JP2003189273A - Speaker identifying device and video conference system provided with speaker identifying device - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a video conference system capable of always and accurately displaying a speaker on a monitor screen. <P>SOLUTION: If a human detecting part 3 for detecting a person detects a person in an input image in a voice source direction obtained by a direction detecting part 2 for detecting the voice source direction of an input voice, a CGROM 6 and a superimposition generation circuit 4 display a prescribed marker indicating a speaker around the image of the detected person on the monitor screen. The human detecting part 3 is provided with a face extracting means 3a for extracting a face color of a person to detect the person, a facial contour detecting means 3b for detecting the facial contour, eyes, nose and lips of the person, a lips detecting means 3c for detecting the speaker by detecting lips' movement. The direction detecting part 2 can be provided with a first storage memory for storing voice source characteristics inputted in advance as a video conference participant, and a voice source characteristics comparing means for comparing the characteristics of a newly inputted voice source with those of the voice source stored in the first storage memory. <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 speaker identification device for a video conference system and a video conference system equipped with the speaker identification device. In particular, when a video conference is held by a plurality of participants, a speaker identification device of a video conference system having a function of adding a marker (mark) for identifying the speaker and displaying a marking on a monitor screen and the speaker The present invention relates to a video conference system including a person identification device.

[0002]

2. Description of the Related Art As a conventional technique of a video conference system for identifying a speaker and displaying a marker (mark) at the time of a video conference, for example, Japanese Patent Laid-Open No. 8-3 is available.
As disclosed in Japanese Patent No. 7655 "TV conference system having a speaker identification display function", only a voice input direction is identified, a direction in which a speaker is present is determined, and marking display is performed. There is. That is, in the technique disclosed in Japanese Patent Laid-Open No. 8-37655, when a plurality of participants hold a video conference, a voice direction detector for detecting the direction of sound generation, that is, a voice from a direction detecting means. The position (coordinates) on the monitor screen is obtained based on the direction data, and the camera device automatically moves to the position of the speaker (speaker) who is speaking in the video conference. , When the speaker is imaged, a marker (mark) that allows the speaker to be identified is provided, and the participant who is speaking in the video conference can be easily identified. Is said to be possible.

[0003]

However, in the conventional video conference system having such a speaker identification display function, it is not possible to identify the speech itself of the participants of the video conference. When a sound, that is, a sound is generated, a marker (mark) is added based on only the direction in which the sound is generated. Therefore, there is a problem that a marker (marker) for speaker identification is displayed even when a voice or a noise other than the speech of the participant of the video conference occurs.
For example, the marking display of the marker (mark) will be displayed even if there is a sound other than the speech of the participant, such as the sound of dropping a sneeze or a pen, or the sound of hitting an object with the microphone. Marking is displayed at a position different from the position showing the speaker who really needs the video conference, or the direction of the camera device has moved in an unpredictable direction. May cause the monitor screen to be very difficult to see or may be confusing.

The present invention has been made to solve the above problems, and in a video conference system for inputting and transmitting an image and a sound, for example, a direction detecting means for detecting a sound source direction from the input sound, and an input image. A human detection means for detecting a person from the human body, and a marking means for displaying a predetermined marker (mark) which is set in advance around the image of the person detected by the human detection means. Then, in the sound source direction obtained by the direction detecting means, when a person is detected by the human detecting means, first, it is determined that the voice is the voice of the speaker in the video conference, and the marking means displays the voice. A video conferencing system that enables a predetermined marker indicating the speaker to be displayed around the image of the person on the displayed monitor screen. There is provided St. videoconferencing system with a beam of speaker identification device and the speaker identification device.

[0005]

A speaker identification device of a video conference system and a video conference system equipped with the speaker identification device according to the present invention are configured by the following specific technical means.

A first technical means is a speaker identifying device of a video conference system for identifying a speaker in a video conference system for inputting and transmitting an image and a voice, and a direction detecting means for detecting a sound source direction from an input voice. Human detecting means for detecting a person from the input image, and marking means for displaying a predetermined marker around the image of the person detected by the human detecting means on the displayed monitor screen. A video conference system for displaying a marker indicating a speaker by the marking unit as a speaker in a video conference when a person is detected by the human detection unit in the sound source direction obtained by the direction detection unit. The speaker identification device is used.

A second technical means is the video conference system according to the first technical means, wherein the human detecting means further comprises face extracting means for extracting the color of the face of the person to detect the person. The speaker identification device is a feature.

A third technical means is the face contour according to the first or second technical means, wherein the human detecting means detects a human by extracting the contour of a person's face and the contours of eyes, nose or mouth. The present invention is characterized in that it is a speaker identification device of a video conference system further comprising extraction means.

A fourth technical means is the video conference system according to any one of the first to third technical means, wherein the human detecting means further comprises a lip detecting means for detecting that a person's lip is moving. The speaker identification device is used.

A fifth technical means is the fourth technical means, wherein when the lip detecting means detects that the lips of a person are moving, the marking means indicates the speaker. The present invention is characterized by being used as a speaker identification device of a video conference system for displaying markers by marking.

A sixth technical means is a case where the human detecting means detects a person in the sound source direction obtained by the direction detecting means in the fifth technical means, and the lip detecting means. In the above, when it is detected that the lips of a person are moving, the speaker identification device of the video conference system displays the marker indicating the speaker by the marking means. is there.

A seventh technical means is the method according to any one of the first to sixth technical means, wherein the direction detecting means stores a characteristic of a sound source, and a characteristic of a newly input sound source. And a sound source feature comparison unit that compares the feature of the sound source stored in the first storage unit, and the human detection unit further includes a second storage unit that stores the image position of the detected person. When it is detected that the feature of the sound source newly input by the sound source feature comparison unit matches the feature of the sound source stored in the first storage unit, the second storage unit stores the feature. Speaker identification of a video conference system in which the stored image position of the person is read and the marker indicating the speaker is marked and displayed by the marking means around the image of the person displayed on the monitor screen. It is characterized in that a location.

An eighth technical means is the same as the seventh technical means, wherein the sound source feature comparing means has a direction of the newly input sound source among the features of the sound source stored in the first storing means. In the case where it is detected by comparing the features of the sound source newly read by the sound source feature comparing means with the feature of the sound source in
When the person detecting section detects a person in the sound source direction obtained by the direction detecting section, the characteristics of the newly input sound source are stored in the first storage memory again, and the human detecting section is stored. It is a speaker identification device of a video conference system that stores the image position of the person detected in 1. in the second storage memory.

