JP2006201933A - Image classifying device and method - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an image classifying device for updating contents of a group of images to the optimal one according to various conditions without changing the designations of the group, and for holding the images belonging to the group as those that are always optimal to a user. <P>SOLUTION: This image classifying device is provided with a database 11 for recording information related with images and the group of the images, an update object acquiring part 14 for acquiring the group of the images to be updated, an update condition acquiring part 15 for acquiring conditions for updating the group of the images and an image classifying part 17 for updating the contents of the group obtained by the update object acquiring part 14 according to conditions obtained by the update condition acquiring part 15. <P>COPYRIGHT: (C)2006,JPO&NCIPI

Description

  The present invention relates to an image classification apparatus and method for classifying images into several groups.

  In order to efficiently manage and view a large amount of images (image group), it is required to classify the images accumulated in a file server or the like into several groups. Conventionally, there is a method of automatically classifying image groups into groups. Here, a name assigned to a group of image groups is referred to as “ROLL”.

  Even if the images are classified into groups, it is desirable to update the images belonging to the set when a new image is added. As a method for updating such a set, for example, there is a method described in Patent Document 1.

In the method described in Patent Document 1, the image group is automatically classified into groups based on the relevance of the feature amount of the image, and the entire image is reclassified after a certain period of time or when a certain number of images are added. ing.
JP 2001-645244 A

However, the above method has the following problems.
An advantage of ROLL is “accessibility by name to an image group”. However, for example, if there is a ROLL named “Yesterday's walk” that handles an image one day before the current date and time, the image that should belong to it should naturally vary depending on the current date and time. Therefore, it is necessary to update the contents according to the current date and time. However, the contents of the ROLL once created are fixed. Therefore, even if a name “walking yesterday” is given, the content does not match the name as the day passes. This causes confusion for the user, and as a result, convenience is lost.

  Similar problems also occur in the following cases. For example, when inputting the ROLL “Yesterday's walk” into the image quality correction software, the user should naturally prioritize the input with the poor image quality. In addition, when inputting the same ROLL into the album creating software, the user often wants to omit the poor image quality as much as possible and put the good image quality in the album preferentially. In this way, apart from “content assumed from the name”, it is conceivable that the content required for the ROLL changes according to conditions such as a request from the user or application. Even in such a case, a function for updating the contents of the ROLL in response to a request from the application is required.

  On the other hand, although the method described in Patent Document 1 updates the group itself, since the reclassification timing is fixed, flexible reclassification according to various conditions is difficult. Moreover, the classification method described in Patent Document 1 does not have an element corresponding to the name of ROLL for intuitively grasping what kind of image group. Therefore, it is hard to say that the convenience for users is high.

  Therefore, the present invention has been made in view of such problems, and an object thereof is to provide an image classification device and an image classification method that enhance user convenience.

  In order to solve the above problems and achieve the above object, an image classification apparatus according to the present invention includes a storage unit that stores an image group composed of a plurality of digital images, a name of the image group, and a purpose of use of the image group. A designation unit for designating a condition determined by the detection unit, a detection unit for detecting the image according to a condition determined according to a purpose of use of the image group designated by the designation unit, and an image detected by the detection unit. And changing means for changing the contents of the image group based on the image group.

  In this way, the contents of the image group can be changed based on the image whose contents are detected by the detecting means.

  Here, the condition designated by the designation means may be at least one of the name of the image group and the image quality of the image.

  According to this configuration, at least one of the name of the image group and the image quality of the image can be specified by the specifying means.

  Here, the designation unit may designate an image quality of the image, and the image classification device may further include an image quality correction unit that corrects an image quality of an image included in the image group changed by the change unit.

  According to this configuration, when the image quality is designated by the designation unit, it is possible to perform image correction of the images included in the image group.

  Here, the designation unit may designate an image quality of the image, and the image classification device may further include an image browsing unit for browsing an image included in the image group changed by the changing unit.

  According to this configuration, when the image quality is designated by the designation means, the images included in the image group can be browsed.

  Here, the image processing apparatus may further include a determination unit that determines whether or not the image group needs to be updated, and the designation unit may perform the designation for the image group that is determined to be updated by the determination unit.

  According to this configuration, there is an effect that the designation unit can perform designation based on the determination result of the determination unit.

  Here, the detection means detects an image that matches the name of the image group designated by the designation means among images that do not belong to the image group, and the change means detects the detected image as the image group. May be added to

  According to this configuration, there is an effect that an image that matches the name of the image group can be added to the image group.

  Here, the detection means includes a dictionary unit that stores a condition corresponding to a keyword, an extraction unit that extracts a keyword from the name of the image group, and a condition corresponding to the keyword extracted by the extraction unit from the dictionary unit. You may provide the reading part to read, and the matching search part which searches the image which satisfy | fills the conditions read by the said reading part.

According to this configuration, there is an effect that it is possible to search for an image to be read and matched based on a keyword extracted from the name of the image group.
Here, the detection means detects an image that does not match at least one of the name of the image group designated by the designation means and the image quality of the image among the images belonging to the image group, and the change means is detected. The image may be excluded from the image group.

  According to this configuration, an image that does not match at least one of the name of the image group or the image quality of the image can be excluded from the image group.

  Here, the detection means includes a dictionary unit that stores a condition corresponding to a keyword, an extraction unit that extracts a keyword from the name of the image group, and a condition corresponding to the keyword extracted by the extraction unit from the dictionary unit. You may provide the read-out part to read, and the inconsistency search part which searches the image which does not satisfy the conditions read by the said read-out part.

  According to this configuration, there is an effect that it is possible to retrieve an image that does not match by reading a condition based on a keyword extracted from the name of the image group.

  Here, the dictionary unit may store a period during which an image is captured as a condition corresponding to a date and time as a keyword.

  According to this configuration, there is an effect that the period during which an image is captured can be stored as a condition corresponding to the date and time as a keyword.

  Here, the dictionary unit may store a person whose image is captured as a condition corresponding to the person as a keyword.

  According to this configuration, there is an effect that a person whose image is captured can be stored as a condition as a condition corresponding to the person as a keyword.

  Here, the dictionary unit may store, as a condition, a place where an image is captured as a condition corresponding to a place as a keyword.

  According to this configuration, there is an effect that a place where an image is captured can be stored as a condition corresponding to a place as a keyword.

  Here, the dictionary unit may store the brightness of the image as a keyword, the contrast of the image as a condition corresponding to the hue of the image, or the condition as a condition.

  According to this configuration, there is an effect that the brightness or darkness of the image or the hue can be stored as a condition as a condition corresponding to the brightness of the image as a keyword or the hue of the image.

  Here, the dictionary unit may store, as a condition, the number of times the image has been viewed as a condition corresponding to a viewing history as a keyword.

  According to this configuration, there is an effect that the number of times of viewing an image can be stored as a condition corresponding to the viewing history as a keyword.

  In this case, the designation means has the image quality of the image as being out of focus, not out of focus, tilted, not tilted, whiteout, no whiteout, blackout, and blackout. Any one of them may be specified.

  According to this configuration, there is an effect that the presence / absence of defocusing, the presence / absence of tilt, the presence / absence of overexposure, and the presence / absence of blackout can be designated as image quality.

  Note that the method, program, and storage medium for performing the image classification have similar means and effects.

  The image classification device and the image classification method according to the present invention can improve user convenience. It is possible to update to an image belonging to an image group that matches the date and time when the image was taken, and to update an image belonging to the image group to the image quality of the designated image.

