JP2003271934A - System and method for recognizing face image - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To easily and objectively recognize which well-known person a user looks most like among his/her favorite category. <P>SOLUTION: The well-known person having the face image most similar to the face image taken by the user in the category designated by the user is determined, and the face of the well-known person is provided to the user. <P>COPYRIGHT: (C)2003,JPO

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a face image recognition system for determining the similarity between a face image sent from a user and a face image registered in advance by using a face recognition technique, and this face image recognition system. The present invention relates to a face image recognition method used in.

[0002]

2. Description of the Related Art Conventionally, in ordinary daily conversation, it is often discussed about who is like a person such as a celebrity (actor, athlete, singer, politician, etc.). For example, phone,
When explaining what a person looks like in a letter or the like, it is often the case that a celebrity is quoted and the person's appearance is explained according to which celebrity he or she looks like. In this way, when explaining which celebrity oneself or a third person resembles, based on the judgment of a person such as the person or a friend,
Explains who is similar to what celebrity.

However, when judging who looks like a celebrity with a human sense, there is no basis and reliability is low. Therefore, there is a demand for a service that objectively determines what kind of celebrity the person looks like. Also,
If the most similar celebrity does not meet his taste, he may upset him. For this reason, there is a service that determines the celebrity who most resembles the celebrity of a particular genre according to the wishes of the person or how similar the celebrity the person thinks is similar. Is requested.

[0004]

As described above, since it is judged by a human sense which person is similar to which person, the reliability is low. Therefore, the face is similar to the person's face. It is desired to be able to objectively recognize whether the face is similar to that of the person specified by the user. An object of the present invention is to provide a face image recognition system and a face image recognition method capable of performing

[0005]

A face image recognition system according to the present invention includes a storage unit in which feature amounts of human faces classified into a plurality of groups are stored in advance, and an image capturing unit for capturing a face image of a user. Means, designating means for designating a group of persons whose similarity is determined with the face image photographed by the photographing means, face image photographed by the photographing means, and the designation stored in the storage means. Calculating means for calculating the degree of similarity with the face feature amounts of all persons belonging to the group designated by the means, and the face image photographed by the photographing means based on the degree of similarity calculated by the calculating means. It has a judging means for judging a person with a similar face and a guiding means for guiding the person judged by this judging means.

The face image recognition system of the present invention has a user terminal that can be used by a user and a server device that can communicate with the user terminal. The user terminal captures a face image of the user. Image capturing means, a designating means for designating a group of persons whose similarity is to be determined with the face image captured by the image capturing means, and the image capturing means for capturing the image among the persons belonging to the group designated by the designating means. And a guide means for guiding a person whose face is determined to be most similar to the face image by the server device. The server device stores in advance face feature amounts of persons classified into a plurality of groups. Belonging to the group designated by the designated storage unit, the face image photographed by the photographing unit of the user terminal, and the designation unit of the user terminal stored in the storage unit. Based on the similarity calculated by the calculating means and the calculating means for calculating the similarity with the facial feature amount of all the people, the person having the face most similar to the face image photographed by the photographing means is determined. And a judging means for judging.

The face image recognition method of the present invention is a method used in a face image recognition system having a storage means in which feature amounts of human faces classified into a plurality of groups are stored in advance. A photographing step of photographing the object, a designation step of designating a group of persons whose similarity is determined with the face image photographed by this photographing step, a face image photographed by the photographing step and stored in the storage means. The calculation step of calculating the degree of similarity with the feature amounts of the faces of all the persons belonging to the group specified by the specifying step, and the photographing step by the photographing step based on the similarity calculated by the calculating step. It has a determination step of determining a person whose face is most similar to the face image, and a guidance step of guiding the person determined by this determination step.

[0008]

BEST MODE FOR CARRYING OUT THE INVENTION Embodiments of the present invention will be described below with reference to the drawings. FIG. 1 is a diagram schematically showing the configuration of the entire face image recognition system. As shown in FIG. 1, the face image recognition system includes a mobile terminal (user terminal).
1, a personal computer (user terminal) 2, a telephone station 3, and a face image recognition center server (server device) 4. The mobile terminal 1 and the personal computer 2 can communicate with the face image recognition center server 4 via the telephone station 3. The face image recognition center server 4 has a collation dictionary 4a as a database in which the feature amount of the face of a famous person or the like is stored.

The mobile terminal 1 is composed of a portable terminal device such as a mobile phone. The mobile terminal 1 is entirely controlled by a control unit (not shown) provided in the mobile terminal body 10. In addition, as shown in FIG. 1, the mobile terminal 1 includes a mobile terminal main body 10 provided with a display unit 11, an operation unit 12, a camera 13, an antenna 14, and the like. The display unit 11 is composed of a liquid crystal display device or the like. The display unit 11 displays operation guidance for the user. The operation unit 12 is composed of keys such as a numeric keypad. In the operation unit 12, the user inputs information such as the operation mode setting and the telephone number of the transmission destination. The camera 13 is composed of a digital camera or the like. The camera 13 captures a face image of a user or the like. The antenna 14 transmits / receives wireless data to / from the telephone station 3 when communicating with the face image recognition center server via the telephone station 3.

The personal computer 2 has a main body 2
Within 0, a memory such as RAM, ROM, HDD, various processing units, and a control unit including a CPU and the like are provided. The personal computer 2 is entirely controlled by a control unit (not shown) provided in the main body 20. As shown in FIG. 1, the personal computer 2 has a main body 20 to which a display unit 21, an operation unit 22, a camera 23, a modem 24, and the like are connected. The display unit 21 is composed of a cathode ray tube display device, a liquid crystal display device, or the like. In this display unit 21,
Operation guides for users are displayed. The operation unit 22 is composed of a keyboard, a mouse, or the like. The operation unit 22 is used by the user to input operation mode settings and the like. The camera 23 is composed of a digital camera or the like. The camera 23 captures a face image of a user or the like. The modem 24 communicates with the face image recognition center server 4 via the public line and the telephone station 3.

Next, a schematic configuration of the face image recognition center server 4 will be described. FIG. 2 is a block diagram showing a schematic configuration of the face image recognition center server 4. As shown in FIG. 2, the face image recognition center server 4 has a main control unit 31, a storage unit 32, a communication interface 33, a similarity calculation unit 34, and a collation dictionary 4a. The main controller 31 controls the entire face image recognition center server 4. The storage unit 32 includes a memory for storing control programs and the like, a memory for temporarily storing work data, and the like. The communication interface 33 is an interface for communicating with the mobile terminal 1 or the personal computer 2 via the telephone station 3. The similarity calculation unit 34 calculates the similarity between the face image supplied from the mobile terminal 1 or the personal computer 2 and the face image whose face feature amount is registered in the matching dictionary in advance. It is a thing. The collation dictionary 4a is a storage unit in which facial feature amounts for a plurality of persons such as celebrities are registered in advance. Further, in the collation dictionary 4a, facial feature amounts are classified and stored for each genre (group) of the celebrity.

FIG. 3 is a diagram showing an example of the structure of the collation dictionary 4a. In the example shown in FIG. 3, a plurality of dictionaries created for each genre are stored in the collation dictionary 4a.
For example, in FIG. 3, a dictionary storing face data of teens singers, a dictionary storing face data of actresses in teens, a dictionary storing face data of beautiful women in the world, and face data of baseball players in teens. A stored dictionary, a dictionary storing face data of soccer players in their teens, and the like are stored in the matching dictionary 4a. In this way, if you create a dictionary for each genre,
At the time of matching face data, a dictionary corresponding to the genre selected by the user is selected and matching is performed with the face image taken by the user.

Further, in the example shown in FIG. 3, the case where a dictionary is created for each genre has been described, but at the time of matching face data, the genre selected by the user is used as a search condition to search the matching dictionary 4a. However, a famous person belonging to the genre may be selected each time. In this case, the collation dictionary 4a stores various data (for example, name, sex, age, place of origin, field of activity, group (team) name, etc.) of each celebrity corresponding to the face data of each celebrity.
Is stored, and all the celebrities who match the conditions are selected based on various conditions specified by the user when the similarity is determined. As a result, it is not necessary to create a plurality of dictionaries for each genre, and the user can specify the genre by combining various conditions.

Next, the operation of the face image recognition system configured as described above will be described. FIG. 4 shows the mobile terminal 1.
Alternatively, the degree of similarity between the face photographed by the personal computer 2 and the face of a celebrity whose face feature amount is stored in the matching dictionary 4a is determined, and the celebrity with the highest degree of similarity is stored in the mobile terminal 1 or the personal computer 2. It is a flow chart for explaining the operation of notifying. In Figure 4,
The operation when the user specifies the genre of the celebrity to be verified after the face is photographed is shown. Although the case of using the mobile terminal 1 will be described here, the same applies to the case of using the personal computer 2.

First, the user photographs a face image with the camera 13 (step S1). The captured image is captured in the mobile terminal body. The control unit of the mobile terminal 1 performs a process of extracting a face image from the captured image (face image extraction process) (steps S2 to S4). Here, it is assumed that the camera 13 continuously captures images.

In this case, the control unit of the portable terminal 1 detects the moving body from the continuously photographed images (step S).
2). The moving body is detected by obtaining the difference between the image read previously and the image read next, that is, the difference between the images before and after. When the moving body is detected, the control unit of the mobile terminal 1 determines whether the detected moving body has a predetermined size (step S3). When the moving body detected by this determination has a predetermined size, the control unit of the mobile terminal 1 extracts a face image by detecting a portion of the moving body that looks like a human face (step S4). If the detected moving body does not have a predetermined size, or if the face image cannot be extracted, the control unit of the mobile terminal 1 returns to step S1 and repeats the face image extraction process. Run.

When the face image is extracted by the above extraction processing, the control unit of the mobile terminal 1 selects the genre to which the celebrity whose similarity is to be determined belongs (step S5). The selection of this genre is determined by the operation of the operation unit 12 by the user based on the selection screen displayed on the display unit 11.

For example, when the user designates a professional baseball player as a genre, the user may be able to select a baseball team he / she likes among the professional baseball players. As a result, among professional baseball players, the result that the user is similar to the player of the baseball team that the user dislikes is not provided to the user, and the celebrity who is similar to the user's request is not provided. It becomes possible to notify.

The user who selects the genre as described above confirms the extracted face image and the selected genre, and issues a transmission request. As a result, the control unit of the mobile terminal 1 transmits the selected genre and the extracted face image to the face image recognition center server 4 (step S6).

The main control unit 31 of the face image recognition center server 4
When receiving the information indicating the genre and the face image from the mobile terminal 1 (step S7), extracts the feature amount (face data) of the face from the received face image (step S8).
Further, the main control unit 31 uses the collation dictionary 4a for the face data of the celebrity belonging to the genre received from the mobile terminal 1.
(Step S9). Here, when the dictionary corresponding to the genre selected by the user is provided in the matching dictionary 4a, the main control unit 31 selects the dictionary corresponding to the genre selected by the user as the dictionary to be matched. To do.

When face data is extracted from the face image received from the portable terminal 1 and face data of a received celebrity of the genre is selected from the collation dictionary 4a, the main control unit 31 receives the received face image. A process of calculating the similarity between the face data extracted from the face data and the face data of all the celebrities selected from the matching dictionary 4a (similarity calculation process) is performed (step S10).

When the similarity for all celebrities of the genre is calculated by this similarity calculation processing, the main control unit 3
1 determines the celebrity with the highest similarity (step S11). By this determination, a celebrity whose face is most similar to the face image taken by the mobile terminal 1 is determined in the genre. This determination result is transmitted to the mobile terminal 1 (step S12).

When this judgment result is received (step S1)
3), the control unit of the mobile terminal 1 displays the received information indicating the famous person on the display unit 11 as the famous person whose face most resembles the captured face image (step S14). This allows the user to recognize who is the most similar celebrity in the selected genre.

Further, in the operation example shown in FIG. 4, the face image is photographed and then the genre is selected. However, after the genre is selected, the face image is photographed and collation is performed. You can Further, when the determination result is displayed in step S14, the user may continuously select another genre to determine the most similar celebrity in the other genre. In this case, the similarity determination may be performed using the captured face image, or the face image may be newly taken and the similarity determination may be performed.

Further, in the operation example shown in FIG. 4, the user is provided with a judgment result as to whether or not the face image photographed by the user is the most similar to the face of a famous person in the genre designated by the user. How much the face image taken by the user resembles a particular celebrity may be provided to the user as a determination result. In this case, when selecting a genre, the user specifies a particular celebrity, not the genre. On the other hand, the face image recognition center server 4 calculates the similarity between the face image captured by the user and the feature amount of the face of the celebrity designated by the user, and the calculated similarity is expressed as data such as percentage. Make it available to users. This allows the user to objectively recognize how similar the captured face image is to a particular celebrity.

As described above, in the present embodiment, the celebrity of the face most similar to the face image photographed by the user in the genre designated by the user is determined and provided to the user. It is a thing. This allows the user to easily and objectively recognize which celebrity is most similar to his / her favorite genre. Also, since the user does not let the user specify the genre, the user will not be provided with the judgment result that the user is making a celebrity in the genre that the user does not want. You can offer a celebrity who is.

[0027]

As described above in detail, according to the present invention, it is possible to objectively recognize what kind of person the photographed face looks like among the persons in the genre designated by the user. It is possible to provide a face image recognition system and a face image recognition method.

[Brief description of drawings]

FIG. 1 is a diagram showing a schematic configuration of a face image recognition system according to an embodiment of the present invention.

FIG. 2 is a block diagram showing a schematic configuration of a face image recognition center server shown in FIG.

FIG. 3 is a diagram showing a configuration example of a collation dictionary in a face image recognition center server.

FIG. 4 is a flowchart for explaining the operation of this face image recognition system.

[Explanation of symbols]

1 ... Mobile terminal (user terminal), 2 ... Personal computer (user terminal), 3 ... Telephone station, 4 ... Face image recognition center server (server device), 4a ... Collation dictionary (storage means), 11, 21 ... Display Department (guide means), 12, 22 ...
Operation unit (designating unit), 13, 23 ... Camera (imaging unit), 31 ... Main control unit (determination unit, selection unit), 32 ...
Storage unit, 33 ... Communication interface, 34 ... Similarity calculation unit (calculation means)

   ─────────────────────────────────────────────────── ─── Continued front page    (72) Inventor Jun Oka             1-1 Shibaura, Minato-ku, Tokyo Co., Ltd.             Toshiba headquarters office (72) Inventor Nobuyuki Tada             1-1 Shibaura, Minato-ku, Tokyo Co., Ltd.             Toshiba headquarters office F term (reference) 5B057 DA12 DA16 DB02 DC04 DC33                       DC36                 5L096 BA08 CA02 DA04 FA59 HA03                       JA03 JA11 KA17

Claims (6)

[Claims] 1. A storage unit in which feature amounts of human faces classified into a plurality of groups are stored in advance, a photographing unit for photographing a face image of a user, and a face image photographed by the photographing unit. Designating means for designating a group of persons to be subjected to similarity determination, face images photographed by the photographing means, and faces of all persons belonging to the group designated by the designating means stored in the storage means. Calculation means for calculating the degree of similarity with the feature amount, and determination means for determining the person having the face most similar to the face image photographed by the photographing means on the basis of the degree of similarity calculated by the calculation means. A face image recognition system comprising: a guide unit that guides the person determined by the determination unit. 2. A storage means in which various data are stored in advance corresponding to the facial feature amounts of a plurality of people, a photographing means for photographing a face image of a user, and a face image photographed by the photographing means. Specifying means for specifying the condition of the person who is the target of the similarity determination with, and all persons who meet the condition specified by the specifying means are selected based on various data stored in the storage means. Selecting means, calculating means for calculating the degree of similarity between the facial images photographed by the photographing means and the facial feature amounts of all the persons selected by the selecting means, and based on the degree of similarity calculated by the calculating means And a guide means for guiding the person judged by the judging means, and a judging means for judging a person having a face most similar to the face image photographed by the photographing means. Recognition system Beam. 3. A face image recognition system having a user terminal that can be used by a user and a server device that can communicate with the user terminal, wherein the user terminal includes a photographing means for photographing a face image of the user. A designation means for designating a group of persons whose similarity is to be determined with the face image photographed by the photography means, and a face image photographed by the photography means among persons belonging to the group designated by the designation means. Guide means for guiding a person whose face is determined to be most similar to each other by the server device, wherein the server device stores in advance facial feature amounts of persons classified into a plurality of groups. The storage means, the face image photographed by the photographing means of the user terminal, and all the persons belonging to the group designated by the designation means of the user terminal stored in the storage means Calculation means for calculating the degree of similarity with the feature amount, and determination means for determining the person having the face most similar to the face image photographed by the photographing means on the basis of the degree of similarity calculated by the calculation means. A face image recognition system, comprising: 4. A face image recognition system having a user terminal that can be used by a user and a server device that can communicate with the user terminal, wherein the user terminal includes a photographing means for photographing a face image of the user. Of the persons who meet the conditions designated by the designating means, the designating means for designating the condition of the person whose similarity to the face image photographed by the photographing means is photographed by the photographing means. And a guide means for guiding a person who is determined by the server device that the face is most similar to the face image, and the server device includes various data corresponding to facial feature amounts of a plurality of persons. Storage means that is stored in advance, and a selection means that selects all persons who meet the conditions specified by the specification means of the user terminal based on various data stored in the storage means, Based on the similarity calculated by this calculating means, a calculating means for calculating the similarity between the face image photographed by the photographing means of the user terminal and the feature amounts of the faces of all the persons selected by the selecting means. A face image recognition system, comprising: a determining unit that determines a person having a face most similar to the face image captured by the image capturing unit. 5. A face image recognition method used in a face image recognition system having a storage unit in which feature amounts of human faces classified into a plurality of groups are stored in advance, in which a face image of a user is captured. And a designation step for designating a group of persons whose similarity is judged with the face image photographed by this photographing step, the face image photographed by the photographing step and the designation stored in the storage means. Based on the similarity calculated by the calculation process of the facial feature amount of all persons belonging to the group specified by the process, and the similarity calculated by the calculation process A face image recognition method comprising: a determination step of determining a person with a similar face; and a guidance step of guiding the person determined by this determination step. 6. A face image recognition method used in a face image recognition system having a storage means in which various data are stored in advance in association with a plurality of facial features of a plurality of people. And a designating step for designating the conditions of the person whose similarity is to be determined with the face image captured by this capturing step, and all the persons who meet the conditions designated by this designating step are stored in the storage means. A selection step of selecting based on various data stored in the calculation step, and a calculation step of calculating the degree of similarity between the face images photographed in the photographing step and the facial feature amounts of all the persons selected in the selection step. And a determination step of determining a person having a face most similar to the face image photographed in the photographing step based on the similarity calculated in the calculation step, and guiding the person determined in the determination step. A face image recognition method, comprising: