search step of searching for a desired one of the registered image data using the registered keyword as a search condition. In the program,
The registration step includes:
Image data obtaining step of obtaining the plurality of image data,
A keyword obtaining step of obtaining a keyword to be assigned to the obtained image data,
Storing a first dictionary that manages semantic classification information, a second dictionary that manages part-of-speech information, and an information table that manages keywords and classification of the keywords;
A keyword determining step of determining whether or not the acquired keyword is a keyword managed by the information table;
A semantic classification obtaining step of obtaining semantic classification information of the keyword from the first dictionary;
A part-of-speech obtaining step of obtaining part-of-speech information of a keyword from the second dictionary;
When the acquired keyword is not the managed keyword, the keyword is determined based on the semantic classification information acquired for the keyword in the semantic classification acquisition step and the part of speech information acquired for the keyword in the part of speech acquisition step. And a classification determining step of determining the classification of
The program according to claim 1, wherein the search step includes a display step of displaying the classified image data and a keyword assigned to the image data. Causing a computer to execute a registration step of registering a plurality of image data and a keyword to be assigned to the image data, and a search step of searching for a desired one of the registered image data using the registered keyword as a search condition. In a computer readable medium storing a program,
The registration step includes:
Image data obtaining step of obtaining the plurality of image data,
A keyword obtaining step of obtaining a keyword to be assigned to the obtained image data,
Storing a first dictionary that manages semantic classification information, a second dictionary that manages part-of-speech information, and an information table that manages keywords and classification of the keywords;
A keyword determining step of determining whether or not the acquired keyword is a keyword managed by the information table;
A semantic classification obtaining step of obtaining semantic classification information of the keyword from the first dictionary;
A part-of-speech obtaining step of obtaining part-of-speech information of a keyword from the second dictionary;
When the acquired keyword is not the managed keyword, the keyword is determined based on the semantic classification information acquired for the keyword in the semantic classification acquisition step and the part of speech information acquired for the keyword in the part of speech acquisition step. And a classification determining step of determining the classification of
The medium according to claim 1, wherein the search step includes a display step of displaying the classified image data and a keyword assigned to the image data.

Images