In a ninth technical means according to the seventh or eighth technical means, the sound source feature comparison means receives the new input from among the characteristics of the sound sources stored in the first storage means. When the feature of the sound source in the direction of the sound source is read out and the feature of the newly input sound source is compared by the sound source feature comparing unit and it is detected that they do not match, the direction detecting unit obtains When no person is detected by the human detecting means in the sound source direction,
The present invention is a speaker identification device of a video conference system in which the marking display of the marker indicating the speaker by the marking means is kept in the original state.

A tenth technical means is the method according to any one of the seventh to ninth technical means, wherein the second storage means uses the face contour extracting means included in the human detecting means in the input image. The human feature detecting means further comprises a facial feature storing means for extracting and storing facial features relating to a contour of a person's face and / or contours of eyes, nose or mouth, and the human detecting means stores the face stored in the facial feature storing means. It further comprises face feature comparison means for comparing the feature with a face feature newly extracted from the input image by the face contour extraction means, and the feature of the sound source newly input by the sound source feature comparison means and the first feature. 1 When it is detected that the features of the sound source stored in the storage unit match, the face feature comparison unit uses the face features stored in the face feature storage unit and the face contour extraction unit. Newly If the detected facial features are compared with each other and it is detected that they match, the image position of the person stored in the second storage means is read out and the person displayed on the monitor screen is displayed. The speaker identification device of the video conference system, in which the marker indicating the speaker is marked and displayed by the marking means around the image.

According to an eleventh technical means, in the tenth technical means, in the face feature comparing means, the face feature stored in the face feature storing means and the face contour extracting means of the human detecting means are newly added. Speaker comparison of the video conference system in which the marking display of the marker indicating the speaker by the marking means is kept in the original state when the extracted facial features are compared and it is detected that they do not match. The device is a device.

A twelfth technical means is the above-mentioned first to eleventh.
In any one of the above technical means, the direction detecting means,
Further comprising first storage means for storing the characteristics of the sound source, and sound source characteristic comparison means for comparing the newly input characteristics of the sound source with the characteristics of the sound source stored in the first storage means,
The human detection means further comprises a second storage means for storing the detected image position of the person, and a lip detection means for detecting that the detected person's lips are moving, and the sound source feature comparison means. When it is detected that the feature of the sound source newly input and the feature of the sound source stored in the first storage unit match, the feature of the sound source matches by the lip detecting unit. When it is detected that the lips of the person in the direction are moving, the image position of the person stored in the second storage unit is read and the image is displayed around the image of the person displayed on the monitor screen. The present invention is a speaker identification device of a video conference system in which the marker indicating the speaker is marked and displayed by the marking means.

A thirteenth technical means is the marking according to the twelfth technical means, when the lips of the person in the direction in which the features of the sound sources match each other are not detected by the lips detecting means. The present invention is a speaker identification device of a video conference system in which the marking display of the marker indicating the speaker by the means is kept in the original state.

A fourteenth technical means is the above-mentioned first to thirteenth.
In any one of the above technical means, the direction detecting means,
The method further comprises time detection means for detecting a duration time during which the input voice is continuously input, and the duration time of the input voice detected by the time detection means is equal to or longer than a preset predetermined duration time. When it is determined that the length of time has passed, the speaker detection device of the video conference system is activated by the human detection means to detect the person from the input image. is there.

The fifteenth technical means is the first to fourteenth aspects.
In any of the above technical means, the direction detecting means,
The input sound source further comprises time detection means for detecting a duration time during which the input voice is continuously input, and the duration time of the input sound source detected by the time detection means is equal to or longer than a predetermined duration time set in advance. The speaker identification device of the video conference system in which the operation for detecting the sound source direction of the input voice is started by the direction detecting means when it is determined that is there.

A sixteenth technical means is the fourteenth or the first
In the technical means of 5, the direction detecting means further comprises a voice level detecting means for detecting the voice level of the input voice, and the voice level of the input voice detected by the voice level detecting means is a predetermined voice. When it is detected that the level is equal to or higher than the level, the speaker identification device of the video conference system is started, in which the operation of detecting the duration by the time detector is started.

The seventeenth technical means is the above-mentioned first to sixteenth.
In any of the technical means of, the human detection means,
Detecting a person in the input image, the marking means, for each person detected by the person detecting means, of the markers color-coded and / or shape-coded around the image of the person displayed on the monitor screen. The present invention is characterized by being used as a speaker identification device of a video conference system that displays markings.

[0023] In an eighteenth technical means, in the seventeenth technical means, the marking means colors each person detected by the human detecting means by color coding around the image of the person displayed on the monitor screen. And / or when performing marking display of the markers divided into shapes,
Around the image of the person at the image position in the direction of the sound source obtained by the direction detecting means, a marking display of a speaker marker of a color and / or shape showing a speaker is obtained by the direction detecting means. A non-speaker marker of a color and / or a shape indicating a non-speaker is displayed around the image of a person at an image position other than the sound source direction as a speaker identification device of a video conference system. It is what

A nineteenth technical means is the first to the eighteenth.
In any of the technical means of, the human detection means,
When detecting a person in the input image, the detection time storage means for storing the detected detection time is further provided, and the detection time stored in the detection time storage means is defined as the participation time of the detected person. The speaker identification device of the video conference system displayed on the monitor screen is characterized by the above.

The twentieth technical means is the first to nineteenth aspects.
In any of the technical means of, the human detection means,
One or more second storage means capable of storing the image positions of one or more persons detected from the input image,
Further, the human detection means detects all the persons in the input image at every preset constant period, and when the image position of the detection result of each person is obtained, the detection result is obtained. Position comparison means for comparing each of the image positions of the persons with the image positions of the one or more persons stored in the second storage means, and the position comparison means includes the second storage means. If any of the image positions of the person stored in the above does not match any of the image positions of the person of the detection result, the person of the image position stored in the second storage means does not match. As a person leaving the seat, the leaving time storage means for storing the time when the person is no longer detected as the leaving time, and conversely, in the position comparing means, any one of the image positions of the persons of the detection result is In the second storage means If it does not match any of the stored image positions of the person, it is assumed that the person at the image position of the detection result that does not match newly joins, and the time when the person is newly joined is joined. A participation time storage means for storing the time, and the leaving time stored in the leaving time storage means, conversely, the participation time stored in the participation time storage means,
The present invention is characterized by being used as a speaker identification device of a video conference system displayed on a monitor screen.

The twenty-first technical means are the first to twentieth aspects.
In any one of the above technical means, the direction detecting means,
Speaker of the video conference system further comprising a sound source detection storage unit that stores the time when the sound source of the input voice is detected as a sound source detection time, and displays the sound source detection time stored in the sound source detection storage unit on a monitor screen. It is characterized by being used as an identification device.

The twenty-second technical means is the first to twenty-first aspects.
In any one of the technical means, the video conference further comprising at least a camera device for image input that is movable in the horizontal direction, and controlling the position of the camera device to face the sound source direction detected by the direction detecting means. The present invention is characterized by being used as a speaker identification device of the system.

A twenty-third technical means is the sound source direction detected by the direction detecting means in the twenty-second technical means,
When controlling so that the position of the camera device is directed, when a person is not detected by the human detecting means in the detected sound source direction, and / or the marker indicating the speaker is newly marked and displayed. When the desired state is not detected, the camera device is used as a speaker identification device of the video conference system that returns the camera device to the position before the operation.

A twenty-fourth technical means is the first to twenty-third aspects.
In any one of the technical means described above, at least the camera device for image input that is movable in the horizontal direction further includes a zooming device capable of performing zooming, and the camera device is provided in the sound source direction detected by the direction detecting device. And a speaker identifying device of a video conference system for controlling the zooming means to perform zooming.

A twenty-fifth technical means is the sound source direction detected by the direction detecting means in the twenty-fourth technical means,
When the position of the camera device is directed and the zooming means is controlled to perform zooming, when no person is detected by the human detecting means in the detected sound source direction, and / or When a state in which the marker to be displayed is not newly marked is not detected, the speaker device is a speaker identification device of the video conference system that returns the camera device to the position before the operation.

A twenty-sixth technical means is the first to the twenty-fifth aspects.
The video conference system is provided with the speaker identification device of the video conference system according to any one of the above technical means.

[0032]

BEST MODE FOR CARRYING OUT THE INVENTION An example of an embodiment of a speaker identification device of a video conference system according to the present invention and a video conference system equipped with the speaker identification device will be described below with reference to the drawings. FIG. 1 is a block configuration diagram showing an example of the configuration of a video conference system including a speaker identification device according to the present invention.

As shown in FIG. 1, a video conference system equipped with a speaker identifying apparatus according to the present invention includes a direction detecting section 2 for detecting a sound source direction from an input voice and a human detecting for detecting a person from an input image. And various types of CG (Computer Gra) to be displayed on the monitor screen.
CGRO, which is a ROM that stores information
The M6 and the superimpose generation circuit 4 for superimposing and displaying the CG information from the CGROM 6 on the monitor screen are provided as marking means for displaying the identification marker. The operation of the video conference system according to the present invention will be described with reference to FIG.

First, the input voice is converted by the voice input unit 1 into a voice signal composed of an electric signal, and the direction detection unit 2
To the audio codec 7. In the direction detection unit 2, the position of the speaker is detected as position information on the monitor screen based on the voice signal, and the position information is given to the camera control unit 8 and the human detection unit 3.

On the basis of the position information of the speaker, the camera control section 8 causes at least the horizontally movable camera device 9 to perform a turning operation and a zooming operation to generate a sound source (that is, a video conference). The position of the speaker in (1) is controlled so as to be positioned at the center of the monitor screen or is controlled so as to be enlarged and displayed by zooming, thereby performing position control or zoom control of the camera device 9.

Here, the camera control unit 8 includes the direction detection unit 2
When the turning control or the zoom control is performed so that the position of the camera device 9 is directed to the utterance direction (sound source direction) of the sound detected by the human detection unit, the human detection unit detects the direction of the sound source detected by the direction detection unit 2. 3 is not able to detect a person, and / or, for example, it is a voiced voice of a feature that does not match the feature of the voice (sound source) of the participants of the video conference, and the speaker is specified. If it is not possible (that is, if the state in which the marker indicating the speaker should be newly marked and displayed cannot be detected), it is possible to return the camera device 9 to the original position before the operation.

Alternatively, as described above, the camera device 9 is provided with the zooming means, and the camera control unit 8 controls the camera device 9 according to the user's instruction or in the sound source direction detected by the direction detection unit 2. It is possible to control the position so that the camera device 9 is automatically zoomed, and the human detection unit 3 cannot detect the person in the sound source direction detected by the direction detection unit 2. If and / or if the speaker cannot be specified (that is, if the state in which a marker indicating the speaker should be newly marked and displayed cannot be detected), the original camera device before the operation It is also possible to return to the position 9 and return the zooming state of the camera device 9 to the state before the operation.

In the camera device 9 as the image pickup device, the picked-up image information is converted into an electric signal as a video signal. The converted video signal is transferred to the video decoder 10
Is digitally processed by and is sent to the human detection unit 3 as video data. In the human detection unit 3, based on the position information input from the direction detection unit 2 and the video data input from the video decoder 10, a captured image of a person is located at a position indicated by the position information in the video data. It detects whether or not data exists.

When it is detected that the imaged image data of a person is present at the position indicated by the position information, various CGs (Computers) are used in the superimpose generation circuit 4 which provides marking means. Gra
CGRO, which is a ROM that stores information
Marker data that is predetermined as a marker (mark) indicating the speaker is read out from M6, and the marker display position (that is, the marker display position calculated from the position information on the video data and the captured image data of the person who is the speaker (ie, , Around the image of the person of the speaker), the composite video data in which the marker data indicating the read speaker is superimposed on the video data is created.

The generated composite video data is encoded image data that has been compression-encoded by the image codec 5, while the audio signal from the audio input unit 1 is compression-encoded by the audio codec 7. Encoded voice data, the encoded image data and the encoded voice data are multiplexed by the multiplexing circuit unit 11 and sent to the partner terminal through the communication line 12.

Here, the composite video data (that is, imaged by the camera device 9 and produced by the superimpose generation circuit 4 which is a marking means, and is recorded by the video decoder 10).
The image data coded by the image codec 5 is further compressed and encoded by the image codec 5 and the encoded image data obtained by combining the image data digitally processed with the marker data indicating the speaker is FIG. 2 shows a conceptual diagram of an example of an image which is decoded by the image codec in the terminal and is displayed on the monitor screen of the video conference system. In FIG. 2, reference numeral 101 denotes a speaker among participants of the video conference, 102 denotes a participant of the video conference who is not currently speaking, and 100 denotes
It is a marker (indicated by an arrow in FIG. 2) indicating that the speaker.

The human detecting section 3 extracts the color of the human face by using the color information from the video data obtained by digitally processing the video information from the camera device 9 by the video decoder 10. The face extraction means 3a that can identify (extract) and detect a person included in the video data. Further, the human detection unit 3 is
The video data from the camera device 9 is digitally processed by the video decoder 10 to extract the contour of a person's face and the contour having individual features such as eyes, nose, or mouth from the video data. Face contour extraction means 3b capable of detecting a person included in
Is also equipped.

