JP6484402B2 - Image surveillance system - Google Patents

Description

  The present invention relates to an image monitoring system.

Conventionally, there exists a system that determines the position of an object using two stereo cameras.
Further, the entire monitoring space is monitored at a constant cycle by switching and displaying the images of a plurality of cameras manually or automatically.

  As another prior art, a technique for synthesizing a human image captured by a visible light camera and a far-infrared camera is disclosed. Specifically, by comparing the far-infrared image acquired from the far-infrared camera and the visible light image acquired from the visible light camera, and correcting the reflected portion based on the comparison result, a dark region with a small amount of incident light, That is, even in a non-visible region that is difficult to capture directly with a visible light camera, it is possible to detect a target object without missing it (see, for example, Patent Document 1).

JP 2011-243037 A

An object of the present invention is to specify the same person among persons reflected in images taken by a plurality of cameras.
Another object of the present invention is to synthesize and display a face image of a person determined to be the same among images taken by a plurality of cameras as an image closer to the front.

An image monitoring system of the present invention is an image monitoring system having a first camera, a second camera disposed opposite to the first camera, and an image processing device, wherein the image processing device is a first camera. The horizontal distance information and the vertical distance information of the first person on the image photographed by the camera from the center of the image to the center of the face of the first person, and the first person on the image photographed by the first camera the height information of the person, the height information of the second person on the image captured by the second camera, on the basis of, the first person, the from the center of the image captured by the second camera first A horizontal distance and a vertical distance to the center of the face of the second person are calculated, and the first person specifies the second person.

The image monitoring system of the present invention is the image monitoring system described above, wherein the image monitoring system further includes a system management server, and the system management server is configured such that when the face of the first person is other than the front, select close face image in front in the face image of a person was taken by multiple cameras, the selected face image is a face image, characterized in that combined with the image of the first person other than the front .

  Furthermore, the image monitoring system of the present invention is an image monitoring system having a plurality of cameras and an operation terminal. The operation terminal allocates a monitoring space for each camera, and the allocated monitoring space has a blind spot due to an obstacle. When a camera image is output to a monitor, the image of another camera that captures a blind spot space is combined with the blind spot portion.

According to the present invention, it is possible to identify the same person among the persons shown in images taken by a plurality of cameras.
Further, it is possible to synthesize and display a face image of a person determined to be the same among images taken by a plurality of cameras as an image closer to the front.

It is a block diagram for demonstrating the image monitoring system which is one Example of this invention. It is the installation figure and display screen figure of the camera arrange | positioned facing. It is a figure for demonstrating one Example of this invention. It is a figure for demonstrating the display content of the monitor A and the monitor B. FIG. It is the figure which faced the front of the person's face part, and the figure which faced diagonally. It is a figure of the direction of a person's face in the surveillance space which has arrange | positioned four cameras which are one Example of this invention, and a corresponding camera. It is a figure for demonstrating the image | video imaged with the four cameras for every continuous time which is one Example of this invention. It is a top view for demonstrating the monitoring space. It is a figure for demonstrating the camera allocation area on a top view. It is a figure for demonstrating the monitoring screen of the operating terminal which is one Example of this invention. It is a figure for demonstrating the left-right monitoring camera using the mirror which is another Example of this invention. It is another block block diagram of the image monitoring system which is one Example of this invention.

Hereinafter, an embodiment of the present invention will be described with reference to the drawings.
FIG. 1 is a block diagram for explaining an image monitoring system according to an embodiment of the present invention.
1, the image monitoring system includes cameras 101-1 to 101-n, image processing apparatuses 102-1 to 102-n (referred to as image processing apparatuses 102 in the representative case), and IP encoders 103-1 to 103-n. A network 104, a system management server 105, and operation terminals 106-1 to 106-n (referred to as operation terminals 106 in the representative case).
As illustrated in FIG. 12, the image processing apparatus 102 may be configured to be directly connected to the network 104 without being interposed between the camera 101 and the IP encoder 103.

