JP2007067972A - Conference system and control method for conference system - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide an effective image in a conference or the like by preferentially photographing a human figure deeply relating to the conference or a human figure making a speech deeply relating to the relevant conference, for example. <P>SOLUTION: A participant recognition section 3 performs participant identification, a processing section 5 acquires human figure information of an identified human figure from a database 11 holding human figure information relating to human figures for each human figure, and an inference section 10a determines a priority order for each human figure, based on the recognized human figure information of each of the human figures. A camera 1 is controlled, based on the priority order determined for each human figure. <P>COPYRIGHT: (C)2007,JPO&INPIT

Description

  The present invention relates to a conference system applicable to, for example, a video conference system and a conference system control method.

  2. Description of the Related Art Conventionally, a video conference system that can hold a conference between a plurality of conference participants in a remote place has been realized. This system consists of a camera, a display, a microphone, etc. When a conference is performed by shooting the entire conference room with a fixed camera, or when a conference participant manually controls a camera installed on the other side. There is a case where a desired video is displayed or a case where a speaker is identified by voice and a speaker is photographed. The video was displayed on the display, and participants at remote locations had a meeting through the video displayed on the display.

  For example, Japanese Patent Application Laid-Open No. 2004-151820 discloses a device that can easily and instantly indicate a camera posture or the like desired by a user during a video conference or remote monitoring. This is because the video taken by the camera with the pan head and the video taken by the wide-angle camera are projected on the display unit, and the camera can move and shoot when controlling the attitude of the remote camera with the pan head while watching the video. The camera platform can be controlled while viewing the video of all areas. Thus, the user can feel as if he / she is holding the camera and controlling the shooting range while staying at the place like a photographer and can easily operate the camera.

  Further, Patent Document 2 discloses a device that aims to provide a conference photographing device that is easy to install and has no limitation on the number of participants. This is a participant who collects surrounding voice with an omnidirectional microphone, extracts the voiceprint of the participant from the voice, identifies the speaker from the voiceprint data, and speaks a rotating camera located at the center of the participant The speaker can be photographed by facing the person.

JP 2000-032319 A JP-A-10-304329

  Therefore, an object of the present invention is to improve the above-described technique and, for example, image effective in a meeting or the like by preferentially photographing a person who is deeply related to the meeting or a person who makes a remark related to the meeting. Is to provide.

According to a first aspect of the conference system of the present invention, the personal information of a person recognized by the recognition means from a recognition means for performing personal recognition of the person and a first database that holds the personal information related to the person for each person. Personal information acquisition means for acquiring the priority, priority determination means for determining the priority for each person based on the person information of each person recognized by the recognition means, and the priority determined for each person. And a control means for controlling a plurality of imaging devices.
According to a second aspect of the conference system of the present invention, there is provided an audio data acquisition unit that acquires audio data, a recognition unit that recognizes a person based on the audio data, and a keyword that extracts predetermined keyword information from the audio data Based on extraction means, priority determination means for determining the priority of the person recognized by the recognition means based on the keyword information, and priority determined for each person recognized by the recognition means, And a control means for controlling a plurality of imaging devices.
A first aspect of a conference system control method according to the present invention is a conference system control method having a plurality of imaging devices, in which a recognition step for performing personal recognition of a person and person information related to the person are held for each person. A personal information acquisition step for acquiring the personal information of the person recognized by the recognition step from the database, and a priority for determining the priority for each person based on the personal information of each person recognized by the recognition step And a control step for controlling the imaging device based on the priority determined for each person.
According to a second aspect of the conference system control method of the present invention, there is provided a conference system control method having a plurality of imaging devices, an audio data acquisition step for acquiring audio data, and recognition of a person based on the audio data. Performing a recognition step, a keyword extraction step for extracting predetermined keyword information from the voice data, a priority determination step for determining a priority of the person recognized by the recognition step based on the keyword information, And a control step of controlling at least one imaging device that captures an image based on the priority determined for each person recognized in the recognition step.
The program of the present invention causes a computer to execute the first or second aspect of the control method of the conference system.
The computer-readable recording medium of the present invention is characterized in that the program is recorded.

  In the present invention, the priority is determined for each person based on the person information of each person and the keyword information included in the audio data, and the photographing apparatus is controlled based on the priority. Therefore, according to the present invention, a high priority is given to, for example, a person who has personal information that is closely related to the meeting, a person who makes a speech that is deeply related to the meeting, and the like. It is possible to provide an effective image for a meeting or the like by preferentially shooting the image.

  DESCRIPTION OF EXEMPLARY EMBODIMENTS Hereinafter, preferred embodiments to which the invention is applied will be described in detail with reference to the accompanying drawings.

  In the video conference system according to the embodiment of the present invention, the participants are prioritized according to the participant information such as the attributes and background information of the participant and the content of the conference, and the person who is deeply related to the agenda is focused on. Do camera work.

  FIG. 1 is a configuration diagram of a video conference system according to an embodiment of the present invention. The video conference system according to the present embodiment includes a plurality of cameras 1 (imaging devices) for photographing participants, a display 2 for displaying participants at external points, and a microphone 4. In addition, the participant recognition unit 3 includes an internal or external connection to the display 2, a processing unit 5 that associates the participant with the participant information, a recording unit 6, an inference unit 10a that calculates the priority of the participant, and participant information. It has a database 11 and a keyword database 12. The processing unit 5, the recording unit 6, the inference unit 10a, the database 11, and the keyword database 12 are configured by an information processing device such as a personal computer.

  Next, a method for acquiring participant information by the participant recognition unit 3 will be described with reference to FIG. FIG. 2 shows two embodiments of a method for acquiring participant information. In the example illustrated in FIG. 2A, an example in which the participant recognition unit 3 is configured by the reader 8 is illustrated. The reader 8 is installed at the entrance of the conference room and recognizes the participant by reading the personal information in the tag 7 possessed by the participant. The tag 7 held by the participant communicates with the reader 8 in a non-contact manner by wireless communication. However, as another embodiment, the tag 7 communicates with the reader 8 in a contact type, and the personal information recorded therein is contact-type communication. Alternatively, a communication device that causes the reader 8 to read may be used. The processing unit 5 refers to the database 11 from the read participant ID, and acquires information on the participant (participant information). The inference unit 10 a calculates the priority of the participant based on the participant information acquired by the processing unit 5, and records the participant information and the priority in association with each other in the recording unit 6.

  In the example illustrated in FIG. 2B, the participant recognition unit 3 includes the camera 1 and the image processing unit 9a and recognizes the participant through image recognition. Specifically, a participant who enters the conference room is photographed with the camera 1. The image processing unit 9a extracts the participant feature information from the image data captured by the camera 1, and recognizes the participant from the extracted feature information and the person feature information registered in the feature database 13a. Consists of. Then, the processing unit 5 acquires an attribute and a history (participant information) from the database 11 based on the recognized identification information of the participant. The inference unit 10 a calculates the priority based on the participant information acquired by the processing unit 5, and records the participant information and the priority in the recording unit 6. In addition, although not shown here, the participant recognizing unit 3 when entering the room may recognize the participant's fingerprint, iris, vein, etc., and perform authentication of the participant by biometrics by the same subsequent processing. .

  Here, the database 11 used for acquiring participant information will be described in detail. FIG. 3 is a diagram schematically illustrating a configuration example of data stored in the database 11. In the attribute, title, affiliation, specialization, name, etc. are recorded, and in the career, past affiliation, work in charge, etc. are recorded. In addition, record keywords in related fields, personal papers and patents. Articles and patents may be collected from an external database connected to the network instead of a database connected to the in-house network. These pieces of information can be used to determine how deeply the participant is related to the agenda of the conference and whether the participant has the right to make a decision on the agenda.

  In order to perform shooting while paying attention to a specific participant in the video conference system, it is necessary to recognize the position of the participant who is where. A method for acquiring the position of the participant will be described with reference to FIG. FIG. 4 shows two embodiments of the method for acquiring the position of the participant.

  FIG. 4A shows a method of acquiring the location information of the participant by connecting the flow line analysis and the participant information. This method is as follows. First, personal recognition is performed by the participant recognition unit 3 at the entrance of the conference room before the participant enters the room. The order in which the participants are recognized is stored in the recording unit 6. Each participant recognized at this time is assigned a number in the order of entry. When the participant enters the room, the participant who has entered the room is photographed by the camera 1 installed in the room, and the flow line analysis is performed by the second image processing unit 9b. Specifically, each participant who has entered the room is tracked by the camera 1, and the movement line of the participant and the final position information are accumulated as a flow line analysis result, whereby the flow line of the participant can be grasped. Then, by associating the order in which the participants are recognized by the participant recognition unit 3 with the result of the flow line analysis corresponding to the order in which the rooms are entered, it is possible to specify who is sitting where. The result is recorded in the recording unit 6 together with the participant information as the location information of the participant.

  FIG. 4B shows a method for acquiring participant information and position information by the participant recognition method of FIG. In this case, the image processing unit 9a recognizes the participant by recognizing the image based on the image data taken by each camera 1 after the conference participant is seated, and the direction and distance from the camera 1 to the participant are determined. By calculating, recognition of the participant and acquisition of the location information of the participant can be performed. After the recognition of the participant, the processing unit 5 refers to the participant information such as the attribute and career of the participant from the database 11, and the inference unit 10a calculates the priority based on the participant information, and the position information At the same time, processing for recording in the recording unit 6 is performed.

  Here, in the case of FIG. 4A, in order to prevent a mismatch between the participant information and the position information due to a deviation in the order of recognition of the participants when entering the conference room and the order in which the participants enter the room, FIG. You may use together the method shown in b). However, in this case, since the participant information is obtained by referring to the database 11 before entering the room, it is not necessary to obtain the participant information by referring to the database 11 again after entering the room, and it is also necessary to calculate the priority. Absent. Further, since the participant is specified at the time of entering the room, the image recognition by the method shown in FIG. If it is determined that there is a mismatch, the other participants who are participating in the conference are recognized and the correct participants are associated with the position information.

  Next, a processing flow of the information processing apparatus for prioritizing when a participant is recognized will be described with reference to the flowchart of FIG. When the recognition information of the participant is received from the participant authentication unit 3 in step S1, the processing unit 5 inquires the database 11 and acquires the corresponding participant information in step S2. At this time, the participant's attribute information, career information, the number of patents of the participant, and points of contents are acquired as the participant information. Further, in step S3, the processing unit 5 counts the number of participants from the acquired number of participant information. In step S4, a priority is calculated by the inference unit 10a from the keyword related to the conference and the participant information, and recorded in the recording unit 6 together with the participant information. For keywords related to the conference, the chairperson or the participant inputs in advance the keywords related to the conference and the content of the conference using the in-house intranet conference reservation system to which the information processing apparatus is connected. Store it. In subsequent step S5, when the number of participants currently recognized is more than one, the inference unit 10a performs prioritization based on the priority calculated in step S4. By performing this process for all the participants, it is possible to rank all the participants in the priority ranking.

  Here, the priority calculation method in step S4 will be described. The priority is calculated from the participant's right to determine and relevance. The decision right is a value indicating how much the participant has the decision right of the conference. The degree of association is a value indicating how much the participant is related to the agenda. The decision right is given a numerical value according to the job title included in the attribute information. For example, when the position is a general manager and a general position, the decision-making authority is higher for the general manager. In addition, a social influence degree that considers the number of hits when a name is searched using a search system connected to a network may be added to the decision right.

  Next, the relevance will be described below. The degree of association is calculated from the sum of the value obtained from the degree of matching of each keyword and the depth of association, the number of keyword matches and the value obtained from the history. FIG. 6 shows an example of a tree structure keyword table registered in the keyword database 12.

  First, the degree of keyword matching and the depth of association will be described. For example, when a keyword “template matching” is registered as a keyword related to the agenda, if the keyword “template matching” is included in the participant information, the keyword matching degree is 1. If there is a keyword “matching” that is not “template matching” but similar, the keyword matching degree is calculated as 0.9, for example. Furthermore, the degree of association is considered in the degree of matching, and the degree of association is 1 if the matched keyword is a keyword that exists at the end of a branch of the tree structure. However, the closer the word is to the trunk of the tree structure, the broader the keyword, so the related depth is close to zero. For example, even if the degree of matching is high for a keyword such as “image processing”, the meaning included in “image processing” is broad (close to the trunk of the tree structure). Therefore, when both the value corresponding to the keyword matching degree and the value corresponding to the related depth are considered, for example, when these values are multiplied, the obtained value becomes a value smaller than 1.

Next, the number of keyword matches and the number of years will be described. This is calculated by referring to the database 11 and based on the number of matched keywords and the history of the detected keywords. For example, if the total number of keyword types that match a keyword related to the agenda in the participant information is the same for multiple participants, the participation that is currently related to this job rather than the participants that were related to the past The person is more relevant. In addition, regarding the number of keywords, it is difficult to determine the presence or absence of relevance at any one of the numbers of detected keywords, so fuzzy is used to provide ambiguity. FIG. 7 shows an example of fuzzy variables and fuzzy rules in this case. FIG. 7A shows a consequent part fuzzy variable, and FIG. 7B shows a fuzzy rule. In the solution synthesis method, the calculation of taking the center of gravity in the Min-Max method can be defined as Equation 1 where y ₀ is the center of gravity to be obtained, y is the horizontal axis, and μ (y) is the combined fuzzy set.

In this way, the degree of association is calculated from the sum of the value obtained from the keyword matching degree and the depth of association, and the value obtained from the keyword matching number and the history. By using this relevance level for priority calculation, participants who are related to the agenda have higher priority even if they have the same position.
For convenience, only the value obtained from the keyword matching degree and the depth of association may be used as the degree of association, or only the value obtained from the keyword matching number and the number of years may be used as the degree of association.

  As explained above, it is possible to flexibly deal with unscheduled participants by automatically prioritizing participants when the conference is held based on the agenda of the conference, related keywords and attributes of participants, and background information. . Furthermore, even if the chairperson does not know the past backgrounds and specialized fields of the conference participants, the priorities of the participants can be given each time according to the content of the conference.

  An example of camera work based on the priorities of the participants described so far is shown in FIG. As described above, the location information of the participant is recorded in the recording unit 6 together with the participant information. Based on this position information, the participant 1 is photographed by the camera 1. When the number of participants exceeds the number of cameras 1, control is performed to assign the cameras 1 installed at appropriate positions to the participants in order of priority of the participants other than the camera 1 shooting the speaker. It is. Further, for example, the participant with the highest priority may be controlled to perform zoom-up shooting as shown in FIG. As for the participant with the second highest priority order, as shown in FIG. 8 (b), it may be controlled that several people are zoomed together and photographed. Other participants may be controlled to take a wide angle image as shown in FIG. By viewing the video taken in this way, a speaker or a participant participating in the conference can see the reaction of the participant suitable for the agenda. Note that camera work is not limited to the one shown here.

  As described above, by performing camera work according to priority, it is possible to preferentially photograph decision makers and participants who are closely related to the agenda, and speakers who are on the speaker side or the listener side You can see the responses of the decision makers and participants closely related to the agenda.

  Next, as another embodiment of the present invention, a video conference system that detects and recognizes keywords from the voices of the participants, prioritizes the participants, and performs camera work according to the agenda and the flow of the story will be described. .

