JP2001357067A - Image data retrieving method and computer-readable storage medium - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide an image data retrieving method and a computer-readable storage medium capable of more easily retrieving image data related to an image, in which a human figure is a subject. SOLUTION: Concerning image data in a database, image data corresponding to the image photographing a human figure as a subject and other image data are identified on the basis of teacher data expressing the features of the face of the human figure, for example. Thus, image data can be efficiently retrieved without necessity for a retriever himself/herself to retrieve whether the human figure is photographed as a subject or not.

Description

DETAILED DESCRIPTION OF THE INVENTION

[0001]

The present invention relates to a database for storing a large amount of image data including image data obtained by photographing a person or the like as a subject.
The present invention relates to an image data search technique capable of detecting specific image data.

[0002]

2. Description of the Related Art In events such as cherry-blossom viewing and travel, many shots are taken with a person as a subject, but it is often difficult to organize the shots. In other words, schools, corporations, and other organizations have a large number of people participating in the event, and trying to distribute the photos to each person included in the photos taken requires manual work such as identifying the people and counting the required number of photos. And the identification of the person may be wrong.

[0003]

On the other hand, it is already possible to digitize a visible image indirectly from a photograph or a negative film, or directly through a digital still camera or the like, and store the digitized image in a database as image data. Is being done. As a result, there is an advantage that even a large amount of images can be saved without taking up space and maintaining the initial image quality.

[0004] However, when the image data is stored in the state as described above, the image cannot be confirmed in the state of the image data when the person is identified as described above. There is a problem that a person identification operation is required, and it takes more time.

SUMMARY OF THE INVENTION In view of the above problems, an object of the present invention is to provide an image data search method and a computer readable storage medium that can more easily search for image data of an image in which a person is a subject. I do.

[0006]

According to a first aspect of the present invention, there is provided a method for retrieving image data in which, when a person is included in an image corresponding to image data, additional data including information on the person is created. A database is constructed by storing the image data in association with the image data, and when searching for image data in the database, image data corresponding to an image of a person is automatically generated based on the additional data. Is characterized.

According to a second aspect of the present invention, there is provided an image data search method according to the present invention, based on feature data extracted from image data stored in a database, wherein the image data from which the feature data is extracted corresponds to an image in which a person is shown. And a process of determining whether or not the image data is to be processed.

According to a third image data search method of the present invention, when an image corresponding to image data stored in a database is searched, feature data is extracted from the image data to be searched. By comparing the feature data with the teacher data registered corresponding to the information on the contact information of the specific person, the image data corresponding to the image in which the specific person is captured is identified and identified. Image data and / or an image corresponding to the image data is sent to the contact information of the specific person.

[0009]

According to the first image data search method of the present invention,
If a person appears in the image corresponding to the image data, create additional data including information about the person,
A database is constructed by storing the image data in association with the image data, and when searching for image data in the database, image data corresponding to an image of a person is automatically generated based on the additional data. Therefore, even if a huge amount of image data is stored in the database, it is possible to easily retrieve data corresponding to the image in which the person is captured from the additional data including the information on the person.

Further, it is preferable that the additional data includes information for specifying a person. For example, if the information for specifying the person is text information on the name of a specific person appearing in the image, by inputting the name of the specific person at the time of search, the image of the person If the image data can be searched, or if the information for specifying the person is text information related to the address or e-mail address of a specific person shown in the image, the searched image of the person is shown. The image data can be stored in a storage medium and sent to the address, or the image data can be attached to an e-mail and sent.

[0011] Preferably, the additional data includes information on the number of persons. For example, if the information for specifying the person is text information on the number of persons (for example, three) appearing in the image, when the image is printed and distributed to all persons in the image, ,
It is convenient because only three sheets need to be printed. It should be noted that if the number of persons appearing in the image is zero, it means that no person is present in the image.

Further, if the additional data is created by automatically extracting the information on the person from the corresponding image data, it is not necessary to input such additional data, which is convenient. Note that the additional data is
As a method of automatically extracting from the corresponding image data, after extracting characteristic data to be described later from the image data, information about a person specified from the characteristic data is newly associated with the image data as additional data. It is possible to memorize, but it is not limited to this.

According to a second aspect of the present invention, there is provided an image data search method according to the present invention, based on feature data extracted from image data stored in a database, wherein the extracted image data corresponds to an image of a person. As described above, even if additional data including information about a person is not stored in association with the image data, the feature data is directly extracted from the image data as described above. By extracting, it is possible to determine whether or not a person is included in the image corresponding to the image data. Here, the characteristic data refers to data representing the color or shape of a person in numerical values, and for example, data that can take a unique value (or range) for each gender, age, race, or specific person. But not limited to that.

Further, teacher data is obtained in advance for a specific person, and the discriminating process is performed when the extracted feature data is compared with the teacher data.
It is preferable to include a process of determining that the specific person is included in an image corresponding to the image data from which the feature data has been extracted, when the difference is equal to or smaller than a threshold value. This is because the state of the person in the image is various, and the feature data extracted therefrom is considered to have a certain range, and does not always completely match the teacher data as the reference for specifying the person. Here, the teacher data refers to data representing the color and shape of a person in the form of numerical values in the same manner as the feature data. For example, the teacher data takes a unique value (or range) for each gender, age, race, or specific person. It refers to what can be obtained, but is not limited to it as in the case of the feature data. In addition, it is preferable that the threshold value can be changed, because the accuracy of the determination that the specific person is captured can be improved.

The feature data and the teacher data are
The difference having a plurality of feature amounts and being compared with the threshold value,
Preferably, the difference is obtained by multiplying a difference between the plurality of feature amounts by a predetermined parameter, and the parameter can be further changed. This is because if the result of identifying the image data by comparing the extracted feature data with the teacher data is bad, it is possible to perform a more accurate search by changing the parameter.

Further, it is preferable that the characteristic data include data relating to the probability of being a person. The data relating to the probability of being a person means that, for example, when a numerical value obtained based on a specific subject in an image ranges from 0 to 1, the closer to 0, the more certain that the subject is not a person. The data is such that the closer to 1, the more certain that the person is.

Further, the data relating to the certainty of the person includes the outline, color, brightness, gradient of color / brightness,
When the data includes at least one of the frequency components, based on these data, data regarding the probability of being the person can be calculated as a specific numerical value, and it can be more accurately and efficiently determined whether or not the subject is a person. Can be determined. Each time a search is performed from image data, these data are extracted as feature data, and these data are compared with teacher data from which these data have been acquired in advance.

Further, it is preferable that the teacher data includes information on the specific person. For example, if the information on the specific person is text information on the name of a specific person appearing in the image, the name of the person appearing in the image corresponding to the image data to be searched is obtained from the teacher data. Can also be specified.

Further, it is preferable that the information on the specific person includes information on contact information of the specific person. For example, if the information on the specific person is text information on the address or e-mail address of the specific person shown in the image, the information is shown in the image corresponding to the image data to be searched from the teacher data. It is also possible to specify a person's address or mail address and send the image print to the address, or attach image data to a mail and send it to the mail address.

Further, it is preferable that the teacher data can be registered. Here, registration means that the teacher data is acquired in advance and stored together with the name of a specific person in a predetermined area such as a hard disk of a personal computer.
At the time of searching for image data, quick search can be performed by reading teacher data from the name of a specific person and comparing it with image data extracted from the image data to be searched.

It is preferable that the registration of the teacher data be performed in the background because it can be performed in parallel with other operations.

Further, by counting the number of persons in the image based on the feature data from the image data corresponding to the image in which the person is captured, for example, based on the retrieved image data, When printing an image so that it can be distributed to another person, the required number of prints can be immediately determined, which is convenient.

Further, displaying the counted number of the photographed persons is convenient because the searcher can immediately grasp the total number of prints and can calculate necessary charges and the like.

It is convenient if the displayed number of people can be edited. This is because it is not always necessary to print an image for a person appearing as a background, and it is convenient for a searcher to be able to adjust the number of prints by editing. Here, editing means increasing or decreasing the displayed number of people.

It is preferable that the image is printed by the number of the counted or edited persons.

Further, for each image data determined to be image data corresponding to an image of a person by the above-described determination processing, printing of an image, storage of image data,
If at least one of the transfer of the image data is performed, the trouble of the searcher can be reduced, which is convenient.

According to a third aspect of the present invention, there is provided an image data search method in which teacher data is registered in correspondence with information relating to a specific person's contact information, and characteristic data is extracted from image data stored in a database. By comparing the obtained feature data with the teacher data, it is determined whether or not the image data from which the feature data has been extracted is image data corresponding to an image of the specific person, Image data determined to be image data corresponding to an image of a specific person and / or an image corresponding to the image data,
Since the information is sent to the contact information of the specific person, it is not necessary for the searcher to manually sort the image prints, and the sorting and sending can be performed efficiently.

[0028]

Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram illustrating a data search device according to the present embodiment. The data search device includes a main body 1 having a large-capacity storage device, a display device 2 as a display, an input device 3 as a keyboard,
A printer 4 capable of printing images. This data retrieval device is preferably connected to the Internet and forms part of a print order network, for example. For example, when a storage medium CD storing a program for executing the image data search method is inserted into the reading unit 1a of the main body 1, the program is read out to the memory of the main body 1 and executed. ing. Examples of such a data search device include a personal computer and a display such as a monitor and a printer, or a game machine and a display such as a monitor and a printer.

(First Embodiment) In the main body 1, an enormous amount of image data is stored in a database. The image data includes text data (additional information) such as the name, gender, date of birth, address, telephone number, fax number, e-mail address, and the number of persons appearing in the image corresponding to the image data. Data) are stored in association with each other.

Here, when it is desired to distinguish between image data corresponding to an image of a person and other image data, the main body 1 first searches the database for a search target according to a searcher's instruction. The additional data corresponding to the image data is read out, and if one or more data is stored as the number of persons therein, it is determined that a person is included in the image corresponding to the image data. If data of 0 is stored as the number of persons, it is determined that no person is included in the image corresponding to the image data. This makes it easy to automatically discriminate image data corresponding to an image of a person from other image data. In addition, data representing the type of image (here, a numeral) is associated with image data as additional data, such as “1” for an image of a person in advance and “2” for an image of only a landscape. Alternatively, it may be automatically determined based on the additional data whether or not the image data corresponds to an image of a person.

Further, when it is desired to extract only image data corresponding to an image in which a specific person is shown, the searcher inputs the name of the specific person via the input device 3 so that the main body 1 can be extracted.
First reads, from the database, additional data corresponding to the image data to be searched, and if the name of the person included therein matches the input name, an image corresponding to the image data includes: Judge that the person with that name is shown. As a result, it is possible to easily extract the image data corresponding to the image in which the specific person is captured. After the image data is extracted, if necessary, the image data is attached to the e-mail address of the person in the e-mail and transmitted, or the image printed based on the image data is sent to the address of the person in the e-mail. Or can be sent to

Such additional data can be stored at the same time as the image data by being manually input by the photographer at the time of photographing, or can be newly created while viewing the displayed image based on the stored image data. You can also enter it manually,
In some cases, for example, data of a specific person (preliminarily input) from a cell phone worn by a specific person is wirelessly transmitted to a digital still camera, and attached to image data at the time of photographing as attached data. Is also possible.

However, it can be said that it is difficult to always input the name of the person appearing in the image as additional data for all the image data. Also,
Although it is theoretically possible to create additional data individually for a large amount of existing image data, it is often impossible in practice due to time constraints and the like. On the other hand, if data relating to a person is directly extracted from the image data and analyzed, and the person appearing in the image corresponding to the image data can be specified, a vast amount of image data can be easily searched. . In the second embodiment described below, image data is searched by extracting feature data from image data.

(Second Embodiment) In this embodiment, specifically, the color, brightness, color gradient, brightness gradient, frequency component, A registration means for pre-calculating and registering data relating to the contour as general or specific human teacher data, registering the registered teacher data, and feature data extracted from image data in the database (in terms of components) And a search unit for searching for image data corresponding to an image in which a human face is captured or a specific human face is captured by comparing the same with teacher data). Here, registration means storing the teacher data of the face of a general person, or the teacher data of a specific person, the name of the specific person,
It means that it is stored in association with an address. Hereinafter, registration and search means will be described.

[Registration Means] (1) Registration of Teacher Data Registration of teacher data will be described with reference to the flowchart of FIG. First, a representative face image of a person is created (step S101). For example, based on image data corresponding to an image of a person, a face area including various hairstyles is extracted from the image, and a predetermined size (for example, 256 pixels horizontally, 384 pixels vertically, and 8 bits each for RGB)
Enlarge or reduce to match t).

With respect to the N images created in this way, an average of RGB values is obtained for 256 pixels horizontally and 384 pixels vertically, and this is used as an average face image of a person. The R value of the horizontal i and vertical j pixels of the face image is represented by the following equation, where R (k) i and j are the R values of the pixels at the i and j positions of the k-th image. It is.

(Equation 1)

With respect to the average face image obtained in this manner, feature data composed of a plurality of feature amounts (values representing color, brightness, etc., which will be described later) is obtained here (step S10).
2). Specifically, the face is divided into a plurality of regions, and feature amounts indicating the person's face, such as the color, brightness, and frequency component values of each region, are calculated. A specific calculation example is shown below. The face image is divided into 16 × 4 × 16 areas. The following features are obtained for each area. Color: The average value of R, G, B in the area may be used, or R / (R
+ G + B), G / (R + G + B), or B (R + G + B). Also, the most frequently appearing RGB in the region
May be combined. Brightness: The average value of R, G, and B in the area is defined as brightness.
In order to simplify the processing, the average value of G may be used. Color gradient: The sum of the absolute values of the color differences between the target pixel and the four pixels vertically and horizontally surrounding the target pixel is defined as a color gradient. Brightness gradient: The sum of the absolute values of the brightness differences between the target pixel and the four pixels vertically and horizontally surrounding the target pixel is defined as a brightness gradient. Frequency component value: The brightness in the area is discretely Fourier-transformed, and for example, if 1.0 is the Nyquist frequency f,
The response values of 0.25f, 0.5f, and 0.75f are defined as frequency component values. Face outline: The target area is binarized according to the magnitude of the brightness gradient described above. As a result, edges within the region are extracted. The binarized image data is used as face outline data. The process is completed by registering feature data including these feature amounts as teacher data of an average person's face.

The representative face teacher data need not be one type. Various teacher data can be created for gender, age, race, or the average face of the entire human race. For example, two types of average face of men and average face of women may be used, or 10s, 20s, 30s, etc.
・ You may have an average face for each age. Also,
You can divide it with or without glasses,
If the target is other than Japanese, it may be classified according to the type of person. This data is stored in advance on a recording medium (for example, CD-R
OM, DVD).

(2) Update and New Registration of Teacher Data The teacher data is not an average face, but a face of a specific person can be newly registered as teacher data, and the content can be updated when necessary. . The items of the feature amount stored as the teacher data of the face of the specific person are the same as those obtained with the above-described average face teacher data. For example, since the face of a person changes with age, even if teacher data is registered at a young age, for example, there is a risk that an image of an adult cannot be searched using such teacher data. In contrast, according to the present embodiment, when identifying a person after becoming an adult,
New teacher data can also be created. The procedure is described below,

After the target person A is an adult, a face image for preparing teacher data is prepared by photographing a face of the target person A with a digital camera or extracting the face from an image of the target person A as a subject. The prepared face image is registered as image data for teacher data. The teacher data is calculated and stored by the method described above. This registration work is performed in the background, and the image search can be performed in parallel during the registration work. Specifically, it becomes possible by lowering the priority of the thread (unit of program execution) corresponding to the registration work,

(3) Registration of Data Other Than Teacher Data Registration of data other than teacher data will be described with reference to the flowchart of FIG. Here, by analyzing an image in which a subject is a person, data as a source of analysis when determining how many persons are included is registered. Such an analysis is based on, for example, how many times the skin color region appears in an image in which a plurality of people are captured, since the skin color region appears at a predetermined cycle when the image is scanned in the arrangement direction of the face images. This is based on the idea that the number of people in the picture can be determined. However, in the natural world, there are a large number of objects having colors similar to skin color. Therefore, based on not only the color but also the brightness, the gradient of the color / brightness (the face is three-dimensional, the color / brightness partially changes due to the unique shadow), the frequency and the contour of the face Then, an image that is considered to be close to the face is extracted. Such extraction can be performed using a generally known technique called template matching.

(I) Registration of Data such as Color and Brightness and Matching Degree Here, representative representative data of a representative (eg, average Japanese) face obtained by the method (1) will be used. .
Using such representative teacher data, pattern matching is performed by the following method in order to check whether or not a (Japanese) person is included in the target image. The template matching is performed between the feature data of the target image and the representative teacher data. Here, as an example, the size of the target image is 640 pixels horizontally and 480 pixels vertically, and the size of the template is 12 pixels vertically and 18 pixels horizontally. This size corresponds to the size of the face, and if the size of the face to be extracted changes, this size also needs to be changed. 12 × 1 of the target image corresponding to the position of the template
The area of 8 pixels is divided into 4 × 4, and in each area, feature data including features such as brightness, color, frequency components, and the like are obtained and registered in the same manner as the representative teacher data (step S201). The absolute value of the difference between the obtained feature amount and the feature amount of the representative teacher data is obtained, and the absolute value is added as in the following equation to obtain the difference from the representative teacher data. (Difference from representative teacher data) = A x (brightness difference) + B x
(Color difference) + C × (frequency component value difference) + D × (brightness gradient) + E × (color gradient) + F × (face outline) where A, B, C, D, E, F is an arbitrary constant, and is set so that the rate of correctly identifying an image including a person is highest. This value can be changed by the user. The matching degree is obtained from the difference from the representative teacher data obtained in (Step S202). The matching degree is a value indicating the likelihood of being a person. Here, a value between 0 and 1 is taken, and the closer to 1, it means that the probability of being a face is higher. The following formula shows how to obtain it. Matching degree = 1−maximum difference between teacher data and difference between teacher data The template position may be shifted by one pixel in the raster direction, or may be shifted by a plurality of pixels in order to reduce processing time. It does not matter if the target image is reduced to 1 / n in advance and the processing is performed.
When faces of various sizes are to be extracted, the same processing is performed by changing the size of the template according to the size of the face. Of the matching degrees of each area obtained in this way,
An image having a first threshold value (for example, 0.6) or more is grasped as an image corresponding to an image in which a person is captured, and the matching degree, the coordinate value of the center of the region, and the size of the template are all set in the target image. It is stored as feature data. For example, matching degree = 0.62, abscissa = 128 (pixel), ordinate = 256, template size = 15 × 10 matching degree = 0.7, abscissa = 170 (pixel), ordinate = 200, template size = 30 × 20 Matching degree = 0.9, abscissa = 200 (pixel), ordinate = 100, template size = 30 × 20 Matching degree = 0.94, abscissa = 190 (pixel), ordinate = 240, Template size = 45 × 30 Matching degree = 0.95, abscissa = 10 (pixels), ordinate = 80, template size = 15 × 10 Are stored in addition to the feature amounts such as the gradient, color, color gradient, frequency component, and face contour. Here, the storage of the feature values such as brightness, brightness gradient, color, color gradient, frequency component, face contour, and the like is performed by storing only the feature value of the region exceeding the first threshold obtained in Is preferable because the capacity of the database can be reduced. further,
The number of matching degrees exceeding a default threshold (for example, 0.9) is exceeded, and this is registered as the number of persons as follows. Number of people = 3

(Ii) Storage of Matching Degree of Specific Person A method of calculating and storing a matching degree with characteristic data of a specific person registered as teacher data will be described below. (3) Pattern matching is performed with teacher data of, for example, the person A registered as teacher data, centering on the center position of the area recognized as including the person obtained in (i). At this time, values determined in advance are used for the size of the template matching and the coordinates of the center position of the area. If the matching degree obtained in step (1) is larger than the first threshold value (for example, 0.6), the matching degree obtained in step (2) is stored as the matching degree with the person A (step S203). Example) Matching degree = 0.62, abscissa = 128 (pixels), ordinate = 256, template size = 15 × 10 Matching degree = 0.7, abscissa = 170 (pixels), ordinate = 200, template size = 30 × 20 matching degree = 0.9, abscissa = 200 (pixel), ordinate = 100, template size = 30 × 20 matching degree = 0.94, abscissa = 190 (pixel), ordinate = 240, Template size = 45 × 30 Matching degree = 0.95, abscissa = 10 (pixels), ordinate = 80, template size = 15 × 10 People A: Matching degree = 0.92 B: Matching degree = 0.9 Specific The extraction of a person may be performed on all registered teacher data of a specific person, or a list of registered teacher data of a specific person is displayed, and a user is selected from the list. May be selectable. Thus, the process of storing data other than the teacher data is completed.

In the present embodiment, the above-described matching method is used, but the present invention is not limited to this. For example, "Introduction to Computer Image Processing" published by Soken Shuppan
(Hideyuki Tamura (supervised), edited by Japan Industrial Technology Center) can be used as appropriate.

[Search] Based on the characteristic data obtained for each image data as described above, an image in which a person is photographed as a subject (when the degree of matching of the characteristic data with the teacher data is high) Other images (when the matching degree of the feature data with the teacher data is low) can be distinguished from the search, and the following search can also be performed. Searching for Images with Three Subjects For example, if the user sets the threshold to 0.89 and selects it, the data search device counts the number of image data whose matching degree of feature data is 0.9 or more. I do. The matching degree is 0.62, 0.7, 0.9, 0.94,
When 0.95 is registered, the number of persons is counted as three, and a list of images corresponding to the image data counted as three is displayed. Also, the default threshold 0.9
Is used without changing, the data of the number of persons included in the feature data can be used as it is. In this way, an image in which the number of subjects is three can be quickly searched. It is desirable that such a number can be displayed. If the number of persons included in the feature data is zero, it can be understood that the image does not include a person as a subject, and thus can be distinguished from an image in which a person is captured as a subject. Retrieval of Image Including Specific Person For example, when searching for an image in which person A is captured, an image in which the person A is included in a feature data item of each image at a value equal to or greater than a predetermined threshold value is included. Search for. At this time, it is preferable that the name of the specific person included is displayed together with the image. The user changes parameters and searches for a predetermined image. (3) Coefficients A to F of the formula for obtaining teacher data in (i)
By changing the value of (parameter) by the user, a desired search can be performed. For example, an image in which the number of subjects is three or more can be searched. In this case, the difference is obtained by multiplying the brightness, the brightness gradient, the color, the color gradient, the frequency component, and the coefficient of the face contour data included in the feature data by multiplication by a set coefficient. By comparing with, the number of persons can be counted, and based on the counted number of persons, images of the number of persons designated by the user can be searched. In addition, the number of subjects in a specific image can be counted based on parameters changed by the user as described above. In the present embodiment, this corresponds to the editing of the number of persons, but the editing includes the fact that the user actually checks the image and increases or decreases the number of persons by manual input.

The retrieval of the solid image data will be specifically described with reference to the flowchart of FIG. First, in step S301, the data search device determines whether or not to change the coefficient and the threshold from the default according to a user's instruction.
In step S303, the obtained matching degree is compared with a threshold value in step S303. When it is determined that the coefficient and the threshold should not be changed from the default, the data search device immediately compares the obtained matching degree with the threshold in step S303.

If the degree of matching is higher than the threshold value, the data search apparatus counts up by one in step S304, and if it is low, keeps the count, and performs this for all of the degrees of matching recorded as feature data (step S304). S305). Further, in step S306, the data search device sets the final count value as the number of persons, and in step S307, displays the corresponding image in an image display device such as a monitor in step S308 if the search condition is satisfied,
On the other hand, if it does not match the search condition, the process is terminated without displaying the image in step S309.

As described above, when all the image data are searched by the data search device, an image collection in which a person is captured, an image collection in which no person is captured, and an image collection in which a specific person A is captured (FIG. 2) (A), FIG. 2 (b)), an image collection showing B (FIG. 2 (b)), or an image collection showing A and B (FIG. 2 (b)). Based on the image collection sorted in this way, the number of prints can be printed by the printer 4 for the number of persons counted there, or stored in a specific recording medium as necessary, Can be given to the specific person A. If the contact information of the specific person A is registered in the database, the contact information of the specific person A is searched, and the specific person A's E-ma
The image data can be automatically transferred to the il address (Aa in FIG. 2C), or the image printed at the specific person A's home can be mailed. In addition, by automatically printing the number of copies, the required number of copies can be automatically printed without the user confirming how many people are in the image.

As described above, the present invention has been described with reference to the embodiments. However, the present invention should not be construed as being limited to the above embodiments, and it is needless to say that modifications and improvements can be made as appropriate. is there. For example, when the number of persons displayed by counting by the data search device is different from the operator's intention, the displayed number of persons may be changed. Also, if the number of people counted by the data search device as a whole is larger than the number of people to be counted in the actual image, if the threshold value compared with the matching degree is increased, the data search device counts as a whole. The number of people can be reduced. Conversely, when the total number of persons is smaller than the intended number, the number of persons to be counted can be increased as a whole by lowering the threshold value compared with the matching degree. In addition, a user who handles the data search device may be a general consumer, and a user who is a general consumer performs search and print orders, and constructs and manages a database on the company side via the Internet. Alternatively, a printing operation according to the order may be performed.

[0050]

According to the present invention, it is possible to provide an image data search method and a computer-readable storage medium that can more easily search for image data of an image in which a person is a subject.

[Brief description of the drawings]

FIG. 1 is a diagram illustrating a data search device according to an embodiment;

FIG. 2 is a diagram illustrating an example of an image of a person.

FIG. 3 is a flowchart showing registration of teacher data.

FIG. 4 is a flowchart showing registration other than teacher data.

FIG. 5 is a flowchart illustrating a search of image data.

[Explanation of symbols]

 DESCRIPTION OF SYMBOLS 1 Main body 2 Display device 3 Input device 4 Printer

──────────────────────────────────────────────────続 き Continued on the front page (51) Int.Cl. ⁷ Identification code FI Theme coat ゛ (Reference) H04N 5/76 H04N 5/76 BF term (Reference) 5B050 BA12 BA15 DA04 EA04 EA06 EA08 EA18 EA20 FA05 5B075 ND08 ND23 NK06 NK21 NK39 PQ03 PQ05 PR06 QM08 UU24 5C052 AC08 DD02 DD04

Claims (20)

[Claims] When a person is included in an image corresponding to image data, a database is constructed by creating additional data including information on the person and storing the additional data in association with the image data. When retrieving image data in the database, an image data retrieval method is characterized in that image data corresponding to an image of a person is automatically determined based on the additional data. 2. The image data search method according to claim 1, wherein the information on the person includes information for specifying the person. 3. The image data search method according to claim 1, wherein the information on the person includes information on the number of the person. 4. The image data search according to claim 1, wherein the additional data is created by automatically extracting information on the person from corresponding image data. Method. 5. A process for determining, based on feature data extracted from image data stored in a database, whether or not the image data from which the feature data has been extracted is image data corresponding to an image of a person. An image data search method comprising: 6. The teacher data is acquired in advance for a specific person, and the discriminating processing is such that when the extracted feature data is compared with the teacher data, the difference is equal to or less than a threshold value. 6. The image data search method according to claim 5, further comprising a process of determining that the specific person is included in an image corresponding to the image data from which the feature data has been extracted. 7. The feature data and the teacher data have a plurality of feature amounts, and the difference to be compared with the threshold is obtained by multiplying a difference between the plurality of feature amounts by a predetermined parameter. 7. The image data search method according to claim 6, wherein said parameter can be further changed. 8. The image data search method according to claim 7, wherein the feature data includes data on the probability of being a person. 9. The data relating to the probability of being a person is data including at least one of a contour, a color, brightness, a gradient of color / brightness, and a frequency component of the person. Image data search method described in. 10. The image data search method according to claim 6, wherein the teacher data includes information on the specific person. 11. The image data search method according to claim 10, wherein the information on the specific person includes information on contact information of the specific person. 12. The image data search method according to claim 6, wherein said teacher data can be registered. 13. The image data search method according to claim 12, wherein the registration of the teacher data can be performed in the background. 14. The image processing apparatus according to claim 5, wherein the number of persons in the image is counted based on the feature data from image data corresponding to an image of the person. Image data retrieval method. 15. The image data search method according to claim 14, wherein the number of the counted persons is displayed. 16. The image data search method according to claim 15, wherein the number of the displayed persons can be edited. 17. The image data search method according to claim 14, wherein printing of images is performed by the number of counted or edited persons. 18. For each piece of image data determined to be image data corresponding to an image of a person by the determining process, at least one of image printing, image data storage, and image data transfer. 18. The image data search method according to claim 5, wherein one of the steps is performed. 19. Registering teacher data in correspondence with information on contact information of a specific person, extracting feature data from image data stored in a database, and extracting the extracted feature data with the teacher data By comparing, the image data from which the feature data is extracted,
Determining whether the image data corresponds to an image in which the specific person is captured; and determining whether the image data is image data corresponding to an image in which the specific person is captured and / or corresponding to the image data. Sending an image to be sent to a contact of the specific person. 20. A computer-readable storage medium storing a program for executing the image data search method according to claim 1.