  The images captured by the cameras 101-1 to 101-n are subjected to moving object detection by the image processing devices 102-1 to 102-n, and a person is identified based on the size and shape of the detected object. Specify the size of. The video is digitized by the IP encoders 103-1 to 103-n and distributed to the network 104.

The system management server 105 manages the entire monitoring space with three-dimensional global coordinates, and registers camera position information and structure information in the monitoring space in advance on the database. The system management server 105 manages the position coordinate information of the detected person in real time, and plots the person position on the monitoring plan view.
Further, the system management server 105 determines whether or not they are the same person based on the position information and size information of persons sent from a plurality of image processing apparatuses. In the case of the same person, all face images of the person are cut out from the images captured by a plurality of cameras and managed as face images of the same person. The system management server 105 has a WEB server function that provides a monitoring screen for performing a monitoring operation from the browser to the operation terminals 106-1 to 106-n.

The operation terminals 106-1 to 106-n browse the video on which the person is displayed on the above-described monitoring operation screen. In response to a request from the supervisor, the system management server provides a face image of the person facing the front, or provides a plan view on which the position information and locus of the person are plotted. Alternatively, it is possible to automatically provide an image in which the person faces the front, or to provide an image in which the person is not hidden in the blind spot.
The operation terminals 106-1 to 106-n may be PCs (Personal Computers).

FIG. 2 is an installation view and a display screen view of the cameras arranged opposite to each other.
Hereinafter, a method of identifying the same person from videos captured by two cameras installed opposite to each other and developing position information in global coordinates will be described.
2A, 211 is the camera A, 212 is the vertical angle of view of the camera A, 213 is the installation height ZA of the camera A, 221 is the camera B, 222 is the vertical angle of view of the camera B, 223 Is the installation height ZB of the camera B, 231 is the ground, and 241 is an object (person).
Here, ZA = ZB.
2B is a photographed image of the camera A, 251 is a monitor A that displays the photographed image captured by the camera A, and 252 is a person on the screen of the camera A.
FIG. 2C is a photographed image of the camera B, 261 is a monitor B that displays the photographed image captured by the camera B, and 262 is a person on the screen of the camera B.

FIG. 3 is a diagram for explaining an embodiment of the present invention.
On the xyz coordinate axis, camera A is the origin, and the position of the person's face is (x, y, z).
3A is an installation view of the xy coordinates of the camera, 211 is the camera A, θHA is the horizontal angle of view of the camera A, 221 is the camera B, θHB is the horizontal angle of view of the camera B, and 241 is the object. (Person), x is the horizontal distance from the camera A to the person, y is the vertical distance from the camera A to the person, xb is the horizontal distance from the camera B to the person, and YA is the horizontal imaging of the camera A at the point where the person is present. A range YB is a horizontal imaging range of the camera B at a point where a person is present.

  3B is an installation view of the xz coordinate of the camera, θVA is the vertical angle of view of camera A, hA is the height of camera A, θVB is the vertical angle of view of camera B, and hB is the height of camera B. Z is the height of the person, ZA is the vertical imaging range of the camera A at the point where the person is, ZB is the vertical imaging range of the camera B at the point where the person is, and 231 is the ground.

  3C is a photographed image of the camera A, 251 is a monitor A that displays the photographed image taken by the camera A, 252 is a person in the video of the monitor A, HMA is a horizontal dimension of the screen of the monitor A, VMA is the screen vertical dimension of the monitor A, HAC is the horizontal distance from the center of the monitor A to the person 252, VAC is the vertical distance from the center of the monitor A to the person's face center, VA is the height of the person in the monitor A, VAF Is the height of the face of the person on the monitor A.

3D is a photographed image of the camera B, 261 is a monitor B that displays the photographed image captured by the camera B, 262 is a person in the video of the monitor B , and HMB is a screen horizontal dimension distance of the monitor B. , VMB is the screen vertical dimension of the monitor B , HBC is the horizontal distance from the center of the monitor B to the person, VBC is the vertical distance from the center of the monitor B to the person's face center, VB is the height of the person on the monitor B, VBF Is the height of the person's face on the monitor B.

Next, a method for identifying the same person from videos captured by different cameras A and B according to an embodiment of the present invention will be described with reference to FIG.
Here, hA = hB, θHA = θHB, and θVA = θVB.
Let L be the horizontal distance between camera A and camera B.
Since the ratio of the distance to the person and the ratio of the size of the person in the screen are similar, the following equation holds.

The horizontal distance x from the camera A to the person is obtained by (Expression 2) from the relationship of (Expression 1).
VA: VB = (L−x): x (Equation 1)
x = LVB / (VA + VB) (Equation 2)
The vertical distance y from the camera A to the person is obtained by (Expression 4) from the relationship of (Expression 3).
y: HAC = YA / 2: HM A / 2 (Equation 3)
y = xtan (θHA / 2) × 2HAC / HM A
= (2LHACVBtan (θHA / 2)) / (HM A (VA + VB))
... (Formula 4)
The person's height z is obtained by (Expression 6) from the relationship of (Expression 5).
(Z-hA): VAC = z: VA (Formula 5)
z = hAVA / (VA-VAC) (6)
The horizontal distance HBC from the center of the monitor B to the person is obtained by (Expression 8) from the relationship of (Expression 7).
HBC: HM B / 2 = y: (L−x) tan (θHB / 2) (Expression 7)
HBC = HACVB / VA (Equation 8)
The vertical distance VBC from the center of the monitor B to the center of the person's face is obtained by (Expression 10) from the relationship of (Expression 9).
VBC: VM B / 2 = (z−hB): ZB / 2 (Equation 9)
VBC = (hAVM B VAC (VA + VB)) / (4LVA (VA−VAC)
tan (θVA / 2)) (Equation 10)

In the above equation, hA, HM A , HMB, VM A , VMB, L, θHA, and θVA are constants, and HAC, VAC, VA, and VB are position information in the person's screen. This is a calculated value.
That is, HBC and VBC can be obtained from the above values using (Equation 8) and (Equation 10), and from the position information of the person imaged by the camera A 211, the image of another camera (here, camera B221) can be obtained. It is possible to specify which person is the same person.
Then, the three-dimensional position coordinates x, y, and z of the person's face in the monitoring space can be obtained from (Expression 2), (Expression 4) , and (Expression 6).
In the above description, the position information of the predetermined person is obtained from the results displayed on the monitor A 251 and the monitor B 261. However, the position information of the predetermined person may be obtained from captured images captured by the camera A 211 and the camera B 221.

Next, an operation of combining and displaying a face image of a person determined to be the same among images taken by a plurality of cameras as an image closer to the front will be described with reference to FIGS.
FIG. 4 is a diagram for explaining the display contents of the monitor A and the monitor B.
In FIG. 4, 251 is a monitor A, 252 is a person in the image of the monitor A, 411 is a face of the person detected by the image processing apparatus 102, 261 is a monitor B, 262 is a person in the image of the monitor B, Reference numeral 421 denotes a human face detected by the image processing apparatus 102.
It is assumed that the front of the person's face is shown on the monitor A and the back of the person's face is shown on the monitor B.
The image processing apparatus 102 determines the face orientation based on facial feature points, for example, information on eyes, nose, and mouth.
Since the face of the person on the monitor B faces the back, the face image of the person on the monitor A is cut out and synthesized with the video on the monitor B. At this time, the face image of the person on the monitor A is converted into the size of the face image on the monitor B and synthesized. The enlargement / reduction ratio at the time of conversion is assumed to be VBF / VAF.

FIGS. 5A and 5B are a front view and a diagonal view of a human face.
In FIG. 5, reference numeral 501 denotes a detection area of the face portion of a person facing the front, 502 a face of the person facing the front, 503 a person's right eye α, 504 a person's left eye β, and 505 a center point of the right eye and the left eye. 2R and 506 are circles having a diameter of the straight line 505, and the head is seen from above. 507 is the person's right eye α, 508 is the person's left eye β, 509 is the radius R, and 510 is the center of the right eye. A line segment with the center point of the point and the left eye as the end, 511 is a straight line including the center point of the line segment 510 and the center point of the circle, 512 is a distance Lα between the straight line 511 and the left end of the circle, 513 is a straight line 511 and the circle Distance Lβ to the right end, 14 is the detection area of the face portion of the person facing diagonally, 515 is the face of the person facing diagonally, 516 is the right eye α of the person, 517 is the left eye β of the person, 518 is the right eye and the left eye Line segments 2R and 519 including the center point have a straight line 518 as a diameter. 520 is a person's right eye α, 521 is a person's left eye β, 522 is a radius R, 523 is a line segment with the center point of the right eye and the center point of the left eye as ends, and 524 is A straight line including the center point of the line segment 523 and the center point of the circle 525 is a straight line including the center point of the line segment 510 and the center point of the circle in the figure in which the face is facing the front, and 526 is a fan including the straight line 524 and the straight line 525. The mold center angle θ 527 is the distance Lα between the straight line 524 and the left end of the circle, and 528 is the distance Lβ between the straight line 524 and the right end of the circle.

Next, θ, Lα, and Lβ in the figure with the person's face facing the front are as follows.
θ = 0 ° (Equation 11)
Lα = Lβ (Equation 12)
Further, θ (in this case, θ ′), Lα, and Lβ in the figure in which the human face is directed obliquely are represented by the following equations.
θ = θ ′ (Equation 13)
Lα = R−Rsinθ ′ (Equation 14)
Lβ = R + Rsin θ ′ (Equation 15)
Therefore, θ ′ can be obtained from the following equation.
Lα: Lβ = (R−Rsin θ ′): (R + Rsin θ ′) (Expression 16)
sin θ ′ = (Lβ−Lα) / (Lα + Lβ) (Expression 17)
θ ′ = sin−1 ((Lβ−Lα) / (Lα + Lβ)) (Equation 18)
When the image processing apparatus 102 specifies the size of the face and the positions of the right eye and the left eye, the distance Lα and the distance Lβ are obtained.
θ ′ can be calculated from the distance Lα and the distance Lβ in (Equation 18), and the orientation of the person's face can be expressed as an angle. As θ ′ is closer to 0 °, it can be determined that the person is facing the front.

Next, a method for capturing a person's face with a camera in the front direction when four cameras are used will be described.
FIG. 6 is a diagram of the orientation of a person's face and the corresponding camera in a monitoring space where four cameras according to an embodiment of the present invention are arranged.
FIG. 6 is an overhead view of the xy coordinate axes.
It is assumed that the camera A 602 and the camera B 603 are installed opposite to each other on a straight line.
6, 602 is a camera A, 603 is a camera B, 604 is a camera C, 605 is a camera D, 606 is a person's head viewed from above, 607 is a person's face orientation, and 608 is a face orientation relative to the x-axis. 609 is a straight line including the cameras A602 and B603 that are installed opposite to each other, 610 is a straight line including the center point between the cameras A602 and B603 of the straight line 609 and the camera C604, and 611 is a camera A602 and a camera B603 of the straight line 609. 612 is the center line of the straight lines 609 and 610, 613 is the center line of the straight lines 609 and 610, 614 is the center line of the straight lines 609 and 611, and 615 is the straight line of the straight lines 609 and 609. 611 is a center line, 616 is an angle θ1 of the straight line 612 with respect to the x-axis, 617 is an angle θ2 between the straight line 610 and the straight line 613, and 618 is a straight line. Angle θ3,619 respect to the x-axis of 14 is the angle θ4 of the straight line 611 and the line 615.

The orientation θ of the person's face is obtained by the image processing apparatus 102 by the method described above.
In order to capture a person's face in the front direction with the camera, the relationship between the orientation θ of the person's face and the assigned camera is as follows.
The range of θ when the camera A 602 is assigned is (Equation 19).
0 ≦ θ <θ1, 2 (θ1 + θ2 + θ3) + θ4 ≦ θ <360 (Equation 19)
The range of θ when assigning the camera B 603 is (Equation 20).
θ1 ≦ θ <2θ1 + θ2 (Equation 20)
When assigning the camera C604, the range of θ is (Expression 21).
2θ1 + θ2 ≦ θ <2 (θ1 + θ2) + θ3 (Formula 21)
When assigning the camera D605, the range of θ is (Equation 22).
2 (θ1 + θ2) + θ3 ≦ θ <2 (θ1 + θ2 + θ3) + θ4 (Equation 22)

  The system management server 105 periodically manages the current human face orientation and assigned camera. For example, a face image in a specific image such as the image of camera A 602 is converted at an enlargement / reduction ratio obtained by the method described in the description of the embodiment of FIG. 4 according to the face orientation that changes with time. The assigned camera face image can be combined and replaced. The observer can obtain a face image capturing the front direction without depending on the action of the person while viewing the same video.

Next, an image captured by four cameras every continuous time will be described with reference to FIG.
The system management server 105 manages face images in specific videos captured by the camera A 602, the camera B 603, the camera C 604, and the camera D 605.
In FIG. 7, 701 is the image of the camera A602 at time t = 1, 702 is the face of the person in the image 701, 703 is the image of the camera B603 at time t = 1, 704 is the face of the person in the image 703, and 705 is the time t. = 1 is the image of the camera C604 at = 1, 706 is the face of the person in the image 705, 707 is the image of the camera D605 at time t = 1, 708 is the face of the person in the image 707, 709 is the composite image based on the image 5, 710 Is the face of a person in the video 709.
Also, 711 is the image of the camera A602 at time t = 2, 712 is the person's face in the image 711, 713 is the image of the camera B603 at time t = 2, 714 is the person's face in the image 713, and 715 is the time t = 2. 716 is a human face in the video 715, 717 is a video of the camera D605 at time t = 2, 718 is a human face in the video 717, 719 is a composite video based on the video 715, and 720 is a video 719 is the face of a person.
Further, 721 is the image of the camera A602 at time t = 3, 722 is the face of the person in the image 721, 723 is the image of the camera B603 at time t = 3, 724 is the face of the person in the image 723, and 725 is time t = 3. 726 is a human face in the video 725, 727 is a video of the camera D605 at time t = 3, 728 is a human face in the video 727, 729 is a composite video based on the video 725, and 730 is a video. 729 is the face of a person.

For example, it is assumed that the monitor is watching the video 705 captured by the camera C604 on the operation terminal 106.
When the time t = 1, the image facing the front is the image 701 captured by the camera A602. At this time, since the person of the video 705 that the supervisor is viewing is facing the back, the system management server 105 cuts out the face image of the person of the video 701, converts the image size, and synthesizes the face image of the video 705. Thus, a video 709 is generated.
When the time t = 2, the image facing the front is an image 717 captured by the camera D605. At this time, since the person in the video 715 that the supervisor is watching is facing sideways, the system management server 105 cuts out the face image of the person in the video 717, converts the image size, and synthesizes the face image in the video 715. Thus, an image 719 is generated.
When the time t = 3, the image facing the front is the image 723 captured by the camera B603. At this time, since the person in the video 725 viewed by the supervisor is facing sideways, the system management server 105 cuts out the face image of the person in the video 723, converts the image size, and synthesizes the face image in the video 725. Thus, an image 729 is generated.
Note that the above-described synthesized video may be generated by the image processing apparatus 102 or the operation terminal 106.
As described above, according to the present invention, it is possible to monitor a video on which a front image of a face is always displayed without being influenced by the movement of a person.

  Further, for example, when constructing a system in which the image monitoring system of the present invention and the similar face image authentication system are integrated, the system management server 105 assigns a unique ID to each detected person, as well as ID information and the front direction. By providing the similar face image search server with the image synthesized with the face image suitable for the face, it is possible to improve the accuracy of the face authentication process and reduce the processing load.

Next, the current position information and trajectory information of the monitoring target (person) in the image monitoring system that is an embodiment of the present invention will be described.
FIG. 8 is a plan view on the xy coordinate axes for explaining the monitoring space.
In FIG. 8, 802 is a plan view, 803 to 809 are surveillance cameras installed in the surveillance space, 810 is a person, and 811 is a locus of the person.
The system management server 105 can manage the position coordinates of the person for each time by executing the calculation of the person coordinates described in the description of the embodiment in FIG. It can be expressed by 810, and the locus can be generated on the plan view by connecting the position of the person for each time with a straight line. In the present invention, it is possible to specify a position only by a monitoring camera without providing a sensor or the like for specifying a person position, and to provide current position information and trajectory information in a plan view.

Next, a method for generating the trajectory of the target person will be described with reference to FIG.
For example, the camera 803 captures a predetermined monitoring area including the target person 810 and transmits the captured image to the image processing apparatus 102.
The image processing apparatus 102 converts the position of the person 810 among the plurality of persons shown in the monitoring area image into global coordinates by the above-described method.
Next, the camera 806 captures a predetermined monitoring area including the target person 810, and transmits the captured image to the image processing apparatus 102.
The image processing apparatus 102 converts the position of the person 810 among the plurality of persons shown in the image captured by the camera 806 into global coordinates by the above-described method.
If the position information of the global coordinates of the person 810 photographed by the camera 803 and the position information of the global coordinates of the person 810 photographed by the camera 806 are substantially the same at a specific time, the image processing apparatus 102 It can be specified that they are the same person.
Similarly, the image processing apparatus 102 converts the position of the person 810 from the images obtained by capturing the predetermined monitoring area including the target person 810 by the camera 804 and the camera 806 into global coordinates, and converts the converted position. If the information is substantially the same, the person 810 is identified as the same person.
Then, the image processing apparatus 102 can generate the locus 811 of the person 810 by connecting the position information of the global coordinates of the person 810.

Next, a camera allocation area on a plan view on the xy coordinate axes according to an embodiment of the present invention will be described with reference to FIGS.
FIG. 9 is a diagram for explaining a camera allocation area on a plan view.
9, 802 is a plan view, 803 to 809 are surveillance cameras installed in the surveillance space, 910 is an obstacle, 911 is an allocation area of the camera 803, 912 is an allocation area of the camera 804, and 913 is an allocation area of the camera 805 , 914 is an allocation area for the camera 806, 915 is an allocation area for the camera 807, 916 is an allocation area for the camera 808, and 917 is an allocation area for the camera 809.

The operation terminal 106 registers an allocation area for each camera in advance on the xy plan view 802. The allocation area is a division of the monitoring space into a plurality, and the most visible camera is allocated in each area.
If there is an obstacle 910 in the allocated area of the monitoring space, the operation terminal 106 is closest to the camera 807 in the vicinity of the obstacle, but the shadow of the obstacle 910 is not visible when viewed from the camera 807. As shown, this portion is registered as an allocation area for the camera 808. As described above, the operation terminal 106 sets the monitoring space allocation area and registers the camera for each area, whereby the corresponding camera can be selected in accordance with the detected position of the human coordinate.
For example, the operation terminal 106 selects an image captured by the camera 807 when a person is present on the left side of the obstacle 910, and selects an image captured by the camera 808 when a person is present on the right side of the obstacle 910. select.
In addition, when the operation terminal 106 wants to display the back side of the obstacle 910 while outputting the video of the camera 807, the operation terminal 106 combines the video of the camera 808 or 809 capturing the back side of the obstacle 910 with the obstacle 910. Can do. Note that the video of the camera 807 and the video of the camera 808 or 809 may be combined by the image processing apparatus 102 and the combined video may be displayed on the operation terminal 106.

Next, a monitoring screen of the operation terminal according to an embodiment of the present invention will be described with reference to FIG.
In FIG. 10, 1001 is a monitoring screen, 1002 is an image display unit, 1003 is a person in the image, 1004 is a plan view, 1005 is a person position in the plan view, and 1006 is capturing an image of the image display unit. A camera 1007 is a camera that provides a human face image.
The system management server 105 has a WEB server function that provides a monitoring screen for performing a monitoring operation from the browser to the operation terminal, and provides the screen shown in FIG.
The camera that provides the video on the video display unit and the camera that provides the face image are color-coded by the camera icon on the floor plan, so that the monitor can visually grasp which video is being displayed. Is possible. The face image may be automatically synthesized and displayed on the image of the camera capable of capturing the front, the face portion of the video display unit may be touched with a finger, or the face image may be switched to the face image facing the front at the timing of sliding.
In addition, in the case of a camera equipped with an electric pan head and an electric zoom lens, it is possible to further improve the human visibility by operating the assigned camera with PTZ (pan / tilt / zoom) with respect to the human coordinates. is there. The PTZ operation may be automatically performed, or may be performed at the timing when a person on the plan view is touched.

Next, a left and right monitoring camera using a mirror according to another embodiment of the present invention will be described with reference to FIG.
In FIG. 11A, 1101 is a camera, 1102 is a camera housing, 1103 is a horizontal angle of view θ of the camera, 1104 is a left mirror, 1105 is a right mirror, 1106 left and right mirror installation angles, and 1107 is a left and right mirror. The distance x between the boundary and the camera lens, 1108 is the mirror length L, 1109 is the cosine of a right triangle with the straight line 1108 as the hypotenuse, 1110 is the sine of the right triangle with the straight line 1108 as the hypotenuse, 1111 is the left glass surface Reference numeral 1112 denotes a right glass surface, and 1113 denotes a camera image obtained by imaging both mirrors.
In FIG. 11B, reference numeral 1114 denotes a left half video of the video 1113, 1115 denotes a right half video of the video 1113, 1116 denotes a video cut out from the video 1114, and 1117 denotes a video cut out from the video 1115.

In FIG. 11A, the installation angle of the left mirror 1104 and the right mirror 1105 is 90 °. The horizontal angle of view of the camera is the angle of view that captures the ends of the left and right mirrors.
As an installation condition of the camera 1101, the distance x between the camera 1101 and the mirror is obtained by the following method.
tan (θ / 2) = (L / 2 ^ (1/2)) / ((x + L) / 2 ^ (1/2))
... (Formula 23)
x = (L / 2 ^ (1/2)) × (1 / (tan (θ / 2) −1) (Equation 24)

The camera 1101 can obtain a left-right image as one image 1113 by capturing an image projected on the left and right mirrors. The image processing apparatus 102 can divide this image into left and right parts to provide two images, a left image 1116 and a right image 1117.
The position information calculation according to the embodiment of the present invention can be performed by a facing camera as shown in FIG. 2, but one camera is used by using a left and right monitoring camera using a mirror as shown in FIG. Since both the left and right images can be obtained, the number of installed cameras can be halved.

  Although the present invention has been described in detail above, the present invention is not limited to the above-described embodiments, and various modifications can be made without departing from the spirit of the present invention.

  101-1 to 101-n: camera, 102-1 to 102-n: image processing apparatus, 103-1 to 103-n: IP encoder, 104: network, 105: system management server, 106-1 to 106-n : Operation terminal 211: Camera A, 221: Camera B, 251: Monitor A, 261: Monitor B, 602: Camera A, 603: Camera B, 605: Camera C, 610: Camera D, 802: Plan view, 803 , 804, 806, 808, 809: camera, 1001: monitor, 1101: camera, 1102: camera housing, 1104: left mirror, 1105: right mirror, 1111: left glass surface, 1112: right glass surface.

Claims (2)

In an image monitoring system having a first camera, a second camera disposed opposite to the first camera, and an image processing device,
The image processing apparatus, the first person on the image captured by the first camera, the horizontal distance information and the vertical distance information from the center of the image to the center of the face of the first person, the first the height information of the first person on the images taken with the camera, and the height information of the second person on the image captured by the second camera, on the basis of, of the first person, wherein the center of the captured image with the second camera calculates the horizontal distance and vertical distance to the center of the face of the second person, the specific or the first person is the same person and the second person An image monitoring system characterized by: The image monitoring system according to claim 1,
The image monitoring system further includes a system management server,
The system management server, when the face of the first person is other than the front, and select a face image closer to the front in the face image of a person was taken by the plurality of cameras, and the selected An image monitoring system, wherein a face image is synthesized with an image of the first person whose face image is other than a front face.

Images