JP4908543B2 - Conference image reproduction system and conference image reproduction method - Google Patents

Description

The present invention relates to a conference image reproduction system and a conference image reproduction method that handle wide-angle images.

  In recent years, with the development of telecommunications technology, video conferencing systems that take pictures of conferences and transmit acquired images to remote locations have come to be used by many companies and organizations. In order to further improve the convenience of such a system, a number of devices for capturing the state of a conference as a video and a system for transmitting a partial video obtained by cutting out only a speaker have been proposed.

  As such a prior art, for example, a “video conference system” described in Patent Document 1 can be cited. Patent Document 1 discloses a technology relating to a video conference system in which a voice input from a microphone is detected to determine a speaker, and the camera control unit automatically controls the camera based on the determination result to capture the speaker. ing.

  Further, there is a “TV camera device” described in Patent Document 2. Patent Document 2 discloses a technique relating to a television camera device using a fish-eye or super wide-angle lens and a variable directivity microphone. Specifically, an invention is disclosed in which the direction of a sound source position is determined, the sound source position direction is tracked, and an image in the sound source position direction is cut out to generate a video signal.

  However, the conventional techniques as described above have the following problems. In other words, the technique disclosed in Patent Document 1 requires a certain amount of time to change the direction of the camera in the direction of the speaker, and the speaker is projected after a short while after the speaker starts speaking. There was a problem. In addition, there is a problem that the video flows during the movement of the camera, making it difficult to see the conference image. That is, there is a problem that the sense of reality is impaired.

  In addition, in the technique disclosed in Patent Document 2, when installing the TV camera device using a fish-eye or a super-wide-angle lens on a desk or the like, generally less important things such as a ceiling occupy most of the field of view. There is a problem that an important subject such as a human face exists in the peripheral part of the field of view and is affected by peripheral light reduction and aberration. That is, there is a problem that the conference cannot be viewed efficiently.

  In addition, when such a lens is used, there is a problem that the calculation for distortion correction greatly depends on the position of the image, and the calculation burden increases. In addition, the design of such a lens or optical system is very difficult and costly.

  In recent years, in addition to the so-called real-time property that is a feature of the conventional video conference system, there is a demand for checking the content of the conference again.

  The present invention has been made in view of the above, and an object of the present invention is to enable a meeting to be efficiently reproduced while maintaining a sense of reality.

  Another object is to provide a small and inexpensive device.

In order to solve the above-described problems and achieve the object, a conference image generation system according to the present invention includes an image input unit that inputs a wide-angle image, a plurality of audio input units that input sound, and a sound source direction from the sound. Included in the panoramic image, and a sound source direction detecting unit that detects a panoramic image including a region corresponding to the speaker position determined based on the sound source direction from the wide-angle image. An enlargement or reduction process is performed so that the speaker has an appropriate size with respect to the region of the speaker portion to be displayed.

When generating a panoramic image from a wide-angle image, the conference image reproduction system according to the present invention enlarges or reduces the area of the speaker portion included in the panoramic image, so that the speaker in the wide-angle image or the panoramic image Even when the sizes are different, enlargement or reduction processing is performed on the area of the speaker portion in an area in which the speaker has an appropriate size. Therefore, there is an effect that the conference can be efficiently reproduced while maintaining a sense of reality.

FIG. 1 is an explanatory diagram outlining an example of use in which the present invention is installed in a conference scene. FIG. 2 is an external perspective view of the conference image transmission device according to the first embodiment. FIG. 3 is a front view and a plan view of the conference image transmission apparatus according to the first embodiment. FIG. 4 is an explanatory diagram showing a configuration example of the camera unit of the conference image transmission device according to the first embodiment. FIG. 5 is a diagram for explaining an optical path when the hyperboloid mirror of the first embodiment is used. FIG. 6 is a diagram showing a state of a wide-angle image formed on the surface of the CCD by the hyperboloid mirror of the first embodiment. FIG. 7 is a diagram illustrating a configuration example of the conference recording / playback apparatus according to the first embodiment. FIG. 8 is a block diagram illustrating an example of a functional configuration of the recorded image reproduction system according to the first embodiment. FIG. 9 is a diagram for explaining the principle of detection of the sound source direction by the sound source direction detection unit of the first embodiment. FIG. 10 is a diagram for explaining that the direction in which the sound source exists is on a cone. FIG. 11 is an explanatory diagram showing a state of grouping when the sound source direction is detected by dividing four microphones into two groups. FIG. 12 is an explanatory diagram for explaining how to take a set of microphones when a microphone unit is constituted by three microphones. FIG. 13 is a diagram illustrating a data configuration example of the sound source direction according to the first embodiment. FIG. 14 is an explanatory diagram showing a state where a donut image captured from a hyperboloid mirror is transformed into a panoramic image. FIG. 15 is a diagram illustrating a coordinate system of a donut image and a panoramic image among the diagrams for explaining the deformation principle when a hyperboloid mirror is used. FIG. 16 is a diagram illustrating the relationship between the apex angle ψ and the elevation angle φ as viewed from the CCD, out of the diagrams for explaining the deformation principle when a hyperboloidal mirror is used. FIG. 17 is an explanatory diagram schematically showing an example of a conversion table when converting the coordinate system from the coordinate system (u, v) of the donut image to the coordinate system (θ, φ) of the panoramic image. FIG. 18 is an explanatory diagram showing an example of a processing flow of the conference recording / playback system according to the first embodiment. FIG. 19 is a diagram illustrating an example of an external configuration of the image recording / playback system according to the second embodiment. FIG. 20 is an explanatory diagram showing an example of a hardware configuration of the conference image recording / playback system according to the second embodiment. FIG. 21 is an explanatory diagram showing an example of a functional configuration of the conference recording / playback system according to the second embodiment. FIG. 22 is an explanatory diagram showing an example of image extraction in the second embodiment. FIG. 23 is an explanatory diagram illustrating a method of generating partial image data by the image extraction unit according to the second embodiment. FIG. 24 is an explanatory diagram showing an example of a processing flow of the conference recording / playback system according to the second embodiment. FIG. 25 is an explanatory diagram showing an example of an external configuration of an apparatus including the camera unit according to the third embodiment. FIG. 26 is an external configuration diagram of a camera unit configured to capture a donut image using two reflecting mirrors. FIG. 27 is an explanatory diagram illustrating the relationship between the microphone unit and the sound source direction according to the third embodiment. FIG. 28 is a diagram illustrating functional blocks of the conference image transmission device and the conference recording / playback device according to the fourth embodiment. FIG. 29 is a flowchart illustrating an example of a processing flow of the conference image transmission apparatus according to the fourth embodiment. FIG. 30 is a flowchart illustrating an example of a processing flow of the conference image reproduction device according to the fourth embodiment. FIG. 31 is a diagram showing an example of a screen configuration for selecting an image to be reproduced. FIG. 32 is a functional block diagram of the conference image transmission device and the conference recording / playback device according to the fifth embodiment. FIG. 33 is a flowchart illustrating an example of a processing flow of the conference image transmission apparatus 3201 according to the fifth embodiment. FIG. 34 is a flowchart illustrating an example of a processing flow of the conference image reproduction device according to the fifth embodiment. FIG. 35 is a diagram showing an example of a screen configuration for selecting an image to be reproduced. FIG. 36 is an explanatory diagram showing a configuration example (screen example) of an image output from the conference image recording / playback apparatus according to the fifth embodiment. FIG. 37 is an explanatory diagram showing a change in the configuration of the image shown in FIG. 36 after the “MANUAL” button is selected. FIG. 38 is a diagram illustrating an image in which the extraction portion is changed by the direction instruction operation unit according to the fifth embodiment. FIG. 39 shows another example of the screen configuration and is an explanatory diagram showing an example of a quad screen.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

[Embodiment 1]
In the first embodiment, a conference recording / playback system in which the wide-angle image recording / playback system of the present invention is applied to conference recording / playback will be described. Here, first, a usage example of how the conference recording / playback system is used will be briefly outlined. Next, elements constituting the conference recording / playback system (conference image transmission corresponding to an image and audio input unit) will be described. Apparatus and meeting image reproduction apparatus corresponding to the image and audio recording / reproduction unit), and finally the process flow will be described.

(Usage example of conference recording and playback system)
FIG. 1 is an explanatory diagram outlining an example of use in which the present invention is installed in a conference scene. The conference recording / playback system 100 includes a conference image transmission device 200 that inputs a wide-angle image and sound, and a conference image playback device 300 that records and reproduces the image and sound input by the conference image transmission device 200.

  As shown in the figure, the conference image transmission device 200 is installed on the table 1 and collectively captures images in the direction where the conference participants (speakers) 2 are located, that is, the entire periphery overlooking the horizontal plane. Also input the voice. The conference image playback device 300 records an image stored in the cabinet 3 and from the conference image transmission device 200, and plays back the recorded conference content as required according to the user's request. The monitor for is omitted.) During playback, the conference image playback device 300 transforms the entire image captured from the conference image sending device 200 into a rectangular output image.

  Next, each part of the conference recording / playback system 100 will be described.

(External configuration of conference image transmission apparatus 200)
FIG. 2 is an external perspective view of the conference image transmission device 200 according to the first embodiment. FIG. 3 is a front view and a plan view of the conference image sending apparatus 200 according to the first embodiment. The conference image transmission apparatus 200 includes a camera unit 201 that inputs a wide-angle image with the vertical direction as a center or an axis, and a microphone unit 202 that inputs sound. Here, the wide-angle image means an image including at least the entire circumference (360 °) overlooking the horizontal plane.

  As shown in the figure, the conference image transmission apparatus 200 according to the first embodiment has four microphones 221, and the microphone 221 and an image pickup device (CCD) of the camera unit 201 described later are arranged on a pedestal 203. Has been. Further, a hyperboloid mirror 211 of the camera unit 201 described later is disposed so as to face the pedestal 203 by the transparent glass 204. By using the transparent glass 204, an image of the entire periphery can be input without the light incident from the hyperboloid mirror 211 being shielded. Reference numeral 205 denotes a cable for transmitting various data.

(Conference image sending device 200: contents of camera unit 201)
FIG. 4 is an explanatory diagram showing a configuration example of the camera unit 201 of the conference image transmission apparatus 200 according to the first embodiment. The camera unit 201 includes a hyperboloid mirror 211, a lens 212, an aperture 213, and a CCD (Charge Coupled Device) 214 that is a photoelectric conversion element.

  The camera unit 201 controls the timing of the CCD 214 and also performs a digital signal obtained by the drive processing unit 215 and a drive processing unit 215 that performs A / D conversion (analog-digital conversion) on the video signal obtained by the CCD 214. Are provided with a pre-processing circuit 216 that performs pre-processing such as edge enhancement and γ correction, and a motor driving unit 217 that drives the diaphragm 213 to control the iris.

  Here, the optical system will be described. The hyperboloid mirror 211 is a reflecting mirror that enables wide-angle shooting. In the first embodiment, various explanations are given using a hyperboloidal mirror as an example of a reflecting mirror. However, the aspect is not limited as long as a wide-angle image can be captured. An example of another reflecting mirror will be described in Embodiment 3.

  As for the technology for capturing an image by the hyperboloid mirror 211, for example, A. M.M. Bruckstein and T.W. J. et al. Richardson: Omniview Cameras with Curved Surface Mirrors, Proc. of the IEEE Workshop on Omnivisional Vision 2000, pp. 79-84. According to the same magazine, it is shown that an important subject close to the horizontal direction such as a human face can be photographed with a relatively high resolution by using a hyperboloidal mirror.

  FIG. 5 is a diagram for explaining an optical path when the hyperboloid mirror 211 of the first embodiment is used. FIG. 6 is a wide-angle image formed on the surface of the CCD 214 by the hyperboloid mirror 211 of the first embodiment. It is the figure which showed the mode of. As illustrated, the image captured from the hyperboloidal mirror 211 has a donut shape. Note that the central portion in FIG. 6 reflects the direction of the pedestal 203, which is unimportant image information. Therefore, the top portion 218 of the hyperboloid mirror 211 may be painted black to obtain black information. Depending on the mode of use, a reference line may be drawn on the top of the head 218, and when starting the conference image sending apparatus 200, the motor driving unit 217 may be driven to be used for initial settings such as focus adjustment. Good.

  As described above, the camera unit 201 can be constructed by the general-purpose CCD 214 and the hyperboloid mirror 211 having a simple configuration. Therefore, it is possible to capture a desired subject at a high resolution in a batch and to provide an inexpensive camera unit 201.

(Conference image sending device 200: Configuration of microphone unit 202)
Next, the contents of the microphone unit 202 will be described. As described with reference to FIG. 2 or FIG. 3, the microphone unit 202 includes a plurality of microphones 221. Hereinafter, the plurality of microphones 221 will be appropriately referred to as a microphone array. As the microphone 221, various types such as a piezoelectric type and a capacitive type (so-called condenser microphone) can be used. As will be described later, the sound source direction (speaker direction) can be detected by using a plurality of microphones.

(Configuration of Conference Image Playback Device 300)
Next, the configuration of the conference image reproduction device 300 will be described. FIG. 7 is a diagram illustrating a configuration example of the conference image reproduction device 300 according to the first embodiment. The conference image reproduction apparatus 300 includes a CPU (Central Processing Unit) 301 that performs various controls and processes, an SDRAM (Synchronous Dynamic Access Memory) 302, an HDD (Hard Disk Drive) 303, a pointing device such as a mouse, , An input interface (hereinafter referred to as I / F) 304, a power source 305, a display I / F 306, a large capacity recording device 307 such as a DVD (Digital Versatile Disc) -RAM drive, and a conference image An external I / F 308 for connecting to the sending device 200, and connected via a bus 309. The display I / F 306 is connected to a display such as a CRT.

  Next, each component of the conference image reproduction device 300 will be described. The CPU 301 transforms the wide-angle donut shape image shown in FIG. 6 into a rectangular output image according to a predetermined program stored in the HDD 303. Further, the CPU 301 extracts a predetermined area in the sound source direction. This deformation and extraction process will be described later. The SDRAM 302 is used as a work area of the CPU 301 and also as a storage area for each processing program stored in the HDD 303 and other control programs (for example, OS).

  As described above, the external I / F 308 is an interface used when data sent from the conference image sending apparatus 200 is input. Here, examples of data input from the conference image transmission apparatus 200 include wide-angle images (moving image data), audio data, and sound source direction data. The external I / F 308 can adopt various I / Fs. For example, a wired connection such as USB (Universal Serial Bus) or IEEE 1394 may be adopted, or a wireless connection such as IrDA or BlueTooth may be adopted. Also good. Data input from the external I / F 308 is stored in the large-capacity recording device 307.

(Contents of wide-angle conference recording / playback system 100: functional configuration)
Next, image processing for transforming a wide-angle image into a rectangular output image and sound source direction detection processing will be described while describing the functional configuration of the wide-angle conference recording / playback system 100. FIG. 8 is a block diagram illustrating an example of a functional configuration of the conference recording / playback system 100.

  The wide-angle conference recording / playback system 100 has, as its functional configuration, a wide-angle image input unit 801, an audio input unit 802, a sound source direction detection unit 803, a recording unit 804, an image transformation unit 805, and a direction correction unit 806. A region fixing unit 807 and an image / sound output unit 808.

(Wide-angle Conference Recording / Reproducing System 100: Contents of Wide-Angle Image Input Unit 801)
The wide-angle image input unit 801 takes in a wide-angle image with the vertical direction as the center or axis and sends the image data to the recording unit 804. An example of a wide-angle image is the donut-shaped image shown in FIG. The wide-angle image input unit 801 can realize its function by, for example, the hyperboloid mirror 211, the lens 212, the diaphragm 213, the CCD 214, the drive processing unit 215, and the preprocessing circuit 216 shown in FIG.

(Conference Recording / Playback System 100: Contents of Audio Input Unit 802 / Sound Source Direction Detection Unit 803)
The voice input unit 802 inputs voice, converts it into an electrical signal (voice data), and sends the voice data to the sound source direction detection unit 803 and the recording unit 804. The voice input unit 802 can realize its function by the microphone 221 (see FIG. 2 or FIG. 3). A plurality of microphones 221 are arranged as described above, and a sound source direction is detected based on audio data from each microphone 221.

  The sound source direction detection unit 803 receives sound data from the sound input unit 802 and detects the sound source direction. By detecting the direction of the sound source, it is possible to extract (cut out) an image of the speaker (speaker) portion from the wide-angle image, and the conference can be efficiently reproduced while maintaining a sense of reality. Next, the sound source direction detection process will be described.

Here, a method will be described in which the sound source direction detection unit 803 detects the sound source direction based on the arrival time difference of the sound input to the microphone array. FIG. 9 is a diagram for explaining the principle of detection of the sound source direction by the sound source direction detection unit 803. As shown in the figure, when two microphones 221 (referred to as microphone 1 and microphone 2 for convenience) are arranged at a distance l and the sound reaches from the θ direction, the microphone 1 outputs The relationship between the audio data s1 (t) to be output and the audio data s2 (t) output from the microphone 2 can be expressed by the following equation (1), where t is the time v and the sound speed.
s1 (t) = s2 (t− (l · cos θ) / v) (1)

  Expression (1) indicates that the sound data of the microphone 1 has reached the sound data of the microphone 2 with a time advance of (l · cos θ) / v. The sound source direction detection unit 803 uses this arrival time difference to specify the direction of the speaker's voice.

In specifying the direction of the sound source, first, the arrival time difference between the audio data of the microphone 1 and the microphone 2 is detected. This arrival time difference is calculated by, for example, a cross-correlation value between the voice data s1 (t) of the microphone 1 and the voice data s2 (t + dt) of the microphone 2. The cross-correlation value C (t, dt) is calculated by the following equation (2).

  Expression (2) indicates that a product-sum operation is performed using N samples before time t. N is a positive integer indicating the size of the correlation window. Although detailed description is omitted, dt that maximizes C (t, dt) is the arrival time difference.

ここで、θの値域は０°以上１８０°以下とする。 Next, using the microphone interval l, the arrival time difference dt, and the sound velocity v, the angle θ formed by the voice and the base line of the microphone is calculated by the following equation (3).

Here, the range of θ is 0 ° or more and 180 ° or less.

  Note that the direction is detected only in the range of 180 ° on the front side of the microphone 221 only by the above procedure, and the sound source direction is not specified. That is, the angle θ output from the sound source direction detection unit 803 is actually an angle formed by the direction of arrival of the voice and the base line between the two microphones. The actual direction of the voice is 2 as shown in FIG. It exists on either side of the cone of apex angle θ with the midpoint of the two microphones as the apex.

  In order to solve this problem, correction is performed using another microphone pair that is not parallel to the pair composed of the microphone 1 and the microphone 2. FIG. 11 is an explanatory diagram showing a state of grouping when the sound source direction is detected by dividing four microphones 221 into two groups. As illustrated, the grouping combines a certain microphone 221 (for example, microphone 1 (microphone 3)) and a microphone 221 (microphone 2 (microphone 4)) farthest from the microphone.

  By using a pair of two microphones that are the farthest away, the difference in arrival time of the voice is maximized, and the accuracy of direction detection is improved. In the first embodiment, the microphone unit 202 includes the four microphones 221. However, the sound source direction can be detected with high accuracy using three microphones. FIG. 12 is an explanatory diagram for explaining how to take a set of microphones when a microphone unit is constituted by three microphones. As shown in the figure, by arranging the microphones in an equilateral triangle, the direction of the sound source can be detected with high accuracy regardless of which microphone pair is employed. In the illustrated example, the first set and the second set can be employed to detect sound sources in all directions, but the third set may be used in a complementary manner.

  The function of the sound source direction detection unit 803 can be realized by a control unit (not shown) of the microphone 221, for example. Depending on the mode of use, the function may be realized by the CPU 301 (see FIG. 7) on the conference image playback device 300 side. In this case, it is necessary to separately input the sound input from the microphone 221 to the conference image playback device 300 side.

(Wide-angle Conference Recording / Reproducing System 100: Contents of Recording Unit 804)
The recording unit 804 records video data of a wide-angle image output from the image input unit 801, audio data output from the audio input unit 802, and data related to the sound source direction output from the sound source direction detection unit 803. There are various recording methods. For example, moving image data is recorded in a moving image encoding format represented by MPEG. As for audio data, the MPEG audio format may be used or the PCM format may be used.

  With respect to the sound source direction data, it is possible to extract (cut out) a screen, which will be described later, by recording the time when the sound source direction is changed and the azimuth angle and elevation angle at that time. FIG. 13 is a diagram illustrating a data configuration example of the sound source direction. In the figure, the time (Time) when the sound source direction is changed, the azimuth angle (θ) and the elevation angle (φ) of the new sound source direction are recorded. This direction data is recorded in the large-capacity recording device 307 together with moving image data and audio data in the form of a text file or the like.

  In the above-described example, the sound source direction data is not data combined with moving image data or audio data. However, if a streaming format such as RealMedia format provided by RealNetworks is used, the sound source direction data is also one. Can be embedded in a file. In addition, sound source direction data can be written in a file using a content description standard for multimedia information such as MPEG-7.

  In addition, moving image data and audio data may be stored in one file and recorded as in an MPEG program stream. By using such encoding, the recording capacity can be reduced. The recording unit 804 can realize its function by the large-capacity recording device 307, for example. Note that the function may be realized by the HDD 303 depending on the mode of use. For example, a long-time meeting or a regular meeting is recorded on a large-capacity recording device 307 composed of a DVD or the like because of the necessity for storage, and a short-term meeting or the like that has a low necessity for long-term storage. Different usage such as recording in the HDD 303 may be performed.

(Wide-angle Conference Recording / Playback System 100: Contents of Image Deformation Unit 805 and Related Units)
Next, the image deforming unit 805 and the function unit associated therewith will be described. The image deforming unit 805 transforms the donut-shaped (or circular) wide-angle image so as to be a rectangular output image. In general, an image obtained by photographing a wide-angle range at a time includes large distortion unlike an image shape that can be confirmed by human eyes. Therefore, in order to reproduce the conference recorded in the recording unit 804 later, a transformation process is required.

  Deformation processing when the hyperboloid mirror 211 shown in FIG. 2 or FIG. 3 is used will be described. The image deforming unit 805 refers to the donut-shaped image shown in FIG. 6 (hereinafter referred to as a donut image) as a normal image (hereinafter referred to as a panoramic image) having a viewing angle of 360 degrees as shown in FIG. ).

  15 and 16 are diagrams for explaining the deformation principle when the hyperboloid mirror 211 is used. Among these, FIG. 15 is a diagram illustrating a coordinate system of a donut image and a panoramic image, and FIG. 16 is a diagram illustrating a relationship between an apex angle ψ and an elevation angle φ viewed from the CCD 214. In FIG. 16, the lens 212 and the diaphragm 213 are omitted for the sake of simplicity. Here, for convenience of the conversion formula, the optical system of the lenses 212 to CCD 214 will be described as a pinhole camera model.

The meaning of each variable in the figure is as follows.
(U, v): Coordinate in donut image (u0, v0): Coordinate of center of hyperboloid mirror in donut image (θ, φ): Coordinate in panoramic image r: From (u0, v0) to (u, v) Rmax: Radius of pixel unit of hyperboloid mirror in donut image θ: Azimuth angle φ: Elevation angle ψ: Apex angle from optical axis of camera F: Focal point of hyperboloid mirror F ': Pair with hyperboloid mirror Of hyperboloids
(Matches the optical center of the camera)

ここで、 At this time, the following relationship is established between the apex angle ψ and the elevation angle φ.

here,

である。また、φ_maxはドーナツ画像上の半径ｒ_maxの位置に対応する仰角の値であり、これはカメラの仰角方向の撮影限界値を表す。ｒ_maxとφ_maxの値は一般に容易に知ることができる。

It is. Φ _max is the value of the elevation angle corresponding to the position of the radius r _max on the donut image, and this represents the photographing limit value in the elevation angle direction of the camera. The values of r _max and φ _max are generally easily known.

Hereinafter, a modification procedure will be described.
(I): Polar coordinates (r, θ) corresponding to the point (u, v) are obtained by solving the following equation (6).
(U, v) = (r cos θ + u 0, r sin θ + v 0) (6)
(Ii): The vertex angle ψ corresponding to r calculated by the equation (6) is obtained by the following equation (7).

であり、ψ_maxはドーナツ画像上の半径ｒ_maxの位置および仰角φ_maxに対応する頂角ψの値である。ψ_maxの値は、（４）式にφ_maxを代入することにより求めることができる。
（ｉｉｉ）：（７）式により算出されたψに対応する仰角φを、（４）式により求める。 here,

Ψ _max is the value of the apex angle ψ corresponding to the position of the radius r _max and the elevation angle φ _max on the donut image. The value of ψ _max can be obtained by substituting φ _max into equation (4).
(Iii): An elevation angle φ corresponding to ψ calculated by equation (7) is obtained by equation (4).

  Through the above procedure, an arbitrary point (u, v) in the donut image photographed by the hyperboloid mirror 211 can be coordinate-converted to a point (θ, φ) in the panoramic image. That is, the donut image is transformed into a panoramic image.

  If the processing capacity of the processor is low, the transformation process of the image data takes a calculation time, so the conversion from (u, v) to (θ, φ) may be performed by referring to a predetermined conversion table. FIG. 17 is an explanatory diagram schematically showing an example of a conversion table of (u, v) → (θ, φ). The illustrated table stores a point (θ, φ) of the panoramic image corresponding to each point (coordinate (u, v)) of the donut image. Therefore, by using this table, it is possible to perform high-speed image deformation while reducing the processing load.

  The image deforming unit 805 performs the above conversion process and outputs a predetermined image area. That is, in order to efficiently reproduce the conference while maintaining the presence of the conference, the conference recording / playback system 100 extracts and outputs a speaker (speaker) portion of the panoramic image. As shown in FIG. 8, the conference recording / playback system 100 includes a speaker position determination unit 809 and an area determination unit 810 as functional configurations.

  The speaker position determination unit 809 determines the speaker position based on the color distribution of the image or the moving part in the image among the image data input from the wide-angle image input unit 801 or the image data recorded in the recording unit 804. As a determination method based on the color distribution of the image, for example, there is a method of detecting a portion where the skin color is locally large. Note that the determination by the moving part in the image is based on the fact that the speaker's mouth always moves, and in some cases the speaker moves his body by gestures. Therefore, it is possible to determine the speaker position based on the portion having the largest movement amount in the image.

  The area determination unit 810 determines which part of the speaker position determined by the speaker position determination unit 809 is to be extracted. When the table 1 is elliptical, the distance between the camera unit 201 and the speaker is different, and the size of the speaker in the wide-angle image or the panoramic image is also different. Therefore, if the area to be output is a uniform size, the speaker may be too large or too small in some cases. The area determination unit 810 determines the area of the speaker part in an area where the speaker has an appropriate size. Note that the image deforming unit 805 displays the image as enlarged or reduced as appropriate.

  On the other hand, the direction correction unit 806 corrects the direction corresponding to the sound source direction. This is based on the fact that the sound source direction detected by the sound source direction detection unit 803 may not be in a desired direction due to noise such as applause sounds or single words uttered by other than the speaker such as a reply. Further, for example, a request that the right side should be displayed slightly more than the area determined by the area determination unit 810 is practically generated. In particular, there is a case where the speaker wants to correct the direction when making a presentation and writing on the whiteboard. Therefore, the direction correcting unit 806 corrects the sound source direction so as to satisfy such a request.

  Further, the area fixing unit 807 fixes the image area determined by the area determining unit 810 in a direction corresponding to the sound source direction. In other words, the area determination unit 810 determines a relative area of, for example, 160 pixels × 90 pixels, whereas the area fixing unit 807 fixes the area as an absolute position so that the area does not move according to the sound source direction. . This is to prevent the image from blurring when the direction of the sound source moves slightly, such as when the speaker shakes his body.

  As described above, the image deforming unit 805 appropriately outputs the image of the speaker portion without distortion. The image deforming unit 805, the region fixing unit 807, and the speaker position determining unit 809 can realize the functions by, for example, the CPU 301 of the conference image reproduction device 300 illustrated in FIG. 3 and a predetermined program stored in the HDD 303. it can. In addition, the direction correction unit 806 and the area determination unit 810 include, for example, the CPU 301 of the conference image reproduction device 300 illustrated in FIG. 3, a predetermined program stored in the HDD 303, and a pointing device connected to the input I / F 304, The function can be realized with K / B and buttons.

(Conference recording and playback system 100: contents of image / sound output unit 808)
The image / sound output unit 808 associates and outputs the image (moving image data) output from the image deformation unit 805 and the sound simultaneously recorded (input) when the image is captured (input). That is, the image and sound are output in synchronization. Depending on the processing speed of the processor (for example, CPU 301), there is a time lag between sound and image, so the image and sound output unit 808 reproduces the conference with a natural feeling by synchronizing the image and sound. The image / sound output unit 808 can realize its function by, for example, the CPU 301 of the conference image reproduction apparatus 300 shown in FIG. 3 and a predetermined program stored in the HDD 303.

(Conference recording and playback system 100: process flow)
Next, the processing flow of the conference recording / playback system 100 will be described. FIG. 18 is an explanatory diagram showing an example of the processing flow of the conference recording / playback system 100. First, the conference recording / playback system 100 starts recording by pressing a recording start button (not shown) (step S1801). After this starting operation, wide-angle images (doughnut images) centered on the vertical direction or the axis are sequentially input from the camera unit 201, and sound is sequentially input from the microphone unit 202 (step S1802). As for the sound input from the microphone unit 202, the sound source direction is detected at any time using the above-described microphone set.

  Next, the donut image input from the camera unit 201, the sound input from the microphone unit 202, and the detected sound source direction are recorded (step S1803). For recording, a recording time, a file name (meeting name), etc. are given as appropriate for later playback. Note that since the sound source direction is detected, it is not necessary to record all four channels of each microphone 221 (that is, four microphones 221), and the average of any one or four sounds is recorded. That's fine. Although the time difference described above is generated in principle, this time difference is not a problem level in practice considering the size and speed of sound of the conference image transmission device 200.

  Recording is ended when a recording end button (not shown) is pressed in accordance with the end of the conference (step S1804). Through the above steps, it is possible to record the content of the conference composed of the entire surrounding image, that is, the donut image before processing. By recording the image before processing, it is possible to edit later (extracting the image area, correcting the direction of the sound source, etc.).

  Next, playback of a recorded conference will be described. When a reproduction start button (not shown) is pressed, reproduction is started (step S1805). When a plurality of conferences are recorded on a recording medium (for example, a DVD-RAM), playback is started after an index is displayed and the user selects which conference to play back.

  In the recorded donut image, the speaker position is determined using the skin color portion in the sound source direction as a clue (step S1806), and the display area is designated (step S1807). Here, if it is desired to consciously adjust the direction of the image, an instruction to correct the direction is given as appropriate.

  Subsequently, image transformation is performed so that the designated area of the donut image becomes a rectangular image (step S1808). When the image is deformed, the image may be deformed using a conversion formula, but a conversion table may be referred to. Finally, the extracted and properly deformed image is output together with the sound (step S1809). By taking such steps, the conference can be efficiently reproduced while maintaining a sense of reality.

  In the conference recording / playback system 100, it is preferable that the position of the center of gravity of the microphone unit 202 is designed to be on the optical axis of the camera unit 201. The most preferable design is an arrangement in which the center of gravity of the CCD 214 and the center of gravity of the plurality of microphones 221 coincide. By designing or arranging in this way, the coordinate system for calculating the sound source can be matched with the coordinate system for image conversion, and the calculation load can be reduced.

  In the present embodiment, the microphone unit 202 is provided on the pedestal 203, but each participant 2 can also detect the direction of the sound source by owning the microphones 221 having wireless communication means. Is possible. For example, units that transmit radio waves to a plurality of known positions in a conference room can be installed, and the position of each microphone 221 can be detected based on the principle of triangulation from the signal intensity and time difference of the radio waves that have reached the microphone 221. At this time, the direction of the microphone 221 in which the largest signal amplitude is obtained can be detected as the speaker direction. Here, a communication technique such as Bluetooth can be used as the wireless communication means.

  It should be noted that the conference image reproduction device 300 can realize its function by a personal computer. In this case, the function can be realized by storing software for realizing each functional unit in a hard disk and appropriately executing a processing program.

  As described above, the conference recording / playback system according to the first embodiment can capture all the conference participants at once with a simple configuration using a simple optical system using a hyperboloid mirror. In addition, the conference can be reproduced by recording this content. With regard to playback, by transforming and outputting necessary parts, it is possible to reproduce highly realistic conference content centered on the speaker. In particular, since the content of the conference is recorded in all directions, it is possible to look back on the conference scene under the conditions preferred by the user.

[Embodiment 2]
In the second embodiment, a conference image recording / reproducing system that records a wide area image after transforming it into a panoramic image will be described. In the second embodiment, the same components as those in the first embodiment are denoted by the same reference numerals, and the description thereof is omitted. Hereinafter, the appearance configuration, hardware configuration, functional configuration, and processing flow of the conference recording / playback system 1900 will be described in order.

(External configuration of conference recording / playback system 1900)
FIG. 19 is a diagram illustrating an example of an external configuration of the image recording / playback system according to the second embodiment. The image recording / playback system 1900 includes a cross button 1901, an enter button 1902, an image / audio output terminal 1903, and a medium insertion slot 1904. In the conference recording / playback system 100 according to the first embodiment, the conference image transmission device 200 that captures images and sounds and the conference image playback device 300 that records and processes the moving images are separated. The conference recording / playback system 1900 according to mode 2 performs input / storage / processing / playback / output of video and audio in a single casing.

  First, the above-described parts appearing on the exterior will be described. A cross button 1901 is used to move a menu or pointer displayed on a screen (not shown). For example, enter a meeting name and use it to create a meeting file. When a plurality of conferences are recorded, it is also used when selecting a conference file name to be reproduced. In addition, it is also used when correcting the sound source direction, such as inputting the elevation angle of the speaker.

  The decision button 1902 makes various decisions. For example, it is also used when the selection target by the cross button 1901 is determined. Note that this determination button can also be used as a multi-function button by assigning power on / off and playback stop.

  The image / audio output terminal 1903 outputs data processed by the conference recording / playback system 1900, that is, an image signal of an image obtained by cutting out a predetermined speaker without distortion and an audio signal associated with the image. The data format may be the MPEG format or RealAudio format as described above, but here, a signal format transmitted and received by a VIDEO terminal (VHF / UHF terminal) provided in a normal television is adopted. With such a general-purpose signal format, it is possible to reproduce a conference on a normal television without using a special control circuit.

  The medium insertion slot 1904 is a slot for inserting a recording medium for recording the conference. In the first embodiment, a DVD-RAM or the like is assumed. However, here, a PCMCIA socket is employed, and a high-density and large-capacity card type HDD is inserted. Such a configuration makes it possible to reduce the size of the apparatus. Depending on circumstances, a slot into which a DVD-RW or a DAT tape can be inserted may be used. In the case where a mechanical drive unit is provided, a mute structure is adopted so that the conference recording / playback system 1900 does not pick up mechanical noise.

(Conference recording and playback system 1900: hardware configuration)
Next, the hardware configuration of the conference recording / playback system 1900 will be described. FIG. 20 is an explanatory diagram showing an example of a hardware configuration of the conference image recording / playback system according to the second embodiment. In addition to the CPU 301, the conference recording / playback system 1900 includes a RAM 2001, a ROM 2002, an operation unit 2003, an output I / F 2004, a camera unit 2005, a microphone unit 2006, and a removable media unit 2007. The camera unit 2005 is a notation that conveniently shows the camera unit 201 including the optical system shown in FIG. 19, and the microphone unit 2006 is also a microphone unit 202 including the microphone 221 shown in FIG. Is a notation for convenience.

  The RAM 2001 is used as a work area for the CPU 301 and also as a storage area for processing programs stored in the HDD 303 and other control programs (for example, OS). The ROM 2002 stores universal control information and coefficients. For example, the conversion table (correspondence table) shown in FIG. 17 may be stored.

  The operation unit 2003 includes a cross button 1901 and an enter button 1902. The output I / F 2004 includes an image / audio output terminal 1903, a video card, and a video memory, and transmits an image signal and an audio signal to a video input terminal of a television (not shown). The removable media unit 2007 performs drive control of writing / reading of a PCMCIA type large-capacity HDD inserted in the medium insertion slot 1904.

(Contents of conference recording / playback system 1900: functional configuration)
Next, a functional configuration of the conference recording / playback system 1900 will be described. FIG. 21 is an explanatory diagram showing an example of the functional configuration of the conference recording / playback system 1900. The conference recording / playback system 1900 includes, as its functional configuration, a wide-angle image expansion unit 2101 and an image extraction unit 2102 in addition to the functional units described in FIG.

(Conference Recording and Playback System 1900: Contents of Wide Angle Image Expansion Unit 2101)
The wide-angle image development unit 2101 transforms the donut image into a panoramic image. In the conference recording / playback system 100 according to the first embodiment, the image is deformed during playback (see the image deforming unit 805 in FIG. 8). However, in the conference recording / playback system 1900 according to the second embodiment, the image is deformed during recording. To do. In other words, the conference recording / playback system 1900 expands the wide-angle image into a panoramic image before recording in the recording unit 804, and the panoramic image is recorded. Since the expansion process may be calculated using Expressions (4) to (8), description thereof is omitted.

  Note that when the processing capability of the CPU 301 (see FIG. 20) is low, calculation time is required for the deformation processing of the image data, so the wide-angle image and the panoramic image may be associated with each other by referring to a predetermined conversion table. By using such a table, it is possible to perform high-speed image deformation while reducing the processing load.

  The wide-angle image expansion unit 2101 can realize its function by, for example, the CPU 301 and a wide-angle image expansion program stored in the HDD 303. Note that both the conference recording / playback system 1900 and the conference recording / playback system 100 have 100% of the original information, so that it is possible to play back a scene in a desired sound source direction at any time.

(Conference recording and playback system 1900: contents of image extraction unit 2102)
The image extraction unit 2102 cuts out (extracts) a predetermined image area corresponding to the sound source direction from the panoramic image recorded in the recording unit 804, and outputs it to the image / audio output unit 808. For example, when the conference participant A (see FIG. 6) is speaking, the portion corresponding to the participant A is extracted based on the video data developed and recorded as shown in FIG. 14 based on the sound source direction data. Extract. Hereinafter, this cut out image is referred to as a partial image. FIG. 22 is an explanatory diagram showing an example of image extraction. As illustrated, the image extraction unit 2102 generates partial image data in which only the participant A is shown.

  Here, the procedure of image extraction will be described. FIG. 23 is an explanatory diagram illustrating a method of generating partial image data by the image extraction unit 2102 according to the second embodiment. First, a range of angles extracted as partial image data is set in advance. It is assumed that the range of this angle is Δθ in the azimuth angle direction and Δφ in the elevation angle direction. Next, the azimuth angle θ and elevation angle φ detected by the sound source direction detection unit 803 are read. Finally, in the panoramic image data input from the recording unit 804, areas corresponding to the azimuth angle θ and the elevation angle φ ((θ−Δθ / 2, φ−Δφ / 2), (θ + Δθ / 2, φ−Δφ / 2)). , (Θ−Δθ / 2, φ + Δφ / 2), (region surrounded by (θ + Δθ / 2, φ + Δφ / 2)), partial video data is generated.

  Note that the image extraction unit 2102 may extract an image directly from the donut image depending on the mode of use. At this time, as shown in FIG. 15, the partial image data is cut out from the donut image data by accessing only a rectangular area of Δθ × Δφ centered on (θ, φ) in the coordinate conversion table. And deform. Further, when the image deformation unit 805 is included as in the first embodiment, a rectangular area of Δθ × Δφ centered on (θ, φ) is directly applied to the panoramic image generated by the image deformation unit 805. You may make it extract.

  The image extraction unit 2102 can realize its function by, for example, the CPU 301 of the conference recording / playback system 1900 shown in FIG. 20 and the image extraction program stored in the HDD 303. In the second embodiment, the image / audio output unit 808 includes the image (moving image data) output from the image extraction unit 2102 and the audio recorded (input) simultaneously when the image is captured (input). Are output in association with each other. That is, the image and sound are output in synchronization. Depending on the processing speed of the CPU 301 (see FIG. 20), there is a time lag between audio and image, so the audio / video output unit 808 reproduces a natural meeting by synchronizing the image and audio.

(Conference recording and playback system 1900: flow of processing)
Next, the process flow of the conference recording / playback system 1900 will be described. FIG. 24 is an explanatory diagram showing an example of the processing flow of the conference recording / playback system 100. First, the conference recording / playback system 100 starts recording by pressing a recording start button (not shown) (step S2401). After this start operation, wide-angle images with the vertical direction as the center or axis are sequentially input from the camera unit 201, and sound is sequentially input from the microphone unit 202 (step S2402). As for the sound input from the microphone unit 202, the sound source direction is detected at any time using the above-described microphone set.

  Next, the wide-angle image (doughnut image) input from the camera unit 201 is sequentially transformed into a panoramic image (step S2403). The panoramic image, the sound input from the microphone unit 202, and the detected sound source direction are recorded (step S2404). For recording, a recording time, a file name (meeting name), etc. are given as appropriate for later playback.

  When the conference ends and a recording end button (not shown) is pressed, the recording is ended (step S2405). Through the above steps, it is possible to record the entire surrounding image, that is, the contents including all the states of the meeting. Since the stored image includes the entire surrounding image, it can be edited later (extraction of the image region, correction of the sound source direction, etc.) as the user prefers.

  Next, playback of a recorded conference will be described. The conference recording / playback system 1900 starts playback when a playback start button (not shown) is pressed (step S2406). When a plurality of conferences are recorded on the recording medium (PCMCIA type hard disk), an index is displayed to allow the user to select which conference to play.

  The speaker position is determined using the flesh-colored part in the sound source direction in the recorded panoramic image as a clue (step S2407), and the image of the area to be displayed is extracted (step S2408). Finally, the extracted image is output together with sound (step S2409). By taking such steps, the conference can be efficiently reproduced while maintaining a sense of reality.

  As described above, the conference recording / playback system according to the second embodiment can capture all the conference participants at once with a simple configuration by a simple optical system using a hyperboloid mirror. In addition, the conference can be reproduced by recording this content. In addition, since a donut image developed into a panoramic image is recorded, it is possible to construct a system with less load during reproduction.

[Embodiment 3]
In the third embodiment, a mode in which the camera unit and the microphone unit in the conference recording / playback system are different from those in the first or second embodiment will be described. FIG. 25 is an explanatory diagram showing an example of an external configuration of an apparatus including the camera unit according to the third embodiment. As is apparent from the figure, the camera unit 2501 of the conference recording / playback system 2500 includes a mirror body 2502 having a conical shape instead of a hyperboloid mirror. Although a description of the conversion formula from the donut image to the panoramic image is omitted, focusing is performed on the surface of the CCD 214 by appropriately arranging a lens like the lens 212 shown in FIG. Depending on the mode of use, a mirror body having a paraboloid may be used.

  In the above example, the reflecting mirror (the hyperboloid mirror 211, the mirror body 2502 having a conical shape or the mirror body having a paraboloid) has a single configuration. May be used. FIG. 26 is an external configuration diagram of a camera unit configured to capture a donut image using two reflecting mirrors. The camera unit 2600 includes a first reflecting mirror 2601 composed of a parabolic mirror or a hyperboloid mirror, and a second reflecting mirror 2602 that deflects the reflected light reflected by the first reflecting mirror in the CCD direction. Have. Note that the top of the first reflecting mirror 2601 has a hole in order to capture the reflected light from the second reflecting mirror.

  Next, the microphone unit will be described. FIG. 27 is an explanatory diagram illustrating the relationship between the microphone unit and the sound source direction according to the third embodiment. The microphone unit 202 according to the first embodiment and the second embodiment detects a sound source direction based on a difference in arrival time of sound using a non-directional microphone 221. The microphone unit 2701 of Embodiment 3 has four microphones 2702 having directivity, and determines the sound source direction based on the intensity of the sound. For convenience, the four microphones 2702 are referred to as microphones 1 to 4.

  Assume that the voice intensity is 20 for microphone 1, 30 for microphone 2, 20 for microphone 3, and 5 for microphone 4. In this case, it is determined that there is a sound source in the direction of the microphone 2. Comparing the intensities of the microphone 1 and the microphone 3, since both values are the same 20, the sound source direction is finally determined to be the microphone 2 direction (direction shown as θ = 45 ° in the figure).

  Another example will be described. Assume that the sound intensity is 15 for microphone 1, 30 for microphone 2, 25 for microphone 3, and 5 for microphone 4. In this case, it is initially determined that there is a sound source in the direction of the microphone 2. Comparing the intensities of the microphone 1 and the microphone 3, since the intensity of the microphone 3 is greater than the microphone 1, the direction of the sound source is slightly moved from the direction of the microphone 2 to the direction of the microphone 3 (the direction indicated as θ = 30 ° in the figure). ). The amount of movement in this direction may be determined in advance according to the characteristics of the directional microphone. As described above, when the directional microphone 221 is used, it is not necessary to perform calculations such as Expressions (1) to (3), and thus the load on the processor can be reduced.

[Embodiment 4]
In the fourth embodiment, a versatile conference image transmission device and a conference image reproduction device will be described. Here, versatile means that even if there are multiple types of conference image transmission devices and conference image playback devices depending on the configuration of the mirror body that captures wide-angle images and the type of microphone, the conference can be recorded or played back in any combination. Say. In the fourth embodiment, the same components as those in the first to third embodiments are denoted by the same reference numerals unless otherwise specified, and the description thereof is omitted.

  The conference recording / playback system 2800 of the fourth embodiment includes a conference image transmission device 2801 and a conference image playback device 2802. FIG. 28 is a diagram illustrating functional blocks of the conference image transmission device 2801 and the conference image playback device 2802 according to the fourth embodiment. The conference image transmission device 2801 includes a wide-angle image input unit 2811, a voice input unit 2812, a sound source direction detection unit 2813, an elevation angle setting unit 2814, and a data transmission unit 2815 as functional configurations.

  The wide-angle image input unit 2811 takes in a wide-angle image with the vertical direction as the center or axis, and outputs the image data to the data transmission unit 2815. For the input of the wide-angle image, the hyperboloid mirror 211 shown in the first embodiment may be used, and either the conical mirror body 2502 or the parabolic reflector shown in the third embodiment is used. Also good.

  The voice input unit 2812 inputs voice and converts it into an electrical signal (voice data), and sends the voice data to the sound source direction detection unit 2813 and the data transmission unit 2815. For the input of sound, the omnidirectional microphone 221 described in Embodiment 1 may be employed, or the directional microphone 2702 described in Embodiment 3 may be used. The sound source direction detection unit 2813 detects the sound source direction based on the time difference or intensity of the sound input from the sound input unit 2812. Since the principle of detecting the direction of the sound source has already been described, a description thereof will be omitted.

  The elevation angle setting unit 2814 sets an elevation angle that is the height direction of the speaker. As described with reference to FIG. 10, the sound source direction detection unit generally has a large error in the elevation angle direction. Therefore, the elevation angle setting unit 2814 sets the elevation angle from the plane on which the conference image transmission device 2801 is installed. The setting method may be based on, for example, detection of a speaker's image data (skin color data) in addition to the direct setting of the angle ψ using a numeric keypad.

  The data sending unit 2815 sends data on the sound source direction including the wide-angle image, the sound, and the elevation angle to a predetermined data storage unit. Here, the image is transmitted to the conference image reproduction device 2802. In the first to third embodiments, data transmission by wire has been described. However, the present invention is not limited to this, and data may be transmitted wirelessly. Various methods can be adopted as a method for transmitting wireless data. For example, a wireless I / F such as IrDA or BlueTooth can be adopted.

  Next, the conference image playback device 2802 will be described. The conference image reproduction device 2802 includes a data input unit 2821, a recording unit 2822, an image transformation unit 2823, an area determination unit 2824, and an image / audio output unit 2825 as functional configurations. In addition, the conference image reproduction device 2802 includes a direction correction unit 806 and an area fixing unit 807. In the following, although each functional unit will be described, the conference image playback device 2802 can realize its functions by a personal computer. In this case, the function can be realized by storing software for realizing each functional unit in a hard disk and appropriately executing a processing program.

  The data input unit 2821 inputs moving image data obtained by capturing a wide-angle image, audio data synchronized with the moving image data, and data related to the sound source direction from a predetermined data transmission source. Here, the predetermined data transmission source is the conference image transmission device 2801. However, the video data, audio data, and sound source direction data depend on the transmission source device as long as the data type can be recognized. do not do. The type of data can be determined by the file extension or the file header. The wide-angle image is assumed to be a donut image here, but may be a panoramic image. This type is also determined by the extension and header. The data input unit 2821 can employ a wireless I / F such as IrDA or BlueTooth, for example.

  The recording unit 2822 records data on the sound source direction including the moving image data, audio data, and elevation angle of the wide-angle image input by the data input unit 2821. There are various recording methods, but as described above, the MPEG format or the RealAudio format can be adopted.

  The image deformation unit 2823 deforms the wide-angle image so as to be a rectangular output image. The conversion is designed so that the CCD 214 is in focus, and the image captured by the CCD 214 is always a donut image. Therefore, as described above, the image is deformed by referring to a correspondence table (not shown) between the donut image and the panoramic image. At this time, since the final output image is a portion including the speaker, the image deformation unit 2803 performs image deformation only on the image region portion determined by the region determination unit 2824.

  The area determination unit 2824 determines an area to be reproduced based on the data regarding the sound source direction including the elevation angle recorded in the recording unit 2822. As described in the first embodiment, the speaker position determination unit 809 may be used together to improve the accuracy of detecting the speaker position. The image / sound output unit 2825 associates and outputs the image (moving image data) output from the image deformation unit 2823 and the sound simultaneously recorded (input) when this image is captured (input).

  Next, the processing flow of the conference image transmission device 2801 will be described. FIG. 29 is a flowchart illustrating an example of a processing flow of the conference image transmission apparatus 2801 according to the fourth embodiment. First, the system of the conference image transmission device 2801 is activated by the user, and starts an operation of taking in data (image data and audio data) (step S2901). Next, it is determined whether or not capture stop (recording) has been instructed (step S2902). If there is an instruction (step S2902: YES), capture is terminated.

  Unless there is an instruction to stop capturing (step S2902: NO), the image data transmitted from the CCD 214 and the audio data transmitted from the microphone array are continuously input (step S2903). When a certain amount of audio data, for example, the same number of samples as the size N of the correlation window C shown in Expression (2) is input, the sound source direction is detected and sound source direction data is generated (step S2904). The conference image transmission device 2801 sequentially outputs image data, audio data, and sound source direction data to a predetermined transmission destination, for example, a PC (step S2905). Thereafter, the operations from step S2902 to step S2904 are sequentially repeated, and data is transmitted until the user instructs to stop recording.

  Next, the processing flow of the conference image playback device 2802 will be described. FIG. 30 is a flowchart illustrating an example of a processing flow of the conference image reproduction device 2802 according to the fourth embodiment. First, the system of the conference image playback device 2802 is activated by the user (step S3001). Next, an image to be reproduced is selected according to an image displayed on a display (television) (not shown) (step S3002). FIG. 31 is a diagram showing an example of a screen configuration for selecting an image to be reproduced. As shown in the figure, the meeting files are named Meeting1 and Meeting2, and each file is composed of image data (MPEG-2Video), audio data (MPEG Audio), and sound source direction data (TEXT). I understand that.

  Next, the conference image playback device 2802 reads wide-angle image data, audio data, and sound source direction data, and starts a playback operation (step S3003). Subsequently, the conference image playback device 2802 determines whether or not there is an instruction to stop playback (step S3004), and stops playback when instructed. On the other hand, if there is no instruction to stop reproduction (step S3004: NO), it is determined whether or not the time for inquiring the sound source direction data has been reached (step S3005). The time when the sound source direction data is inquired refers to, for example, the time when the sound source direction changes as shown in FIG.

  When the inquiry time is reached (step S3005: Yes), the sound source direction data is accessed, and new sound source directions (values of azimuth angle θ and elevation angle φ) are acquired (step S3006). Subsequently, the conference image reproduction device 2802 extracts the partial image data corresponding to the azimuth angle θ and the elevation angle φ acquired in step S3006 (step S3007), and outputs the extracted partial image data and audio in synchronization. (Reproduction) is performed (step S3008). If the inquiry time has not been reached in step S3005 (step S3005: NO), the currently reproduced partial image data is continued and reproduced (step S3009).

  As described above, the fourth embodiment is efficient while maintaining a sense of reality, even if the conference image transmission device and the conference image playback device are configured separately, such as a video camera and a video deck. Can be reproduced.

[Embodiment 5]
In the fifth embodiment, another configuration example of a versatile conference image transmission device and conference image reproduction device will be described. In the fifth embodiment, the same components as those in the first to fourth embodiments are denoted by the same reference numerals unless otherwise specified, and the description thereof is omitted.

  FIG. 32 is a functional block diagram of the conference image transmission device and the conference recording / playback device according to the fifth embodiment. The conference recording / playback system 3200 according to the fifth embodiment includes a conference image transmission device 3201 and a conference image playback device 3202. The conference image transmission apparatus 3201 has, as its functional configuration, a wide-angle image input unit 3211, an audio input unit 3212, a sound source direction detection unit 3213, a wide-angle image expansion unit 3214, an image extraction unit 3215, and a data transmission unit 3216. And having.

  The wide-angle image input unit 3211 takes in a wide-angle image with the vertical direction as the center or axis, and outputs the image data to the wide-angle image development unit 3214. As in the fourth embodiment, the wide-angle image may be input using any one of the hyperboloid mirror 211, the conical mirror body 2502, and the parabolic reflector. The voice input unit 3212 inputs voice, converts it into an electrical signal (voice data), and sends the voice data to the sound source direction detection unit 3213 and the data transmission unit 3216. The voice input unit 3212 may use a directional microphone or an omnidirectional microphone. The sound source direction detection unit 2813 detects the sound source direction based on the time difference or intensity of the sound input from the sound input unit 3212 and outputs the sound source direction to the image extraction unit 3215 and the data transmission unit 3216.

  The wide-angle image expansion unit 3214 transforms the donut image into a panoramic image and outputs the panoramic image to the image extraction unit 3215 and the data transmission unit 3216. The image extraction unit 3215 extracts an image of a predetermined portion in the speaker direction based on the sound source direction output from the sound source direction detection unit 3213 from the panoramic image output from the wide-angle image development unit 3214. The data sending unit 3216 sends the panoramic image (entire area), the extracted image (partial image in the speaker direction), the sound, and the data related to the sound source direction to predetermined data storage means. Here, it is sent to the conference image playback device 3202.

  Next, the conference image reproduction device 3202 will be described. The conference image reproduction device 3202 includes a data input unit 3221, a recording unit 3222, an image / audio output unit 3223, and a direction correction unit 3224 as functional configurations. In the following, although each functional unit will be described, the conference image reproduction device 3202 can realize its function by a personal computer. In this case, the function can be realized by storing software for realizing each functional unit in a hard disk and appropriately executing a processing program.

  The data input unit 3221 inputs moving image data obtained by capturing a wide-angle image from a predetermined data transmission source, audio data synchronized with the moving image data, and data related to the sound source direction. Here, the predetermined data transmission source is the conference image transmission device 3201, but the video data (entire image and partial image), audio data, and data related to the sound source direction are in a format that can recognize the type of the data. It does not depend on the source device.

  The recording unit 3222 records panoramic images input by the data input unit 3221 and moving image data, audio data, and sound source direction data of partial images in the speaker direction. There are various recording methods, but as described above, the MPEG format or the RealAudio format can be adopted. The image / sound output unit 3223 associates and outputs the partial image (moving image data) in the direction of the speaker output from the recording unit 3222 and the sound simultaneously recorded (input) when the image is captured (input). .

  However, there is a case where a partial image in the direction of the speaker is not properly extracted, or an image other than the speaker, for example, a speaker including two adjacent people or a whiteboard may be displayed. Therefore, in order to satisfy such a request, the conference image reproduction device 3202 includes a direction correction unit 3224. The direction correction unit 3224 corrects the direction corresponding to the sound source direction, and allows the user to select a desired voice direction. The direction selection by the user will be described later.

  Next, the processing flow of the conference image transmission apparatus 3201 will be described. FIG. 33 is a flowchart illustrating an example of a processing flow of the conference image transmission apparatus 3201 according to the fifth embodiment. First, the system of the conference image transmission device 3201 is activated by the user, and starts an operation of taking in data (image data and audio data) (step S3301). Next, it is determined whether or not a capture stop (record stop) is instructed (step S3302). If there is an instruction (step S3302: YES), the capture is terminated.

  Unless there is an instruction to stop capturing (step S3302: NO), image data transmitted from the CCD 214 and audio data transmitted from the microphone array are input (step S3303). When a certain amount of audio data, for example, the same number of samples as the correlation window size N is input, the sound source direction is detected, and sound source direction data is sequentially generated (step S3304). The conference image sending apparatus 3201 sequentially develops the wide-angle image (donut image) input in step S3303 into a panoramic image (step S3305), and generates partial image data in the sound source direction in the developed panoramic image (step S3306). ).

  The conference image transmission device 2801 sequentially outputs panoramic image data, partial image data, audio data, and sound source direction data to a predetermined transmission destination, for example, a PC (step S3307). Thereafter, the operations from step S3302 to step S3307 are sequentially repeated, and data is transmitted until the user instructs to stop recording.

  Next, the processing flow of the conference image reproduction device 3302 will be described. FIG. 34 is a flowchart illustrating an example of a processing flow of the conference image reproduction device 3302 according to the fifth embodiment. First, the system of the conference image reproduction device 3302 is activated by the user (step S3401). Next, an image to be reproduced is selected according to an image displayed on a display (television) (not shown) (step S3402). FIG. 35 is a diagram showing an example of a screen configuration for selecting an image to be reproduced. As shown in the figure, the meeting files are named Meeting1 and Meeting2, and each file includes panoramic image data (MPEG-2Video), audio data (MPEG Audio), sound source direction data (TEXT), and It can be seen that the image data is composed of partial image data (MPEG-2Video).

  Next, the conference image playback device 3302 reads the partial image data and audio data, and starts the playback operation (step S3403). Subsequently, the conference image playback device 2802 determines whether or not there is an instruction to stop playback (step S3404), and stops playback when instructed. On the other hand, if there is no instruction to stop playback (step S3404: NO), it is determined whether or not there is an input from the direction correcting unit 3224 (step S3405). If the direction has been corrected (step S3405: YES), the designated partial image is extracted from the panoramic image and output (reproduced) together with the sound (step S3406).

  On the other hand, if there is no input from the direction correcting unit 3224 (step S3405: NO), the partial image data is output as it is (step S3407). Note that the conference image playback device 3202 sequentially outputs the partial images extracted in advance, so that the Meeting1_pv shown in FIG. 35 may be played back unless the direction is corrected.

  Next, a configuration example of an image output from the conference image recording / playback apparatus 3200 will be described. FIG. 36 is an explanatory diagram showing a configuration example (screen example) of an image output from the conference image recording / playback apparatus 3200. The screen includes not only a speaker direction image 3601 but also user interfaces such as a mode switching unit 3602, a direction instruction operation unit 3603, and a reproduction operation instruction unit 3604.

  Next, each user interface will be described. The mode switching unit 3602 switches whether to reproduce a specific partial image in the wide-angle image data. As shown in FIG. 36, the operation mode can be switched using a radio button. That is, when the radio button drawn as “AUTO” is selected, the partial image that is processed and extracted based on the sound source direction data and recorded in the recording unit 3222 is automatically reproduced. On the other hand, when a radio button drawn with “MANUAL” is selected, a donut image 3605 is displayed as shown in FIG. 37, and a “manual” can be manually selected for a portion to be reproduced by a user operation. Transition to “switching mode”.

  In the manual switching mode, the pointer 3607 is moved by a direction instruction operation unit 3603 that is four buttons on which up, down, left and right arrows are drawn. By moving the pointer 3607, the drawing direction of the partial image data can be moved, and an image with the extracted portion changed can be output as shown in FIG. By this operation, for example, the drawing contents on the whiteboard can be appropriately output. Note that the screen configuration is not limited to FIGS. 36 to 38, and for example, as shown in FIG. 39, a 4-split screen may be output simultaneously. Here, reference numeral 3901 is a GUI for switching the output between the four divided screens and one of them.

  On the other hand, as shown in the figure, the playback operation instruction unit 3604 has a GUI to which playback, stop, pause, fast forward, and rewind functions are assigned from the left. Realize corresponding operation. Although described as software processing here, buttons may be arranged in hardware on the conference image playback device 3302 side, or a remote controller may be separately provided to enhance convenience.

  By providing such a manual switching mode and a 4-split screen, for example, when a scene in which one participant keeps talking for a long time is played back later, it is more intermittent than playing only the video showing the speaker. In general, playing participants other than the speaker gives a sense of realism without feeling bored. As described above, the mode switching unit 3602 and the direction instruction operation unit 3603 are particularly useful when it is desired to view a video other than the portion specified by the direction data such as the facial expressions of the participants other than the speaker.

  The conference image transmission device 3201 according to the fifth embodiment transmits both the panoramic image (entire area) and the extracted image (partial image in the speaker direction). However, depending on the usage, only the partial image is transmitted. It may be sent out. At this time, the sound source direction data does not need to be transmitted to the conference image reproduction device 3202 because no image extraction or sound source direction determination is performed on the conference image reproduction device 3202 side.

  As described above, in the fifth embodiment, as in the fourth embodiment, even if the conference image transmission device and the conference image playback device are configured separately and independently, the content of the conference is efficiently maintained while maintaining a sense of reality. Can be reproduced.

  In the examples so far, the system for recording a conference has been mainly described. However, the present invention is not limited to this application, and can be used as a security camera by being installed on the top, for example. Moreover, it can be used for the purpose of examining the ecology of nocturnal animals. In this case, a high sensitivity CCD is used.

100, 1900, 2500, 2800, 3200 Conference recording / playback system 200, 2801, 3201 Conference image sending device 201, 2005, 2501, 2600 Camera unit 202, 2006, 2701 Microphone unit 203, 2032 Base 204 Transparent glass 211 Hyperboloid mirror 212 Lens 213 Aperture 221, 2702 Microphone 300, 2802, 3202, 3302 Conference image playback device 307 Large capacity recording device 801, 2821, 3211 Wide angle image input unit 802, 2812, 3212 Audio input unit 803, 3213 Sound source direction detection unit 804, 2813 , 2822, 3222 Recording unit 805, 2803, 2823 Image transformation unit 806, 3224 Direction modification unit 807, 2824 Area fixing unit 808, 2825, 3223 Image / audio output unit 809 Speaker position determination unit 810 Region determination unit 1901 Cross button 1902 Determination button 1903 Video / audio output terminal 1904 Medium insertion slot 2003 Operation unit 2007 Removable media unit 2101, 3214 Wide-angle image expansion unit 2102, 3215 Image extraction unit 2502 Mirror body 2601 1 1 reflection mirror 2602 second reflection mirror 2814 elevation angle setting unit 2815, 3216 data sending unit 2821, 3221 data input unit 3602 mode switching unit 3603 direction instruction operation unit 3604 reproduction operation instruction unit

Japanese Patent Laid-Open No. 5-12289 Japanese Patent Laid-Open No. 11-331827

Claims (6)

An image input unit for inputting a wide-angle image;
A plurality of voice input sections for inputting voice;
A sound source direction detecting unit for detecting a sound source direction from the sound;
An image transformation unit for generating a panoramic image including an area corresponding to a speaker position determined based on the sound source direction from the wide-angle image;
Have
A conference image reproduction system , wherein an enlargement / reduction process is performed on an area of a speaker portion included in the panoramic image so that a speaker has an appropriate size . The image input unit
A specular body having a conical shape;
An image sensor for capturing an image;
Have
2. The conference image reproduction system according to claim 1, wherein a wide-angle image including at least the entire periphery overlooking the horizontal plane is input. The sound source direction detector
The conference image reproduction system according to claim 1, wherein a direction of a sound source is detected based on a difference in arrival time of sound using an omnidirectional microphone. further,
A determination unit that determines a speaker position from the image based on a color distribution or a movement amount of the wide-angle image;
An area determination unit that determines an area of the speaker part determined by the determination unit;
The image transformation unit further generates a panoramic image including the speaker portion determined by the region determination unit;
The conference image reproduction system according to claim 1, 2, or 3.   The conference image reproduction system according to claim 1, further comprising an image audio output unit that associates the panoramic image with the audio. An image input process for inputting a wide-angle image;
A plurality of voice input processes for inputting voice;
A sound source direction detecting step for detecting a sound source direction from the sound;
An image transformation step for generating, as a panoramic image, an area corresponding to a speaker position determined based on the sound source direction from the wide-angle image;
Including
A conference image reproduction method, wherein an enlargement or reduction process is performed on an area of a speaker part included in the panoramic image so that the speaker has an appropriate size .

Images