Further, as will be described later, the human detecting section 3
Indicates that the video information from the camera device 9 is the video decoder 10
Also provided is a lip detecting means 3c capable of detecting whether or not the lips of a person are moving from the video data digitally processed by. When it is detected by the lip detecting means 3c that the lips of the person of the participant are moving, it does not depend on the result of the direction detecting section 2 or the person is detected in the sound source direction obtained by the direction detecting section 2. When a person is detected by the unit 3, and the lip detecting means 3
When it is detected that the lips of the person are moving in c, it is considered that the person is speaking, and the speaker is displayed around the image of the person who is the speaker on the monitor screen. Markers may be displayed by marking, and when the lip detecting means 3c detects that the lips of a person in a direction that matches the sound source direction detected by the direction detecting unit 2 are moving. Further, in the case where the characteristics of the uttered sound (sound source) and the characteristics of the uttered sound (sound source) of the video conference participant are matched by the direction detection unit 2, a marker indicating the speaker is displayed on the monitor screen by marking. It doesn't matter.

Further, the direction detecting section 2 and the human detecting section 3 shown in FIG. 1 can be configured as shown in FIG.
FIG. 3 is a block configuration diagram showing another configuration example of the direction detection unit and the human detection unit which configure the video conference system including the speaker identification device according to the present invention. That is,
As shown in FIG. 3, in addition to the direction detection unit 2 shown in FIG. 1, the direction detection control unit 2 ′ further includes a sound source feature extraction unit 21 for extracting features of an input sound (sound source) and an extracted input voice ( Sound source), which is a first storage means for storing the feature of the sound source) as feature data, and the feature of the sound source which is input through the direction detection unit 2 and extracted by the sound source feature extraction unit 21. Sound source feature comparison unit 23 for comparing and collating with all the feature data stored in the storage memory 22.
It has and.

On the other hand, the human detection control section 3 ′ is a person other than the human detection section 3 shown in FIG. 1 when a person is detected in the video data from the video decoder 10 in the human detection section 3. A second storage memory 32 that provides a second storage unit for storing the image position of the second image, and a second storage memory 32 that corresponds to the position in the utterance direction (sound source direction) of the sound indicated by the direction detection unit
A position calculation unit 31 for calculating the image position of the video data in the storage memory 32 is provided.

In the direction detection control unit 2'and the human detection control unit 3'shown in FIG. 3, first, prior to the start of the video conference, the first storage memory 22 and the second storage memory 32 are respectively provided. The feature data indicating the features of the voice (sound source) of each participant of the video conference and the image position of each participant are registered in advance. Here, the uttered voice of the speaker is converted into a voice signal of an electric signal by the voice input unit 1 and is input to the direction detection control unit 2 ', and is passed through the direction detection unit 2 that performs the same function as in FIG. , To the sound source feature extraction unit 21. In the sound source feature extraction unit 21, the voice signal (sound source) of the speaker
Is extracted and stored in the first storage memory 22 as characteristic data.

On the other hand, in the human detection section 3 which performs the same person detection function as in FIG. 1, the video data of the video decoder 10 corresponding to the position in the uttered voice direction (sound source direction) input from the direction detection control section 2 '. When the video of the person is detected at the image position in the inside, the image position of the person is stored in the second storage memory 32.

First storage memory 22 and second storage memory 32
Then, when the video conference starts after presetting the characteristics of the sound (sound source) and the image position of the person regarding the video conference participants, the sound source feature comparison unit 23 sends the sound source feature extraction unit 21. The feature of the new vocal sound (sound source) and the feature data of all the voices stored in the first storage memory 22 are compared and collated, and the new feature is added to the feature data of all the stored voices. When it is detected that a voice that matches the characteristics of the uttered voice (sound source) is present, it is determined that the participant of this video conference is speaking,
The image position of the person is read from the second storage memory 32 via the position calculation unit 31, and a marker indicating that the speaker is a superimpose is generated around the image of the person of the speaker on the monitor screen. The screen is displayed via the circuit 4.

As described above, in the case where it is judged that the speech of the participant of the video conference is based on the characteristics of the sound source which is the uttered voice, the figure which fulfills the same function as the human detecting section 3 shown in FIG. The human detection unit 3 in 3 does not need to perform any operation, and the human detection unit 3 can be assumed not to be activated.

On the other hand, if the features of the uttered voice (sound source) sent from the sound source feature extraction unit 21 do not match the feature data of all the voices stored in the first storage memory 22. When the sound source feature comparison unit 23 finds out, the human detection unit 3 in FIG. 3 having the same function as that of the human detection unit 3 shown in FIG. 1 is activated, and the human detection unit 3 outputs the signal from the video decoder 10. A person is detected from the video data. That is, the human detection unit 3 determines whether or not the image of the person is present at the position in the uttered voice direction (sound source direction) obtained via the position calculation unit 31, and thereby the uttered voice is detected. The voice (sound source)
It is determined whether or not it is the speech of the participant of this video conference.

When the human detection unit 3 determines that no person is present at the position in the uttered voice direction (sound source direction), the uttered voice (sound source) is a noise irrelevant to the video conference. And without any processing,
If it is determined that a person is present at the position in the uttered voice direction (sound source direction), it is considered that the speaker has moved the position and is speaking. The image position information of the participant calculated via the position calculation unit 31 is newly registered in the second storage memory 32, and the feature data of the uttered voice (sound source) is also stored in the first storage memory 22. Register again.

In the prior art, when a new voice is generated, a marker (mark) indicating the speaker is displayed on the monitor screen as a marking or the camera device is turned to detect the source of the voice. A speaker is trying to capture an image so that it fits within the monitor screen. However, even when some noise is generated by a person other than the speaker, the marker (marker) indicating the speaker is marked on the monitor screen or the camera device is swung in the direction of the noise. Was there.

In the video conference system equipped with the speaker identifying means according to the present invention, when the person detecting section 3 does not detect any person, a marker indicating the speaker is newly displayed. In addition, the camera controller 8 shown in FIG. 1 is controlled to send a return command so that the position of the camera device 9 and the zooming state return to the original state. That is, even if the position of the camera device 9 is once swung in the direction of the sound source, although not shown, the position information and zooming information before the camera control unit 8 is swung is shown in FIG. Second storage memory 3 shown
By being stored in 2, it is possible to send a return command for returning the position of the camera device 9 and the zooming state to the original state. Thus, even if the camera device 9 turns in response to a noise other than the speech of the participant, the position of the marker (mark) indicating the speaker or the camera device 9 is changed. The position can be returned to the original position.