  FIG. 9 is a configuration diagram of a video conference system according to another embodiment of the present invention. The information processing apparatus of this embodiment includes a recording unit 6, a second inference unit 10b, a keyword database 12, a second database 13b, and a voice processing unit 14. The voice processing unit 14 detects and recognizes a keyword from the voice of the participant and associates it with the participant. The second reasoning unit 10b calculates the priority based on the speech time and the speech keyword. A basic configuration other than the second feature database 13b that records feature information of audio data for each person (for example, a configuration for detecting the position of the participant and controlling the drive of the camera 1 based on the detection result). Etc.) is the same as that of the first embodiment, and the description thereof is omitted.

  First, the flow of processing for prioritizing when a participant's speech is recognized will be described with reference to the flowchart of FIG. In step S11, the voice processing unit 14 acquires the voice data of the speaker from the microphone 4, and recognizes the speaker from the feature information of the acquired voice data and the feature information of the person registered in the feature database 13a. Next, the voice processing unit 14 records the correspondence between the speaker and the participant in the recording unit 6 in step S12.

  Next, the voice processing unit 14 performs voice division of the voice data in step S13, and extracts keywords from the participant's utterance in step S14. In step S15, the recording unit 6 further deals with the utterance time for the speaker and the participant. Add and register.

  Next, the second reasoning unit 10b calculates a priority from each keyword extracted in step S16, and registers the priority in association with the speaker, the participant, and the speech time in the recording unit 6.

  In the keyword database 12, in addition to keywords registered in advance, keywords that frequently occur during meetings are also determined to be important, and are registered in the keyword database 12 as keywords for calculating priorities. For example, even if a keyword such as “background difference” is not registered in advance, if it is frequently used in a meeting, the priority of meeting participants having attributes and backgrounds related to it is high. Therefore, even if the participant has a low priority when starting the conference, the priority will increase according to the flow of the story. Thereby, a participant's priority can be calculated in real time according to the flow of the talk of a meeting. Therefore, even if the agenda changes in the middle of the content of the conference scheduled in advance, the prioritization of the participants can be performed each time according to the content of the conference (step S17).

  Here, the priority calculation method in step 16 will be described. The priority calculation method in step S16 is a method in which the priority calculation method in step S4 in FIG. 5 further considers whether or not the speaker is a speaker and the total speech time. When speaking, the speaker degree is set to 1. If the speaker level is set to 0 at the moment of silence, it is determined that the speaker is not a speaker even after a short breath. Therefore, it is assumed that the speaker is a speaker until a certain time after speaking. This is shown in FIG. As the speech time, a value converted from the total speech time is used. In the present embodiment, the priority is obtained by adding or multiplying these values to the value of each element used for calculating the priority in step S4 in FIG. However, in this embodiment, since participant information is not used in the priority calculation process, the year of the history in which the keyword is detected with reference to the database 11 is not considered in the calculation of the priority.

  FIG. 12 shows a screen display layout example when displaying the video selected by the display switching unit on the display on the listener side. FIG. 12A shows a case in which a speaker and a decision maker on the speaker side are displayed. Further, FIG. 12B shows a case where video images of participants who are closely related to the agenda are displayed at the same time. The layout of the display method is not limited to the types shown here. The screen configuration may be a single screen instead of a multi-screen. Screen settings can be changed automatically or manually. As for the screen display switching, there are modes such as attention person priority, speaker priority, and hybrid. In the attention person priority mode, a participant with a high priority is displayed on the main screen according to the priority order calculated by the inference unit 10a. A process of displaying on the sub-screen the video that is displayed and shooting the next highest priority participant is performed. In the speaker priority mode, the participant who has the highest priority among the speakers calculated by the second reasoning unit 10b is displayed on the main screen, and the video of the speaker with the next highest priority is captured. Processing to display on the sub screen is performed. In the hybrid mode, a speaker is displayed on the main screen, and a person of interest such as a decision maker is displayed on the sub screen. This display may be reversed. The mode setting may be automatically switched according to the speaking time or manually performed by the participant. For example, the display on the listener side is displayed in the speaker priority mode or the hybrid mode, and the display on the speaker side is displayed in the attention person priority mode.

  According to the above-described embodiment, the camera not only performs shooting while paying attention to the speaker, but also prioritizes the participant according to the agenda and the flow of the talk according to the attribute information and background of the participant, and It is possible to preferentially photograph participants who are closely related to decision-makers and agendas. Therefore, the speaker or other conference participants can see the reaction to the opinions of the speaker or the decision-makers at different points or participants who are closely related to the agenda, and display an effective video. You can have a video conference.

  In addition, when conducting meetings, it is assumed that important opinions often come from participants with knowledge related to the agenda. Therefore, by taking a picture of a person according to the agenda at the center, it is possible to reduce the frequent switching of camera images and the camera's sensitive response to sound.

  Also, an object of the present invention is to supply a storage medium storing software program codes for realizing the functions of the above-described embodiments to the above-described information processing apparatus, and a computer (or CPU or MPU) of the information processing apparatus. Needless to say, this can also be achieved by reading and executing the program code stored in the storage medium.

  In this case, the program code itself read from the storage medium realizes the functions of the above-described embodiments, and the program code itself and the storage medium storing the program code constitute the present invention.

  As a storage medium for supplying the program code, for example, a flexible disk, a hard disk, an optical disk, a magneto-optical disk, a CD-ROM, a CD-R, a magnetic tape, a nonvolatile memory card, a ROM, or the like can be used.

  Further, by executing the program code read by the computer, not only the functions of the above-described embodiments are realized, but also an OS (basic system or operating system) running on the computer based on the instruction of the program code. Needless to say, a case where the functions of the above-described embodiment are realized by performing part or all of the actual processing and the processing is included.

  Further, after the program code read from the storage medium is written in a memory provided in a function expansion board inserted into the computer or a function expansion unit connected to the computer, the function is determined based on the instruction of the program code. It goes without saying that the CPU or the like provided in the expansion board or function expansion unit performs part or all of the actual processing and the functions of the above-described embodiments are realized by the processing.

It is a block diagram of the video conference system which concerns on embodiment of this invention. It is a figure for demonstrating the method of acquiring participant information by a participant authentication part. It is a figure which shows typically the example of 1 structure of the data stored in a database. It is a figure for demonstrating the method to acquire a participant's position. It is a flowchart which shows the flow of a process which prioritizes when a participant is authenticated. It is a figure which shows an example of the keyword table of the tree structure registered into the keyword database. It is a figure which shows an example of a fuzzy transformation and a fuzzy rule. It is a figure which shows the example of a camera work according to a participant's priority. It is a block diagram of the video conference system which concerns on other embodiment of this invention. It is a flowchart which shows the flow of the process which prioritizes when a participant's statement is recognized. It is a figure which shows the relationship between a speaker degree and speech time. It is a figure which shows the example of a screen display layout at the time of displaying the image | video selected by the display switching part on a display.

Explanation of symbols

DESCRIPTION OF SYMBOLS 1 Camera 2 Display 3 Participant recognition part 4 Microphone 5 Processing part 6 Recording part 7 Tag 8 Reader 9 Image processing part 10a Reasoning part 10b Second reasoning part 11 Database 12 Keyword database 13a Feature database 13b Second feature database 14 Voice Processing part

Claims (22)

A recognition means for personal recognition of a person,
Person information acquisition means for acquiring person information of a person recognized by the recognition means from a first database that holds person information related to the person for each person;
Priority determination means for determining priority for each person based on the person information of each person recognized by the recognition means;
And a control unit that controls a plurality of imaging devices based on the priority determined for each person.   The priority determination unit collates the keyword information held in the second database with the person information of each person acquired by the person information acquisition unit, and prioritizes each person based on the comparison result. The conference system according to claim 1, wherein the degree is determined.   The priority determination means changes the priority to be determined according to the degree of coincidence between the keyword information held in the second database and the keyword information included in the person information corresponding to the keyword information. The conference system according to claim 2.   3. The priority determination unit changes the priority to be determined according to the content of keyword information included in the person information that matches the keyword information held in the second database. Or the conference system of 3.   3. The priority determination unit changes the priority to be determined according to the number of keyword information included in the person information that matches the keyword information held in the second database. The conference system described in 1.   The priority determination means refers to the time information corresponding to the keyword information included in the person information that matches the keyword information held in the second database from the first database, and according to the time information 6. The conference system according to claim 2, wherein the priority to be determined is changed.   7. The conference system according to claim 2, wherein the keyword information stored in the second database is keyword information related to a content of the conference.   8. The priority determination unit according to claim 1, wherein the priority determination unit changes a priority to be determined according to information indicating a position of a person included in the person information of each person. 9. Conference system.   9. The conference system according to claim 1, wherein the recognizing unit performs individual recognition of a person based on image data captured by the capturing unit.   The conference according to any one of claims 1 to 8, wherein the recognition means performs personal recognition of a person based on personal information of the person received from a communication device that performs contact or non-contact communication. system. Audio data acquisition means for acquiring audio data;
Recognizing means for recognizing a person based on the audio data;
Keyword extracting means for extracting predetermined keyword information from the voice data;
Priority determination means for determining the priority of the person recognized by the recognition means based on the keyword information;
And a control unit that controls a plurality of imaging devices based on the priority determined for each person recognized by the recognition unit.   The priority determination means collates the keyword information held in the third database with the keyword information extracted by the keyword extraction means, and determines the priority for each person based on the collation result. The conference system according to claim 11.   The priority determination means changes the priority determined according to the degree of coincidence between the keyword information held in the third database and the keyword information extracted by the keyword extraction means corresponding to the keyword information. The conference system according to claim 12, wherein:   The priority determination unit changes the priority to be determined according to the content of keyword information extracted by the keyword extraction unit that matches the keyword information held in the third database. Item 14. The conference system according to Item 12 or 13.   The priority determination means changes priority to be determined according to the number of keyword information extracted by the keyword extraction means that matches the keyword information held in the third database. Item 13. The conference system according to Item 12.   16. The priority determination unit detects a speech time of a corresponding person based on the voice data, and changes the priority to be determined according to the length of the speech time. The conference system according to any one of the above. A flow line analyzing means for analyzing a flow line of each person using image data photographed by the imaging device;
There is further provided a position specifying means for specifying the position of each person based on the order in which the person is recognized by the recognition means and the order in which the person's flow analysis is performed by the flow line analysis means. And
The conference system according to any one of claims 1 to 16, wherein the control unit controls the imaging device based on a position of each person specified by the position specifying unit. It further has a position specifying means for specifying the position of each person using image data photographed by the imaging device,
The conference system according to any one of claims 1 to 16, wherein the control unit controls the imaging device based on a position of each person specified by the position specifying unit. A control method for a conference system having a plurality of imaging devices,
A recognition step for personal recognition of a person,
A person information acquisition step of acquiring person information of the person recognized by the recognition step from a database holding person information related to the person for each person;
A priority determination step for determining a priority for each person based on the person information of each person recognized by the recognition step;
And a control step of controlling the imaging device based on the priority determined for each person. A control method for a conference system having a plurality of imaging devices,
An audio data acquisition step for acquiring audio data;
A recognition step for recognizing a person based on the audio data;
A keyword extraction step of extracting predetermined keyword information from the voice data;
A priority determination step for determining the priority of the person recognized by the recognition step based on the keyword information;
And a control step of controlling at least one imaging device that captures an image based on the priority determined for each person recognized in the recognition step.   A program for causing a computer to execute the conference system control method according to claim 19 or 20.   A computer-readable recording medium on which the program according to claim 21 is recorded.

Images