FIG. 1 is a functional block diagram of an image classification device 10 according to the present invention.
The image classification device 10 is a device that changes an image belonging to a group as necessary for a group of images that have already been classified, that is, ROLL. The database 11, the update determination unit 12, the name analysis unit 13, the update target acquisition unit 14, the update condition acquisition unit 15, the condition analysis unit 16, the image classification unit 17, and the classification registration unit 18 are configured. The

  FIG. 2 schematically shows an outline of ROLL. As shown in FIG. 2, the ROLL is given a name representing its contents, and about several tens of images matching the contents are registered. The user accesses ROLL based on the name.

  The database 11 stores images and feature quantities of the images. The feature amount includes, for example, information on Exif (Exchangeable Image File Format) that records the time, place, focal length, and the like when the image was taken, and information on attributes such as the average color, brightness, frequency component, and subject type of the image. Is included. FIG. 3 is an example schematically showing stored feature values. Various features are associated with the image.

  Further, the database 11 stores information such as ROLL names, images belonging to each ROLL, and attributes possessed by each ROLL. As attributes that each ROLL has, for example, a range of values for each attribute such as an average color, luminance, frequency component, etc. that should belong to the ROLL, a subject that exists in common to the ROLL, an imaging date and time, etc. There are places, etc.

  It should be noted that the image feature amount and the information related to ROLL mentioned here are merely examples, and other types of information can be actually stored.

  The update determination unit 12 determines a ROLL that needs to change an image to which the image belongs, among the ROLLs recorded in the database 11, and acquires a condition for the update. For example, a name that changes depending on the current date and time, such as “walking yesterday”, or a name that changes by the user, such as “going out with friends”, is given as a case where updating is necessary. In some cases, the current date and user will need to be updated if they change.

  As an update determination method, a name processing unit 13 (to be described later) acquires a corresponding condition from the name of the ROLL and checks the condition, or a method for updating in advance when creating the ROLL in the database. There is a method of registering and referring to it.

  The number of ROLLs determined to be updated by the update determination unit 12 may be one or plural. When there are a plurality of ROLLs that need to be updated, the subsequent processing is repeated for each ROLL.

  The name analysis unit 13 determines conditions necessary for the update from the name of the ROLL. As a determination method, specific keywords such as “Yesterday” and “I” are extracted from the name of ROLL, and based on those keywords, “the date is one day before the current date” or “a specific person is included in the subject. The condition is judged with reference to the dictionary 20 for associating with a condition such as “ Note that the determination method is not limited to the above example, and the condition can also be obtained by using a general natural language processing algorithm, such as determining the condition by analyzing the meaning of the name.

  The update target acquisition unit 14 instructs the image classification unit 17 to use ROLL that needs to be updated. Specifically, it is possible to take a method such as designating a ROLL to be updated from a list of ROLLs displayed on the screen through a pointing device or the like. The ROLL list is created by referring to the name of the ROLL, the attributes representing the properties of the images belonging to the ROLL, and the like. Note that the operations described here are examples, and actual operations are not limited to the above examples.

  The update condition acquisition unit 15 instructs the image classification unit 17 on conditions necessary for the ROLL update. Specifically, as with the update target acquisition unit 14, a method in which a user specifies a condition via a pointing device or the like can be used. For example, when a ROLL for “walking yesterday” is input to the image quality correction software, the user designates an element to be corrected such as defocusing or tilting as a condition. In addition, the update condition may be instructed by referring to the viewing history of the image or the location where the image was taken. In addition, any combination of user information (user name, user birthday, information about the user's family, information about the user's friends, information about the user's thoughts) ) And image attributes (image shooting date / time, image shooting location, average color of image, image subject type, image impression), etc. Good. In addition, the operation | movement quoted here about the instruction | indication method of conditions is an example, and actual operation | movement is not limited to said example.

  When instructing a condition, the user may select a desired condition from a list of conditions given in advance by the system, or may input a desired condition from a keyboard or the like.

  The condition analysis unit 16 converts the condition acquired from the update condition acquisition unit 15 into information in a format that can be used by the image classification unit 17. For example, when the condition “blue” is input from the update condition acquisition unit 15, this is converted into information about the type of image feature quantity and the range of the value, such as “range of hue angle of 240 ° ± 15 °”. To do. As a conversion method, there is a method of preparing a dictionary 21 in which a condition input in advance and a conversion result are associated with each other and referencing it.

  Further, the format of information to be converted and the conversion method described here are merely examples, and the actual operation is not limited to the above example.

  The image classification unit 17 changes an image belonging to the ROLL and updates the ROLL by selecting an image group that matches the condition from the database 11 based on the name of the ROLL and information sent from the condition analysis unit 16. . As for a specific configuration of the image classification unit 17, for example, an image that does not meet the conditions may be taken out from the ROLL, and an image that meets the conditions may belong to the ROLL.

  The classification registration unit 18 registers the ROLL information updated by the image classification unit 17 in the database 11 (not shown).

  Instead of determining the update condition, the update determination unit 12 stores a flag for determining whether or not an update is necessary in advance in the database 11 and refers to it to determine the ROLL that needs to be updated. Also good.

  In addition, when the update target acquisition unit 14 designates a ROLL that needs to be updated directly, the update determination unit 12 may not determine the ROLL that needs to be updated.

  Moreover, the update determination part 12 does not need to acquire the conditions for update, when the conditions for direct update are designated by the update condition acquisition part 15. FIG.

  The update target acquisition unit 14 may receive the output result of the update determination unit 12 and send it to the image classification unit 17. In this case, the update target acquisition unit 14 may be omitted.

  Further, the update condition acquisition unit 15 may receive the output result of the update determination unit 12 and send it to the image classification unit 17. In this case, the update condition acquisition unit 15 may be omitted.

  In addition, the condition analysis unit 16 can omit this when the condition acquired from the update condition acquisition unit 15 is in a format that can be used by the image classification unit 17 as it is.

  The processing operation of the image classification device 10 will be described with reference to the flowcharts of FIGS.

  FIG. 4 is a flowchart showing a flow of processing in which the image classification device 10 automatically determines the rolls that need to be updated without going through the update target acquisition unit 14 and the update condition acquisition unit 15 and updates those rolls. is there.

  First, in step S <b> 101, the image classification device 10 determines a ROLL that needs to be updated from the database 11 via the update determination unit 12. In the following, the flow of processing will be described by taking an example of ROLL “Yesterday's walk”. Note that the ROLL names and processing methods given here are merely examples, and are not limited to the above examples.

  First, the update determination method in step S101 will be described. FIG. 5 is a detailed flowchart of step S101.

  In step S201, the name analysis unit 13 determines a condition necessary for updating the ROLL from the name of the ROLL. As a method of name analysis, a method of extracting a specific keyword and acquiring a condition corresponding to the keyword is conceivable. As a method for acquiring the corresponding condition, for example, a method of preparing in advance a dictionary 20 that associates a keyword representing a specific condition and its range as shown in FIG. In this case, the name analyzing unit 13 first decomposes the name “walking yesterday” into keywords “yesterday” and “walking”. As a keyword acquisition method, a technique such as morphological analysis generally performed in natural language processing can be used.

  Next, the name analysis unit 13 acquires a condition corresponding to each keyword by referring to the dictionary. In the above example, it can be acquired that “the shooting date is one day before the current date”. The name analysis method is not limited to the above example, and any method used in natural language processing, such as determining a condition from the meaning of the name, can be used.

  It should be noted that this step can be omitted when attributes necessary for updating the ROLL are recorded in the database 11 in advance. In this case, in step S202, the recorded condition is read with reference to the database 11, and the determination is performed according to the read condition.

  In step S202, it is determined whether or not the ROLL needs to be updated based on the conditions acquired in step S201 for each image in the ROLL. In the case of “walking yesterday”, if there is an image belonging to ROLL that does not have a “shooting date one day before the current date”, it is determined that updating is necessary, and this ROLL is transferred to subsequent processing. Is done.

  If a flag indicating the necessity of ROLL update is stored in the database 11 in advance, steps S201 and S202 can be omitted. In this case, the necessity of updating can be determined by checking the flag.

As described above, the ROLL that needs to be updated is acquired.
In step S102, the update determination unit 12 acquires an update condition for the ROLL determined to be updated. As in the case of step S201, the update condition acquisition method includes a method of acquiring a specific keyword from a name and referring to a dictionary to acquire a corresponding condition. It should be noted that this step can be omitted if the condition has already been determined in step S101. In this case, the condition obtained in step S101 is used as it is for the subsequent processing. Furthermore, when the update condition is stored in the database 11 in advance in step S102, it may be referred to.

  In step S103, the condition analysis unit 16 analyzes the condition acquired in step S102, and converts the information into information that can be used by the image classification unit 17. For example, if the condition is “the image shooting date is one day before the current date”, the condition is converted to the condition “Exif DateTime tag value is one day before the current date”, which is a feature amount indicating the image shooting date and time. To do. For the conversion of the conditions, a method such as referring to the dictionary 21 that associates the conditions as in FIG. 5 can be used. Note that the format of conversion is not limited to the above example, and in practice, conversion of a necessary format can be performed according to the configuration of the image classification unit 17. If the format of the condition acquired in step S102 is one that can be handled by the image classification unit 17, this step can be omitted. For example, in the above example, the image classification unit 17 has a function of converting the condition indicated by the language information such as “the date of shooting of the image is one day before the current date” into information on the feature amount and its value. In this case, this step can be omitted.

  In step S104, the image classification unit 17 updates the ROLL based on the name of the ROLL acquired in step S101 and the condition acquired in step S103. FIG. 6 schematically shows the state of image classification in the above example. Here, classification is performed on two axes, the date when the image was taken and the time within the day. FIG. 7A shows a stage before update. “Yesterday” at this point is the range of the arrows in the figure. FIG. 7B shows a state when one day has elapsed from the time of FIG. As time passes, the actual date represented by “Yesterday” changes, and the image determined as “Yesterday” changes.

  Finally, in step S <b> 105, the classification registration unit 18 registers the updated ROLL in the database 11.

  FIG. 8 shows an example of a specific update. Assume that the current date is December 12, 2004. On December 11, 2004, an image captured on December 10, 2004 is selected from all the images based on Exif information and belongs to the ROLL “Yesterday's walk”. However, in the case where the current date is December 12, 2004 as described above, the contents are inconsistent as it is. Therefore, updating is performed. An image picked up again on December 11, 2004 from all the images is selected based on the Exif information and belongs to the ROLL “Yesterday's walk”. By performing the update in this way, it is possible to achieve consistency with “walking yesterday”.

  As a result of the above operation, when the ROLL “Yesterday's walk” is referred to, the image belonging to it is automatically replaced with the latest one as necessary, and the user can always replace the image by himself / herself. Appropriate images can be seen.

  Similarly, ROLLs such as “My memories” and “Recently viewed images” can be automatically replaced with appropriate images according to the user name and image viewing history, respectively.

  Up to this point, the case where the image classification device 10 itself determines the ROLL that needs to be updated has been described. However, the ROLL and the condition that the user or the system directly updates using the update target acquisition unit 14 and the update condition acquisition unit 15. Can also be specified. For example, consider the case where ROLL “summer trip in Okinawa” is input to image quality correction software and the case where it is input to album creation software. Hereinafter, the operation in such a case will be described.

  FIG. 9 is a flowchart showing a flow of processing in which a user selects ROLLs that require updating and updates those ROLLs.

  In step S <b> 301, the user designates a ROLL to be updated by the update target acquisition unit 14. Here, “summer travel in Okinawa” is selected.

  Next, in step S <b> 302, the user inputs a condition for updating by the update condition acquisition unit 15. It is assumed that data exchange with other applications is performed according to a predetermined format. For example, information can be transferred using tag data as shown in FIG. The Condition tag in FIG. 10 represents the update condition, the attribute tag in the condition tag represents the type of condition, and the value tag represents the value or state of the condition.

  On the application side, conditions such as “out-of-focus”, “tilt”, “out-of-white”, “under-exposure” specified by the user are converted into the format of FIG. Accepts the converted data as input. Note that the format of data exchange with the application is not limited to the example of FIG. 10, and an arbitrary format can be set. Further, the conditions input by the user are not limited to the above example, and in practice any conditions can be set. Further, when inputting conditions, the user may select a corresponding condition from given conditions, or may input an arbitrary condition.

  An external application or the like may automatically specify conditions. For example, the image quality correction software may automatically issue a condition such as “out of focus” or “tilt” and input it to the update condition acquisition unit 15.

  Steps S303 to S305 are the same as steps S103 to S105. That is, the condition input in step S302 is converted into information that can be used by the image classification unit 17, and the image classification unit 17 updates the contents of the ROLL by using the information, and registers the result in the database 11.

  FIG. 11 is a diagram schematically showing how ROLL of “summer trip in Okinawa” is classified when input into image quality correction software and when input into album creation software. Here, the images are classified by two axes, the date and time when the image was taken and the size of the blur. FIG. 11A shows the case of inputting to image quality correction software. In this case, an image having a certain degree of blur on the classification axis is selected. On the other hand, FIG. 11B shows the case of inputting to album creation software. In this case, an image having a defocus degree of a certain value or less on the classification axis is selected. In addition, for images whose shooting times are close to each other, an image with a smaller degree of blur is selected. Note that the ROLL names and conditions listed here are merely examples.

  FIG. 12 schematically shows the result of the above classification. By performing the processing as described above, when the user handles the same “summer trip in Okinawa” ROLL with each software, the images are automatically switched and handled without selecting individual images.

  As described above, according to the present embodiment, when an update condition is input or automatically determined for ROLL, images belonging to ROLL are replaced, and ROLL is automatically updated. This makes it easy to organize a large number of image files.

  Further, when the ROLL is updated, the ROLL name is not changed, and the user is not confused.

  Note that the ROLL update by the ROLL name and the ROLL update by the image quality may be designated at the same time.

  The image classification apparatus and the image classification method of the present invention are useful as an apparatus and method for managing a large amount of image files in a form that is easy for a user to use, and are suitable for an image storage apparatus or the like.

It is a functional block diagram of the image classification device concerning an embodiment of the invention. It is the figure which represented the concept of ROLL typically. It is an example which showed typically the feature-value stored. It is a flowchart which shows the flow of a process of the image classification apparatus which concerns on this Embodiment. It is a flowchart which shows the flow of the process which determines ROLL which needs an update. It is a figure which shows the example of a response | compatibility with the keyword in a ROLL name, and an update condition. It is the figure which represented typically a mode that the content of ROLL was updated according to the present date. It is a specific example of the update of ROLL. It is a flowchart which shows the flow of a process in case a user or a system designates ROLL which is the object of direct update. It is the figure which represented typically the format for exchanging the information regarding ROLL with other applications. It is a figure showing typically how an image is classified based on a user's condition. It is the figure which represented typically the example from which the image which belongs to ROLL changes according to the request | requirement from an application.

Explanation of symbols

DESCRIPTION OF SYMBOLS 10 Image classification apparatus 11 Database 12 Update determination part 13 Name analysis part 14 Update object acquisition part 15 Update condition acquisition part 16 Condition analysis part 17 Image classification part 18 Classification registration part 20, 21 Dictionary

Claims (18)

Storage means for storing an image group composed of a plurality of digital images and a name of the image group;
A designation means for designating conditions determined according to the purpose of use of the image group;
Detecting means for detecting the image according to a condition determined according to a purpose of use of the image group designated by the designation means among the images;
An image classification apparatus comprising: a changing unit that changes contents of the image group based on an image detected by the detecting unit.   2. The image classification apparatus according to claim 1, wherein the condition designated by the designation means is at least one of a name of the image group and an image quality of the image. The designation means designates the image quality of the image,
3. The image classification apparatus according to claim 2, further comprising an image quality correction unit that corrects an image quality of an image included in the image group changed by the change unit. The designation means designates the image quality of the image,
3. The image classification apparatus according to claim 2, further comprising an image browsing unit for browsing an image included in the image group changed by the changing unit. And further comprising a judging means for judging whether or not the image group needs to be updated.
3. The image classification apparatus according to claim 2, wherein the designation unit performs the designation for the image group determined to be updated by the determination unit. The detection means detects an image that matches the name of the image group designated by the designation means among images not belonging to the image group,
The image classification device according to claim 2, wherein the changing unit adds the detected image to the image group. The detection means includes
A dictionary section for storing conditions corresponding to keywords;
An extraction unit for extracting a keyword from the name of the image group;
A reading unit for reading out a condition corresponding to the keyword extracted in the extraction unit from the dictionary unit;
The image classification apparatus according to claim 6, further comprising: a matching search unit that searches for an image that satisfies a condition read by the reading unit. The detection means detects an image that does not match at least one of the name of the image group designated by the designation means and the image quality of the image among the images belonging to the image group;
6. The image classification apparatus according to claim 2, wherein the changing unit excludes the detected image from the image group. The detection means includes
A dictionary section for storing conditions corresponding to keywords;
An extraction unit for extracting a keyword from the name of the image group;
A reading unit for reading out a condition corresponding to the keyword extracted in the extraction unit from the dictionary unit;
The image classification apparatus according to claim 8, further comprising: a mismatch search unit that searches for an image that does not satisfy the condition read by the read unit.   The image classification device according to claim 7 or 9, wherein the dictionary unit stores, as a condition, a period during which an image is captured as a condition corresponding to a date and time as a keyword.   The image classification device according to claim 7 or 9, wherein the dictionary unit stores, as a condition, a person whose image has been captured as a condition corresponding to a person as a keyword.   The image classification device according to claim 7 or 9, wherein the dictionary unit stores, as a condition, a place where an image is captured as a condition corresponding to a place as a keyword.   The image classification device according to claim 7 or 9, wherein the dictionary unit stores the brightness of the image as a keyword, the contrast of the image as a condition corresponding to the hue of the image, or the hue as a condition.   The image classification device according to claim 7 or 9, wherein the dictionary unit stores, as a condition, the number of times an image has been viewed as a condition corresponding to a viewing history as a keyword.   The designation means has an image quality of the image that is out of focus, no out of focus, tilted, no tilt, whiteout, no whiteout, blackout, and no blackout The image classification apparatus according to claim 2, wherein any one is designated. An image classification method in an image classification device that stores an image group composed of a plurality of digital images and a name of the image group, wherein the designation step specifies at least one of the name of the image group and the image quality of the image;
A detecting step of detecting the image according to at least one of a name of the image group designated by the designation means and an image quality of the image among the images;
An image classification method comprising: a changing step of changing the contents of the image group based on the image detected by the detecting means. A program in an image classification device for storing an image group composed of a plurality of digital images and a name of the image group,
A designation step for designating at least one of the name of the image group and the image quality of the image;
A detection step of detecting the image according to at least one of a name of the image group designated by the designation means and an image quality of the image among the images;
A program for causing a computer to execute a changing step of changing the contents of the image group based on an image detected by the detecting means.   A computer-readable recording medium on which the program according to claim 17 is written.

Images