Further, the second storage memory 32 extracts the facial features of the individual person concerning the contour of the person's face and the contours of the eyes, nose or mouth by the face contour extracting means 3b provided in the human detecting section 3. In the case of further comprising a facial feature storing means for storing the facial features, the human detecting section 3 newly extracts from the video data by the facial feature and facial contour extracting means 3b stored in the facial feature storing means. The sound source feature comparing unit 23 further includes a face feature comparing unit that compares the face feature and the feature of the sound source newly input by the sound source feature comparing unit 23 and the first storage memory 2.
When it is detected that the feature data (feature of the sound source) stored in No. 2 match, the face feature comparing means and the human detecting section 3 in the face feature storing means. The face features newly extracted by the face contour extracting means 3b of No. 1 are compared, and when it is detected that they match, the image position of the person stored in the second storage memory 32 is read out, It is also possible to mark and display a marker indicating a speaker around the image of the person at the image position, rotate the camera device 9 to the position of the person, or zoom and enlarge the display.

Here, in the face feature comparison means, the face features stored in the face feature storage means are compared with the face features newly extracted by the face contour extraction means 3b, and they do not match. Is detected, the marking display of the marker indicating the speaker, the position of the camera device 9 and the zooming state are left as they are.

Further, the direction detecting section 2 shown in FIG. 1 can be configured as shown in FIG. Here, in FIG.
It is a block block diagram which shows the further another structural example of the direction detection part which comprises the video conference system provided with the speaker identification apparatus based on this invention. That is, as shown in FIG. 4, in addition to the direction detection unit 2 shown in FIG. 1, the direction detection control unit 2 ″ further includes a voice level detection unit 24 for detecting a voiced sound (sound source) higher than a predetermined voice level. And a timer unit 25 which provides a time detecting means for detecting that the uttered voice exceeding the predetermined voice level continues for a preset predetermined duration or longer.

In FIG. 4, the input uttered voice is converted into a voice signal in the voice input unit 1 and input to the direction detection control unit 2 ″, and the voice level of the uttered voice is measured in the voice level detection unit 24. When it is detected as a speech from the participant that the uttered voice is a voice signal having the predetermined voice level or higher, the timer of the timer unit 25 is activated to start counting elapsed time. If the predetermined duration set in advance has elapsed,
The input sound (sound source) is in a state of continuing for a predetermined duration or longer, and the output signal from the timer unit 25 is
It is sent to the direction detector 2.

Here, before the preset predetermined duration has elapsed, the speech from the participant ends, and the voice level detecting section 24 detects the voiced voice of the predetermined voice level or higher. When it disappears, the timer unit 25 stops counting the timer, and the output signal from the timer unit 25 stops being generated. In the case where the output signal from the timer unit 25 is input, the direction detection unit 2 having the same function as in FIG.
The operation for determining whether or not the audio signal is based on the speech of the participant of this video conference is activated.

For example, since voice information for a short time such as sneezing is not a utterance, the camera device 9 shown in FIG.
It is wasteful to turn to the position of the voice information for a short time, or to display the position of the voice information on the monitor screen so as to indicate the speaker. As shown in the embodiment of the present invention, in the case of the voice information of such a short time, the output signal from the timer unit 25 is not output, and therefore the direction detection unit 2 uses the uttered voice direction. The operation for detecting the position information indicating the position of (sound source direction) is not activated, and the control operation such as turning of the camera device 9 or the person detection operation in the human detection unit 3 is not activated. .

Further, in FIG. 4, it is explained that, in the case of the audio information of such a short time, the human detecting unit 3 is not activated because the direction detecting unit 2 is not activated, but the case is not limited to such a case. In the case where the output signal is directly supplied from the timer unit 25 to the human detection unit 3 and the audio information is for a short time, the human detection unit 3 is not directly activated without going through the direction detection unit 2. It can also be in a state.

As described above, in FIG. 4, the condition for activating the timer unit 25 is that there is a vocal sound at a level equal to or higher than a predetermined predetermined audio level. When it is detected that some identifiable voice continues for a predetermined duration or longer regardless of whether or not the voice is a voice having a level higher than the predetermined voice level. I don't mind.

Further, as described above, the human detecting section 3 shown in FIG. 3 has the utterance voice direction (source direction) from the direction detecting section 2.
In addition to detecting the presence or absence of a person at the position on the video data indicated by the position, the image positions of all the participants in the video conference are detected prior to the video conference. 2 It is possible to register in the storage memory 32. Thus, for example, FIG.
As shown in, it is also possible to change the marker display of the speaker and the non-speaker among the participants by color or shape to display the markings on the monitor screen. It is possible.

FIG. 5 is a conceptual diagram showing an example of the monitor screen display when the marker display for the speaker and the marker for the non-speaker among the video conference participants is displayed. In FIG. 5, an example in which the speaker marker 103 indicating that the speaker 105 is speaking is displayed in a different color from the non-speaker marker 104 indicating the non-speaker 106, for example. Shows.

That is, of the video conference participants, the speaker 1
When 05 speaks, the direction detection unit 2 detects the position information indicating the sound source direction of the speaker 105, compares the position information with the image position stored in the storage memory 32 in advance, and detects the person in the sound source direction. By confirming the existence of
Speaker marker 103 of a different color from the non-speaker marker 104 that identifies 05 and indicates the non-speaker 106 who is not speaking.
Are displayed by.

Therefore, even if all the participants of the video conference are displayed on the monitor screen, all the participants of the monitor screen, that is, in the peripheral position of the image of the person, can see all the participants. On the other hand, speaker markers 103 of different colors are used.
Marking display with the non-speaker marker 104,
It is possible to easily identify whether or not the speaker. The marker displayed in the marking is not limited to the one for distinguishing the speaker from the non-speaker. For example, it may be color-coded and / or shape-coded and / or pattern-coded so that the moderator, the general participant and the observer of the video-conference can be easily distinguished, and any identification useful for conducting the video-conference is possible. Information can also be provided by the marker displayed on the monitor screen in a superimposed manner.

Further, the human detection control section 3'constituting the human detection section 3 does not perform the person detection operation only when the voice is inputted, but the human detection section 3 constantly and periodically detects the person. Can also be detected. Then, at each image position in the video data from the video decoder 10, the time when the person is first detected is read out from a clock circuit (not shown), and the detected time of the detected person is transferred to the video conference. As the participation time of the second storage memory 32 shown in FIG.
Is stored in the detection time storage unit, and the superimposing generation circuit 4 superimposes the detection time read from the detection time storage unit on the screen to display the detection result of a person on the monitor screen. The time when the person participated in the video conference can be displayed.

Alternatively, the human detection control section 3'constituting the human detection section 3 can store one or more image positions of the person detected from the video data from the video decoder 10, that is, the input image. Second storage memory 32 described above
Further, at a predetermined constant period, all the persons in the input image are detected at regular intervals, and the detection result is obtained when the image position of the detection result of each person is obtained. Each of the image positions of the
The apparatus further comprises position detecting means for comparing the image positions of the one or more persons stored in the storage memory 32, and in the position detecting means, one of the image positions of the persons of the detection result is 2 If it does not match any of the image positions of the person stored in the storage memory 32, it is considered that the person at the image position of the detection result has newly participated in the video conference, and the human detection unit 3 determines that The time when the person at the image position of the detection result is newly detected is stored as the participation time in the participation time storage unit of the second storage memory 32 shown in FIG. 3, and the participation time stored in the participation time storage unit is stored. To
It may be displayed on the monitor screen.

For example, FIG. 6 is a conceptual diagram showing an example of a monitor screen display showing the participation status of a certain video conference. In FIG. 6A, reference numerals 109 and 110 denote participants who participated in the video conference at 9:00, and the detection time when each person was detected, that is, the participation time is recorded in the video data on the monitor screen. Participating time displays 107 and 108 are displayed around the image of the person image position of the participant in an overlapping manner.

For example, FIG. 6B shows that from FIG.
It shows the participation status of the video conference after a while, and the participant 112 has participated in the video conference at 9:30, as the participation time display 111 is newly added and displayed on the monitor screen. Recognize. The display of the participation time is stored in the detection time storage unit or the participation time storage unit of the second storage memory 32 shown in FIG. 3 so that the participation of the person is displayed on the monitor screen until the video conference ends. The time can be displayed continuously.

Further, the human detection control section 3'includes one or more second storage memories 32 capable of storing one or more image positions of the person detected from the video data from the video decoder 10, that is, the input image. Further, at a predetermined constant period, all the persons in the input image are detected at regular intervals, and the detection result is obtained when the image position of the detection result of each person is obtained. Further includes position detecting means for comparing each of the image positions of the persons with the image positions of the one or more persons stored in the second storage memory 32. Memory 3
When any of the image positions of the person stored in 2 does not match any of the image positions of the detection result,
The second storage memory shown in FIG. 3 is defined as the time when the person at the non-coincident image position stored in the second storage memory 32 is considered to have left, and the time when the person is no longer detected by the human detection unit 3 is set as the departure time. The leaving time storage unit of 32 can be stored, and the leaving time stored in the leaving time storage unit can be displayed on the monitor screen.

That is, when the human detection section 3 no longer detects the person at the image position, that is, the image area where the person has been detected so far, the time when the person is no longer detected is read from a clock circuit (not shown), The superimpose generation circuit 4 displays the superimposed image on the monitor screen. Thus, the time when the person is no longer detected can be displayed on the monitor screen around the image of the image position of the person who was showing the participant, as the leaving time when the participant left the seat. Further, the leaving time when the person is no longer detected is stored in the leaving time storage unit of the second storage memory 32 shown in FIG. 3, so that the leaving screen of the person is displayed on the monitor screen until the video conference ends. The time can be displayed continuously.

Further, the human detecting section 3 detects the person who uttered the voice, based on the position information based on the voice direction (sound source direction) of the speaker and the video data based on the video information from the camera device 9. In this case, it is also possible to read the time from a clock circuit (not shown) and display the time read from the superimpose generation circuit 4 in an overlapping manner. That is, the direction detection control unit 2'stores the time when the sound source of the input sound is detected as the sound source detection time in the sound source detection storage unit of the first storage memory 22 shown in FIG. 3, and the sound source is displayed on the monitor screen. The sound source detection time stored in the detection storage unit can be displayed around the image of the person image position of each participant. Thus, the time when each participant speaks can be displayed on the monitor screen.

Further, the human detecting section 3 has a person detecting function based on the color of the face of the person, the outline of the face, etc., as described above.
It is also possible to further add lip detecting means 3c for detecting that the lips of a person are moving.

In such a case, as described above, when the voiced voice is input to the voice input unit 1, first, the person detecting function of the human detection unit 3 detects a person in the voiced voice direction (sound source direction). To be detected. After that, the lip detecting means 3c detects whether or not the detected lips of the person are moving. If the movement of the lips can be detected, it is determined that the utterance is the utterance of the person in the input utterance direction (sound source direction), and the turning operation and zooming operation of the camera device 9, and the utterance Marker indicating the person is displayed. On the contrary, when the lip detecting means 3c detects that the detected lips of the person are not moving, the input uttered voice is the speech of the person in the input uttered voice direction (source direction). It is determined that the voice is not produced by the camera device 9, and the turning operation and zooming operation of the camera device 9 and the marking display of the marker indicating the speaker are not performed.

Further, as shown in FIG. 3, the direction detection control unit 2'includes a first storage memory 22 for storing the features of the sound source extracted by the sound source feature extraction unit 21, and the features of the newly input sound source. And a sound source feature comparison unit 23 that compares the feature of the sound source stored in the first storage memory 22, and the human detection control unit 3 ′ stores the detected image position of the person. In the case of including the storage memory 32, in the case where the human detection control unit 3 ′ further includes the lip detecting means 3c for detecting that the detected lips of the person are moving, When it is detected by the sound source feature comparison unit 23 that the newly input feature of the sound source and the feature of the sound source stored in the first storage memory 22 match, and the feature of the sound source matches. Of the person in the direction in which That it is moving, lips detection means 3
When it is detected by c, first, the image position of the person stored in the second storage memory 32 is read out, and the marker indicating the speaker is displayed around the image of the person at the image position. It is also possible to direct the position of the camera device 9 toward the speaker or to perform zooming.

In such a case, the movement of the lips of the person in the direction in which the features of the sound sources are matched is not a short time when the lips detecting means 3c does not detect it. The camera device 9 and the position where the marker indicating the speaker is displayed as marking.
It is also possible to keep the position of the original state.

[0077]

As described above, according to the speaker identification device of the video conference system and the video conference system equipped with the speaker identification device according to the present invention, for example, a sneeze sound, a dropped object, a microphone, etc. When a sound other than the speech of the participant is generated, such as the sound of a thing hitting the object, the camera device accidentally turns in the direction of the sound other than the speech,
It is possible to prevent the marker indicating the speaker from being marked and displayed, and it is possible to always accurately display the participant who is speaking on the monitor screen. It is possible to provide a video conference system that can be conducted.

Further, even when all the participants of the video conference are displayed on the monitor screen, the marker for distinguishing the speaker or the non-speaker may be superimposed and displayed on the screen, or the TV may be displayed. It is also possible to superimpose a marker that is visually easy to identify the moderator of the conference, general participants, observers, etc., and display it on the screen.
A video conference system capable of smoothly performing a video conference can be provided.

Further, it is to provide a video conference system capable of displaying the elapsed time of the video conference such as the participation time when the video conference participant participated in the conference, the time when he / she left the seat, and the time when he / she spoke. You can

[Brief description of drawings]

FIG. 1 is a block configuration diagram showing an example of a configuration of a video conference system including a speaker identification device according to the present invention.

FIG. 2 is a conceptual diagram showing an example of an image displayed on a monitor screen in a video conference system equipped with a speaker identification device according to the present invention.

FIG. 3 is a block configuration diagram showing another configuration example of a direction detection unit and a human detection unit included in the video conference system including the speaker identification device according to the present invention.

FIG. 4 is a block configuration diagram showing still another configuration example of the direction detection unit included in the video conference system including the speaker identification device according to the present invention.

FIG. 5 is a conceptual diagram showing an example of a monitor screen display when marker display for speakers and non-speakers among video conference participants is performed.

FIG. 6 is a conceptual diagram showing an example of a monitor screen display showing a participation situation of a certain video conference.

[Explanation of symbols]

DESCRIPTION OF SYMBOLS 1 ... Voice input part, 2 ... Direction detection part, 2 ', 2 "... Direction detection control part, 3 ... Human detection part, 3' ... Human detection control part, 3
a ... Face extracting means, 3b ... Face contour extracting means, 3c ... Lip detecting means, 4 ... Superimpose generation circuit, 5 ... Image codec, 6 ... CGROM, 7 ... Voice codec, 8 ... Camera control section, 9 ... Camera Device, 10 ... Video decoder, 1
DESCRIPTION OF SYMBOLS 1 ... Multiplexing circuit part, 12 ... Communication line, 13 ... Storage memory, 21 ... Sound source feature extraction part, 22 ... First storage memory, 2
3 ... Sound source feature comparison unit, 24 ... Voice level detection unit, 25 ...
Timer unit, 31 ... Position calculation unit, 32 ... Second storage memory,
100 ... Marker, 101 ... Speaker, 102 ... Participant, 1
03 ... speaker marker, 104 ... non-speaker marker, 105
... speaker, 106 ... non-speaker, 107,108 ... participation time display, 109,110 ... participant, 111 ... participation time display, 112 ... participant.

   ─────────────────────────────────────────────────── ─── Continued front page    F-term (reference) 5B057 AA20 BA02 BA17 CA01 CA08                       CA16 CB01 CB08 CB12 CB16                       CE08 CH18 DA07 DA08 DA16                       DB02 DB06 DB09 DC16 DC25                       DC36                 5C064 AA02 AC04 AC06 AC08 AC09                       AC12 AC13 AC15 AC16 AC18                       AC22 AD02                 5L096 AA02 AA06 BA20 CA02 DA02                       FA06 FA15 GA38 HA03

Claims (26)

[Claims] 1. A speaker identification device of a video conference system for identifying a speaker in a video conference system for inputting and transmitting an image and a voice, and a direction detecting means for detecting a sound source direction from an input voice, and a person from an input image. And a marking means for displaying a predetermined marker around the image of the person detected by the human detecting means on the displayed monitor screen. When a person is detected by the human detection unit in the sound source direction obtained by the direction detection unit, a marker indicating the speaker is displayed as a speaker in the video conference by the marking unit. Speaker identification device. 2. The speaker of the video conference system according to claim 1, wherein the human detection means further comprises face extraction means for extracting the color of the face of the person to detect the person. Identification device. 3. The human detection means further comprises face contour extraction means for extracting a contour of a person's face and contours of eyes, nose, or mouth to detect the person. Alternatively, the speaker identification device of the video conference system according to item 2. 4. The speaker of the video conference system according to claim 1, wherein the human detection means further comprises lip detection means for detecting movement of a person's lips. Identification device. 5. The marker for indicating the speaker is displayed by the marking means when the lips detecting means detects that the lips of the person are moving. The speaker identification device of the video conference system described in 1. 6. A case where a person is detected by the human detecting means in the sound source direction obtained by the direction detecting means, and the lips of the person are detected to be moving by the lip detecting means. The speaker identification device of the video conference system according to claim 5, wherein the marker indicating the speaker is displayed in a marking manner by the marking means when the speaker identification is performed. 7. The sound source in which the direction detecting means compares a feature of a sound source newly stored with a first storage means for storing a feature of the sound source, and a feature of the sound source newly stored and a feature of the sound source stored in the first storing means. The human detection means further comprises second storage means for storing the detected image position of the person, and the feature of the sound source newly input by the sound source feature comparison means and the When it is detected that the characteristics of the sound source stored in the first storage means match,
It is characterized in that the image position of the person stored in the second storage means is read out, and the marker indicating the speaker is marked by the marking means around the image of the person displayed on the monitor screen. The speaker identification device of the video conference system according to claim 1. 8. The sound source feature comparison means reads out the feature of the sound source in the direction of the newly input sound source from the features of the sound source stored in the first storage means, and the sound source feature comparison means. When it is detected by comparing with the characteristics of the newly input sound source that the person does not match, when a person is detected by the human detecting means in the sound source direction obtained by the direction detecting means, The feature of the newly input sound source is stored in the first storage memory again, and the image position of the person detected by the human detecting means is stored in the second storage memory. The speaker identification device of the video conference system according to claim 7. 9. The sound source feature comparison means reads out the features of the sound source in the direction of the newly input sound source from the features of the sound source stored in the first storage means, and the sound source feature comparison means. When it is detected that they do not match by comparing the characteristics of the newly input sound source, and the person detecting unit does not detect a person in the sound source direction obtained by the direction detecting unit. The speaker identification device of the video conference system according to claim 7 or 8, wherein the marking display of the marker indicating the speaker by the marking means is kept in the original state. 10. The second storage means stores the facial features relating to the contour of the face of the person and / or the contour of the eyes, nose or mouth in the input image by the face contour extracting means included in the human detecting means. A face feature storage means for extracting and storing is further provided, and the human detection means is provided with the face feature stored in the face feature storage means and the face newly extracted from the input image by the face contour extraction means. A face feature comparison means for comparing the features with each other, wherein the feature of the sound source newly input by the sound source feature comparison means and the feature of the sound source stored in the first storage means match each other; If detected, the facial feature comparison means compares the facial features stored in the facial feature storage means with the facial features newly extracted by the facial contour extraction means, and finds a match. If detected Reading the image position of the person stored in the second storage means, and displaying the marker indicating the speaker by the marking means around the image of the person displayed on the monitor screen. The speaker identification device of the video conference system according to any one of claims 7 to 9. 11. The face feature comparing means compares the face features stored in the face feature storing means with the face features newly extracted by the face contour extracting means of the human detecting means, and matches them. 11. The speaker identification device of the video conference system according to claim 10, wherein the marking display of the marker indicating the speaker by the marking means is kept in the original state when it is detected that the speaker is not present. 12. The sound source in which the direction detecting means compares a characteristic of a sound source with a first storage means, and a characteristic of a newly input sound source and a characteristic of the sound source stored in the first storage means. The human detection means further comprises a second storage means for storing the detected image position of the person, and a lip detection means for detecting that the detected lip of the person is moving. Further, when it is detected that the feature of the sound source newly input by the sound source feature comparison unit matches the feature of the sound source stored in the first storage unit, the lip detection unit When it is detected that the lips of the person in the direction in which the features of the sound sources match each other are detected, the image position of the person stored in the second storage unit is read out and displayed on the monitor screen. Around the image of the person The speaker identification device of the video conference system according to any one of claims 1 to 11, wherein the marker indicating the speaker is marked and displayed by the marking means. 13. If the lips detecting unit does not detect that the lips of a person in the direction in which the features of the sound sources match each other are detected, the marking display of the marker indicating the speaker by the marking unit is used. The speaker identification device of the video conference system according to claim 12, wherein the speaker identification device remains in the state. 14. The direction detecting means further comprises time detecting means for detecting a duration time during which the input voice is continuously input, and the duration time of the input voice detected by the time detecting means is: When it is determined that the preset duration is longer than a preset duration, the human detection means activates an operation for detecting a person from an input image. Items 1 to 1
3. The speaker identification device of the video conference system according to any one of 3 above. 15. The direction detecting means further comprises time detecting means for detecting a duration time during which the input voice is continuously input, and the duration time of the input sound source detected by the time detecting means is: The operation for detecting the sound source direction of the input sound is started by the direction detecting means when it is determined that the predetermined duration time or more has been set in advance. Item 15. The speaker identification device of the video conference system according to any one of Items 1 to 14. 16. The direction detecting means further comprises a voice level detecting means for detecting a voice level of the input voice, and the voice level of the input voice detected by the voice level detecting means is a predetermined voice level determined in advance. The speaker identification device of the video conference system according to claim 14 or 15, wherein the detection operation of the duration by the time detection means is started when the above is detected. 17. The human detecting means detects a person in the input image, and the marking means, for each person detected by the human detecting means, the periphery of the image of the person displayed on the monitor screen. The speaker identification device of the video conference system according to any one of claims 1 to 16, wherein markers are displayed in different colors and / or shapes. 18. The marking means displays, for each person detected by the human detecting means, a marking display of the marker which is color-coded and / or shape-coded around an image of the person displayed on a monitor screen. When performing, around the image of the person at the image position in the sound source direction obtained by the direction detecting means, a marking display of a speaker marker of a color and / or shape indicating the speaker, and the direction detecting means The marking display of a non-speaker marker of a color and / or a shape indicating a non-speaker is performed around the image of a person at an image position other than the sound source direction obtained by the above. Speaker identification device of the described video conference system. 19. The human detection means further comprises a detection time storage means for storing the detected detection time when detecting the person in the input image, and the detection time stored in the detection time storage means. The speaker identification device of the video conference system according to any one of claims 1 to 18, wherein the time is displayed on the monitor screen as the detected participation time of the person. 20. The human detection means comprises one or more second storage means capable of storing the image positions of one or more persons detected from the input image, and the human detection means further comprises: , Every predetermined period set in advance, all the persons in the input image are detected, and when obtaining the image position of the detection result of each person, each of the image positions of the persons of the detection result And position comparison means for comparing the image positions of one or more persons stored in the second storage means with the position comparison means,
If any of the image positions of the person stored in the storage unit does not match any of the image positions of the person of the detection result, the image positions stored in the second storage unit do not match. If the person has left the seat, the leaving time storage means that stores the time when the person is no longer detected as the leaving time, and conversely, in the position comparing means, any one of the image positions of the persons of the detection result is displayed. If the person does not match any of the image positions of the person stored in the second storage means, it is determined that the person at the image position of the detection result that does not match newly joins the person. And a participation time storage means for storing the time detected as the participation time, and the leaving time stored in the leaving time storage means is conversely stored in the participation time storage means. The above Pressurized time, speaker identification device of the video conference system according to any one of claims 1 to 19, characterized in that displayed on the monitor screen. 21. The direction detection means further comprises a sound source detection storage means for storing a time at which a sound source of the input voice is detected as a sound source detection time, and the sound source detection time stored in the sound source detection storage means is stored. 21. The speaker identification device for a video conference system according to claim 1, wherein the speaker identification device is displayed on a monitor screen. 22. A camera device for image input, which is movable in at least a horizontal direction, is further provided, and the camera device is controlled so as to direct the position of the camera device to the sound source direction detected by the direction detecting means. 22. The speaker identification device of the video conference system according to claim 1. 23. When a person is not detected by the human detecting means in the detected sound source direction when controlling the position of the camera device toward the sound source direction detected by the direction detecting means. 23. And / or, if the state in which the marker indicating the speaker is to be newly marked and displayed is not detected, the camera device is returned to the position before the operation. Speaker identification device for conference system. 24. At least a horizontally movable image input camera device further comprises zoom means capable of performing zooming, and a position of the camera device is set in a sound source direction detected by the direction detection means. 24. The speaker identification device for a video conference system according to claim 1, wherein the speaker identification device is controlled so as to be directed and zoomed by the zoom means. 25. When controlling to direct the position of the camera device to the sound source direction detected by the direction detecting means and to perform zooming by the zooming means,
If no person is detected by the human detecting means in the detected sound source direction and / or a state in which the marker indicating the speaker is to be newly marked and displayed is not detected, the camera device The speaker identification device of the video conference system according to claim 24, wherein the speaker identification device is returned to the position before the operation. 26. A video conference system comprising the speaker identification device of the video conference according to claim 1. Description: