JP2011076573A - Monitoring camera system and abnormality detecting method therefor - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To discriminate an abnormal camera by detecting a fluctuation in the installation position or the direction and posture of each camera and an abnormality of a camera image with high accuracy concerning a monitoring camera system for performing monitoring by using a plurality of cameras and its abnormality detecting method. <P>SOLUTION: A plurality of cameras 6-2 having appearance features different from those of the other cameras are installed so as to photograph at least two other cameras, a camera selecting part 6-12 selects a video picture photographed by the cameras 6-2, and a camera detecting part 6-13 detects a camera image, and identifies a camera to be photographed from an appearance feature of the camera. A camera position calculating part 6-14 calculates an installation position of the camera from a video position of the camera, and a camera posture calculating part 6-15 calculates the posture of the camera. An abnormal camera detecting part 6-16 discriminates whether the camera photographing the camera has an abnormality or whether the photographed camera has an abnormality by checking the installation positions of the other cameras when the previously calculated position is different from the position of this time. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a monitoring camera system that monitors the occurrence of an abnormal situation or emergency using a plurality of cameras, and an abnormality detection method thereof.

  In a surveillance camera system that is used for applications such as finding suspicious persons or abnormal situations from the images using multiple cameras, in order to capture the target object without omission, each camera always has a predetermined surveillance area. Each camera must be installed at a predetermined position and in a predetermined direction and posture so that it can be photographed.

  However, even if each camera is installed at a predetermined position and orientation as expected, somebody may be mischiefed or crafted by the camera during operation, or There may be cases where the installation position, orientation, or posture of the camera moves due to contact with the load. If the position, orientation, and orientation of the camera move, the shooting range will shift and normal surveillance video will not be obtained. In addition, a normal surveillance video may not be obtained due to a failure in the camera itself or the video signal transmission path.

  In such a case, a normal image of the monitoring area cannot be obtained, and if an abnormal camera is left as it is, the occurrence of a suspicious person or an emergency situation is overlooked. In order to prevent such troubles, when an abnormality occurs in the camera image including the deviation of the photographing range, it is necessary to perform a countermeasure process such as immediately notifying the administrator of the abnormal camera.

  In such a handling process, when the number of cameras is small, a supervisor or the like can deal with the situation by visually checking the images of all the cameras and checking the camera in which an abnormality has occurred in the images. However, when dozens or hundreds of cameras are installed in a wide area, it is difficult to detect all camera image abnormalities without omission by human work such as surveillance. There is a demand for a surveillance camera system that automatically identifies and reports a camera in which an abnormality has occurred.

  As such a conventional surveillance camera system, an angle sensor, a position sensor, and the like are attached to the camera, and a method of detecting a change in the installation position, orientation, and orientation of the camera by analyzing the signals of those sensors is used. Proposed. However, with this method, it is not possible to detect an abnormality in the camera image that occurs due to the camera's field of view being obscured.

  On the other hand, as a technique for detecting camera abnormality using video captured by a camera, for example, a predetermined marker (mark) is set at a predetermined location within a shooting range, and the marker is included in the camera video. There is a method for determining whether or not a camera is present at a predetermined position and notifying the camera of an abnormality when it is determined that the camera does not exist. However, with this method, although the camera installation position does not change, it is determined that there is an abnormality in the camera only by the change in the marker position, so it is difficult to identify a true abnormal camera.

  Furthermore, in a surveillance camera system in which dozens or hundreds of cameras are installed in a wide range, the number of markers must be set to tens or hundreds, which is very laborious to install the markers. It takes. In addition, when the marker cannot be arranged due to the external design of the monitoring area, it cannot be adopted.

  To solve such a problem, in a surveillance camera system having a plurality of cameras, the estimated position of the observation target such as a person is calculated by each camera, and the deviation between the estimated position of the reference camera and the estimated position of each camera is calculated. Therefore, a method for obtaining the posture / position of each camera has been proposed (see Patent Document 1). However, with this method, a camera abnormality can be detected only when the observation target is captured by the camera.

JP 2003-279315 A

  The present invention relates to a monitoring camera system that performs monitoring using a plurality of cameras, and does not provide a marker or the like in the monitoring area separately from the camera, and changes in the installation position or orientation / posture of each camera and camera video. It is an object of the present invention to automatically detect an abnormality and accurately identify an abnormal camera.

  A surveillance camera system as an embodiment for solving the above-described problem is to detect a camera image from a plurality of cameras having appearance characteristics different from those of other cameras, and an image screen shot by each of the cameras, A camera detection unit for recognizing appearance characteristics of the camera video and identifying a camera to be shot, and the camera to be shot on the video screens of two cameras that have shot the same camera shot detected by the camera detection unit The camera position calculation unit for calculating the installation position of the camera to be imaged from the image position of the camera, and the installation position of the camera to be photographed previously calculated by the camera position calculation unit and the current installation position are compared. And an abnormal camera determination unit that determines that an abnormality has occurred in the installation position of the camera in a different case.

  Further, according to another aspect of the present invention, there is provided an abnormality detection method in which a plurality of cameras having appearance characteristics different from those of other cameras are installed such that at least two other cameras are photographed. A first process of detecting a camera video from a video screen shot by each camera, recognizing an appearance feature of the camera video, and identifying a camera to be shot, and detected in the first process The second process of calculating the installation position of the camera to be photographed from the video positions of the camera to be photographed on the video screens of the two cameras that photographed the same camera to be photographed, and the previous calculation in the second process. When the installation position of the camera to be photographed is compared with the current installation position, and the two are different from each other, the image position of the other camera taken by each of the two cameras that photographed the camera to be photographed is Calculated in the process of 2 The installation position calculated for the other camera is compared with the installation position calculated last time and the installation position calculated this time. If they do not match, it is determined that an abnormality has occurred in the camera that captured the video. , The third step of determining that an abnormality has occurred at the installation position of the camera to be photographed.

  In a surveillance camera system that uses multiple cameras for monitoring, it is possible to detect changes in the installation position or orientation / posture of each camera and camera image abnormalities without arranging markers (markers) in the surveillance area separately from the cameras. Based on the video of each camera taken by each camera, it can be automatically detected with high accuracy, and an abnormal camera can be identified and notified to the management center or the like.

It is a figure which shows the example which image | photographed the same object with two cameras. It is a figure which shows the example of a coordinate for calculating the position of a detection target. It is a figure which shows the example of a coordinate for calculating the position of a detection target on a XZ plane. It is a figure which shows the flow of the principle which calculates the position and attitude | position of a camera. It is a figure which shows the coordinate system of the relationship between the position of a target object, and the position and attitude | position of a camera. It is a figure which shows the apparatus structure of a surveillance camera system, and the function structural example. It is a figure which shows the example of the camera information database in a surveillance camera system. It is a figure which shows the processing flow in a surveillance camera system. It is a figure which shows the system environment example of the Example to which the surveillance camera system is applied. It is explanatory drawing of the triangulation in an Example. It is a figure which shows the example of an input video from each camera in an Example. It is a figure which shows the example of a change of the camera information database in an Example. It is a figure which shows the example of the image | video of the camera A and the camera B in an Example. It is a figure which shows the example of the image | video of the camera D in an Example. It is a figure which shows the propagation example of calculation of the camera position and attitude | position in an Example. It is a figure which shows the example of the method of detecting the abnormality of the camera in an Example.

  In the disclosed surveillance camera system, when a surveillance area is photographed by a plurality of cameras, each camera is arranged so that not only the surveillance object but also other cameras are photographed simultaneously. Then, the position or orientation / posture coordinate position of each camera is calculated from the video position of the other camera taken by each camera.

  Using the fact that the coordinate position of the other camera position calculated from the video of the camera whose camera position, orientation, or orientation has changed is different from the coordinate position of the other camera position calculated from the video of the normal camera. Detect and identify an abnormal camera whose position or orientation / posture has changed. Hereinafter, calculation of the coordinates of the position and orientation of the camera will be described in detail.

  First, it is assumed that the position, orientation, and orientation of at least two cameras are known as camera installation conditions. Then, at least two cameras whose positions, orientations, and orientations are known are installed so that cameras that are to newly calculate the positions, orientations, and orientations are photographed. Furthermore, the camera whose position, orientation, and orientation are to be calculated is installed so that two or more cameras with known positions are photographed.

(1) Calculation of Camera Position As shown in FIG. 1, if two cameras A and B whose positions, orientations and orientations are known can capture another common camera C (detection target), a triangle is obtained. The position of the camera C can be calculated using the principle of surveying. For example, as shown in FIG. 2, a coordinate system with the focal point position of the left camera as the origin is assumed. The two cameras A and B have a focal length f, the distance between the two cameras A and B is d, and the directions of the optical axes of the two cameras A and B are equal.

In FIG. 2, 2-1 is a video screen of the left camera A, 2-2 is a video screen of the right camera B, and 2-3 is a camera C whose position is to be detected. The coordinate value of the detection target camera C on the video screen 2-1 of the camera A is (xl, yl), and the coordinate value of the detection target camera C on the video screen 2-2 of the camera B is (xr, yr). Further, the coordinate value of the detection target camera C is assumed to be (X _p , Y _p , Z _p ).

という関係式が成立する。 When the arrangement relationship shown in FIG. 2 is viewed from above the Y axis, the positional relationship in the XZ plane is expressed as shown in FIG. 3, and from the principle of similarity of triangles,

The following relational expression holds.

という計算式で算出される。同様に、物体の位置Ｙ_ｐとＺ_ｐも

として算出される。 The position X _p of the object is

It is calculated by the following formula. Similarly, the positions Y _p and Z _p of the object are

Is calculated as

  Therefore, first, as shown in FIG. 4A, the cameras α, β, and γ are arranged so that the other two cameras are photographed with respect to each other. Assume that the posture is known. Then, as shown in FIG. 4B, the camera γ is photographed from the camera α and the camera β, and the position of the camera γ is calculated from the images of the camera α and the camera β by the above formula.

(2) Calculation of camera orientation / posture As shown in FIG. 5, when the camera coordinate system is xyz, the ground coordinate system is XYZ, and the video coordinate system is xy, the camera position, orientation, and orientation are The coordinate values (X ₀ , Y ₀ , Z ₀ ) and the camera rotation angles (ω, φ, κ) around the x-axis, y-axis, and z-axis are represented by six parameters. Between these six parameters, the ground coordinate values (X, Y, Z) of the object P and the image coordinate values (x, y) on the screen of the object P, the following relationship is established. It is known.

From this relational expression, when the position (X ₀ , Y ₀ , Z ₀ ) of the camera γ is known, the ground coordinate values (X, Y, Z) of two or more objects P photographed by the camera γ. ) Is known, the orientation and orientation (ω, φ, κ) of the camera γ can be obtained. That is, if two or more cameras whose positions are already known have been shot on the camera γ whose position has been calculated by the method described in the above (1), the direction and orientation of the camera γ can be calculated. .

  As a camera installation condition in this surveillance camera system, by installing each camera so that two or more cameras whose positions are known are photographed, for example, the camera γ arranged as shown in FIG. The direction and orientation can be calculated from the shooting positions of the cameras α and β whose positions are known in the video of the camera γ. In this way, it is possible to photograph other cameras with each other and calculate the position, orientation, and orientation of each camera from the video data of each camera.

  FIG. 6 shows an example of the configuration of a processing device in this surveillance camera system. This monitoring camera system includes a processing device 6-1 such as a personal computer, and the processing device 6-1 is connected to a plurality of cameras (imaging devices) 6-2 for imaging a monitoring area. The processing device 6-1 includes a storage unit that stores a camera information database 6-11.

  As shown in FIG. 7, the camera information database 6-11 includes, for each camera, a camera ID, camera characteristics (color, etc.), camera position (X, Y, Z), camera orientation / posture (ω, φ, κ). ), Used camera ID used for calculation of position, orientation / posture, previous camera position (X, Y, Z), previous camera orientation / posture (ω, φ, κ), and information of camera video data at the current time Is done.

  The processing device 6-1 further includes a camera selection unit 6-12 that selects video data used for calculation of the position, orientation, and orientation based on the camera position, orientation, and orientation data of the camera information database 6-11. Is provided. In addition, a camera detection unit 6-13 that detects a video of the camera to be shot from a video screen (video data) shot by the camera is provided.

  In addition, a camera position calculation unit 6-14 that calculates the installation position of the camera and a camera posture calculation unit 6-15 that calculates the orientation / posture of the camera from video data captured by the camera whose camera position has been calculated are provided. Also, an abnormal camera detection unit 6-16 that detects an abnormal camera by comparing the previous calculated value of the installation position, orientation, and orientation with the current calculated value is provided.

  FIG. 8 is a flowchart showing the flow of processing in this surveillance camera system. With reference to this flowchart, processing for detecting an abnormality in the installation position and orientation / posture of the camera will be described. First, the camera information database data is initialized. In this initialization, camera IDs are set in order in the camera information database for the number of cameras installed.

  Further, as camera features, features that can individually identify each camera from the camera image, for example, features such as camera color, are set. This camera feature may be a camera-specific pattern, symbol, shape, or the like in addition to the camera color. Also, as the camera installation position, orientation, and orientation, values of the installation position, orientation, and orientation are stored in advance for at least two known cameras, and “unknown” information is provided for other cameras. Stored and information indicating that the camera ID used is “none” is stored (8-1).

  Next, video data captured by each camera is acquired, and the video data is stored in the camera information database (8-2). Further, the current camera installation position and orientation / posture values are copied to the previous camera position and previous camera posture values. The camera selection unit 6-12 reads information on the installation position, orientation, and orientation of each camera from the camera information database 6-11, searches for a camera ID that is not “unknown” in both position, orientation, and orientation, and uses the camera ID as a camera. Notify the detection unit 6-13.

  The camera detection unit 6-13 reads out the video data of the notified camera ID from the camera information database 6-11, and detects the video of the camera from the video screen. Specifically, the shape of the camera is learned in advance, and the video of the camera is detected by extracting a video portion similar to the shape from the video screen.

  When camera images having the same camera characteristics are detected in two or more camera images, the camera position calculation unit 6-14 determines that the same camera has been shot, and determines the camera position as 2 It is calculated and estimated from the camera image positions of the image screens of two cameras by the principle of triangulation (8-3).

  Further, the determined camera feature and the camera feature in the camera information database 6-11 are collated, the camera ID of the determined camera feature is identified, and the position of the camera is written in the current camera position and updated. Further, the camera ID of the video used for calculating the position is written as the use camera ID.

  The calculated value of the camera position is compared with the value of the previous camera position of the camera to determine whether or not they are different (8-4). If the current camera position value is different from the previous camera position, the process proceeds to the flow after the flow 8-5, and if equal to the previous camera position, the process proceeds to the flow after the flow 8-9.

  When the value of the camera position is different from the previous camera position, the following two causes are assumed. The first cause is when the position of the camera to be photographed has actually moved from the previous position. The second cause is a case where an abnormality due to zooming or the like occurs in the video of the photographing camera used for calculating the camera position.

  Therefore, the camera ID of the camera that captured the image different from the previous camera position is checked, and the position of another camera calculated using the camera image is calculated and estimated (8-5). When the data of the position of another camera is not stored, the installation position of another camera (the camera whose installation position is recognized as correct) is calculated.

  It is determined whether or not the installation position of another camera is different from the previous data (normal value) (8-6). If they match, the video of the camera used to calculate the installation position is normal, It is determined that the correct calculation result is output, it is determined that there is an abnormality in the camera to be photographed in which the change in the installation position is detected, and the camera ID is notified to the management center or the like (8-7).

  Also, if the installation position of another camera is also different from the previous data (normal value), it is determined that an abnormality has occurred in the camera that is shooting this video (the camera used for position calculation) The camera ID of the camera is notified to the management center or the like (8-8).

  Next, processing for calculating and estimating the orientation / attitude of the camera is performed as follows (8-9). First, a camera ID whose camera position is known and whose camera posture is “unknown” is searched from the camera information database 6-11. The camera image data at the current time of the camera ID found by this search is read, and the camera image is detected from the image screen.

  The camera feature of the camera image detected on the video screen is compared with the camera feature stored in the camera information database 6-11 to identify the camera ID having the camera feature detected on the video screen, Read the camera position of the camera ID. Since two or more cameras are photographed on the video screen depending on the installation conditions, the direction / posture values of the photographing cameras are calculated and estimated from the video positions of those cameras.

  It is determined whether the calculated camera orientation / posture value is different from the previous camera posture value (8-10). If they are different, this is because the orientation / posture of the photographing camera has actually been changed, or an abnormality has occurred in the image of the photographing camera. Regardless of the cause, an abnormality has occurred in the photographing camera itself, and the camera ID of this photographing camera is notified to the management center or the like (8-11).

  If the calculated camera orientation / posture value is the same as the previous camera orientation value, it is determined that the camera is normal, and it is determined whether the installation position and orientation / posture have been calculated for all cameras (8- 12) If calculation has not been completed for all the cameras, the process returns to the process flow 8-3 and the same process is repeated. When the calculation is completed for all the cameras, the process returns to the process flow 8-1, and the same process is repeatedly executed.

  A specific example when this surveillance camera system is actually used will be described. Here, as shown in FIG. 9, an example of a monitoring system in which a plurality of cameras A, B, C, D, E, F, G, H, and I are directed directly from the ceiling will be described. In this surveillance camera system, a camera equipped with a lens having a field angle of 180 degrees or more, such as a fisheye lens, is installed on the ceiling so that other cameras can be photographed in the video screen even when the camera is pointed directly downward.

Further, it is assumed that the optical axis of the camera is directly below, and that the heights Z of all the camera installation positions (ceilings) are equal. Then, as shown in FIG. 10, the following relational expression is established between the ground coordinates of the two cameras A and B whose installation positions are known and the camera C whose installation position is to be calculated. Here, the values of the ground coordinates of the cameras A, B, and C are respectively (Xa, Ya), (Xb, Yb), and (Xc, Yc). The distance between the cameras A and B is d. The directions of the camera C from the cameras A and B are θ and φ, respectively. For simplicity, Ya = Yb.

  Since each camera is installed directly below the ceiling, as long as each camera captures an image as shown in FIG. 11 and the camera C is captured in the image screens of cameras A and B, The direction (θ and φ) of the camera C is easily obtained. Furthermore, since the camera orientation / orientation has only one degree of freedom (κ), the orientation / orientation can also be calculated by detecting one camera image with a known installation position from one image screen.

  In this embodiment, it is assumed that the positions, orientations, and orientations of the cameras A and B are known, and each camera ID is associated with a color that is a feature of the camera. First, the camera information database is initialized. In the camera information database, as shown in FIG. 12A, since the positions, orientations and orientations of the cameras A and B are known, those values are written, and the colors are assigned to the camera IDs as camera features. Write in advance.

  Since the positions, orientations and orientations of the cameras A and B are known, the video data of the cameras A and B are read, and the camera video portion is detected from the video screen. In camera A, purple, orange and white cameras are photographed as shown in FIG. 13 (a), and in camera B, red, white, orange, yellowish green and black are photographed as shown in FIG. 13 (b). The camera is photographed.

  Comparing the color of the camera video in each video screen, since the white and orange cameras exist in both video screens, the positions of these two cameras are calculated. When camera IDs of white and orange camera characteristics are searched from the camera information database, they are identified as camera D and camera E, so that their position information is updated and camera A is assigned to each of the used camera IDs. And camera B are stored. As a result, the data shown in FIG. 12B is stored in the camera information database.

  Since the cameras D and E are cameras in which only the installation positions are calculated, the direction and orientation are calculated next. In the video screen of the camera D, red, purple, orange, blue and green cameras are photographed as shown in FIG. When the camera ID of each color is searched from the information database for camera IDs whose installation positions are known, they are identified as cameras A, B, and E, and the orientation and orientation are calculated using the position of any camera, Store in the camera information database. As a result, the orientation / attitude values of the camera D are stored in the camera information database as shown in FIG.

  The same processing is performed on the video screen of the camera E, and the orientation / attitude of the camera E is calculated. As a result, the positions, orientations, and orientations of the cameras A, B, D, and E are calculated. Next, by using the images of the four cameras, the same camera image captured on the image screen is detected, and the position, orientation, and orientation of another camera in the vicinity are further calculated and estimated.

  By repeatedly performing this process as shown in FIG. 15, the positions, orientations, and orientations of all cameras can be calculated and estimated. In the initial state, as shown in FIG. 15A, the positions, orientations and orientations of the cameras A and B are known. From this state, as shown in FIG. 15B, the positions of the cameras D and E are calculated from the images taken by the cameras A and B, and the orientations and postures of the cameras D and E are taken by the cameras D and E. Calculated from the image of the camera A or B.

  Next, as shown in FIG. 15 (c), the positions of the cameras G and H are calculated from the images taken by the cameras D and E, and the directions and orientations of the cameras G and H are taken by the cameras G and H. It is calculated from the video of the camera D or E. Further, the positions of the cameras C and F are calculated from the images taken by the cameras B and E, and the orientations and orientations of the cameras C and F are calculated from the images of the cameras B and E taken by the cameras C and F. Next, as shown in FIG. 15 (d), the position of the camera I is calculated from the images taken by the cameras E and F, and the orientation / posture of the camera I is the image of the camera E or F taken by the camera I. Is calculated from

  Each time the position and orientation / posture values of each camera are newly calculated, a comparison with the previous value is performed. If there is no abnormality in the camera, a comparison result that matches the previous value should be obtained. If a comparison result that matches the previous value is obtained for all cameras, the contents of the camera information database are initialized, and the process of calculating the camera position, orientation, and orientation is performed again from the beginning.

  If the position or orientation / posture of any camera is different from the previous one, whether the camera's own position / orientation / posture has changed, or the camera used to calculate the camera's position, orientation / posture To determine if the video is normal. For example, as shown in FIG. 16, it is assumed that the position value of the camera I is deviated from the previous value by the current position calculation, and the images of the camera F and the camera E are used to calculate the position of the camera I.

  When it is determined that the value of the position of the camera I is different from the previous time, the video positions of the cameras whose positions are known, which are taken by the cameras E and F, which are the camera IDs used by the camera I, are investigated. For example, the position of the camera C is calculated by the combination of the camera F and the camera B, and the position of the camera B is calculated by the combination of the camera E and the camera D.

  That is, the position of the camera C is calculated from the images taken by the camera F and the camera B. If the value is different from the previous value, it is determined that the image of the camera F is abnormal. When the position of the camera C is the same as the previous time, the camera F is not abnormal and the camera E is investigated. Therefore, the position of the camera B is calculated from the images taken by the camera E and the camera D, and when the value is different from the previous value, it is determined that the image of the camera E is abnormal. When the position of the camera B is the same as the previous time, it is determined that the camera E is not abnormal and the camera I is abnormal.

  The combination of cameras used for video investigation is not limited to the above-described combination, and may be configured to investigate by a plurality of combinations instead of examining only one type. In the example of FIG. 16, since the position of the camera I is shifted, the positions of the other cameras calculated using the camera E and the camera F are all correct values.

  As described above, when the position of another camera is calculated and the position is the same as the previous time, it is determined that the camera I is abnormal, and a notification to that effect is sent to the management center or the like. If there is an abnormality in the video of the camera E, the position of the other camera calculated using the camera E is also different from the previous result, so that it is determined that there is an abnormality in the camera E.

  In this surveillance camera system, not only when the installation position or orientation / posture of the camera fluctuates, but also when the camera image is abnormal, such as when the camera is zoomed or the camera's field of view is hidden, It is possible to identify a camera in which an abnormality has occurred and notify the camera ID of the camera ID.

  Also, in the surveillance camera system, since the camera itself plays the role of a marker, there is no need to newly install a marker, and if it is detected that the camera has moved, the camera that is shooting the camera or the camera is shot. Since there is an abnormality in the camera, which is confirmed based on the video of another camera, the camera abnormality can be determined with high accuracy.

  In addition, it is possible to identify which camera has an abnormality, and to display the abnormal camera with another camera, so the current state of the camera can be grasped from the image, and the cause of the camera abnormality can be elucidated from the past image. It becomes possible. In addition, when applied to a surveillance camera system that can automatically correct the installation position or orientation / posture to calculate the installation position or orientation / posture of each camera, a deviation occurs in the camera installation position, orientation / posture. In such a case, the deviation can be automatically corrected.

6-1 Processing device 6-2 Camera (imaging device)
6-11 Camera information database 6-12 Camera selection unit 6-13 Camera detection unit 6-14 Camera position calculation unit 6-15 Camera posture calculation unit 6-16 Abnormal camera detection unit

Claims (6)

A plurality of cameras having different appearance characteristics from other cameras;
A camera detection unit for detecting a camera video from the video screen shot by each camera, recognizing features on the appearance of the camera video, and identifying a camera to be shot;
A camera position calculation unit that calculates an installation position of the camera to be photographed from video positions of the camera to be photographed on the video screens of two cameras that photographed the same camera to be photographed detected by the camera detection unit;
An abnormal camera determination unit that compares the installation position of the camera to be photographed previously calculated by the camera position calculation unit with the current installation position, and determines that an abnormality has occurred in the camera installation position when both are different,
A surveillance camera system characterized by comprising:   The abnormal camera determination unit compares the installation position of the camera to be photographed previously calculated by the camera position calculation unit with the current installation position, and when the two are different, the two cameras that photographed the camera to be photographed are compared. For the installation position of the other camera calculated by the camera position calculation unit from the video position of the other camera taken by each of the cameras, compare the installation position calculated last time with the installation position calculated this time, 2. If the two do not match, it is determined that an abnormality has occurred in the camera that has captured the video, and if both match, it is determined that an abnormality has occurred in the installation position of the camera to be photographed. The surveillance camera system as described in. The camera position calculation unit is detected by the camera detection unit, and from the video position of the camera to be photographed on the video screen of the camera that has photographed the camera to be photographed whose installation position is known, the direction of the camera that photographed the camera to be photographed・ Calculate posture,
The abnormal camera determination unit compares the previous orientation / attitude calculated by the camera position calculation unit with the current orientation / attitude of the camera that has photographed the subject camera. 3. The surveillance camera system according to claim 1, wherein it is determined that an abnormality has occurred in the camera that has photographed the photographing camera.   The surveillance camera according to any one of claims 1 to 3, wherein a camera equipped with a lens having an angle of view of at least 180 degrees is installed as the plurality of cameras directly below the ceiling portion of the surveillance area. system. Install a plurality of cameras having different appearance characteristics from each other so that at least two other cameras are photographed;
A first process of detecting a camera video from the video screen shot by each camera, recognizing features on the appearance of the camera video, and identifying a camera to be shot;
A second step of calculating the installation position of the camera to be photographed from the video positions of the camera to be photographed on the video screens of the two cameras that photographed the same camera to be photographed detected in the first process;
When the installation position of the camera to be photographed previously calculated in the second process is compared with the current installation position, and the two are different, the camera was photographed by each of the two cameras that photographed the camera to be photographed. For the installation position of the other camera calculated in the second process from the video position of the other camera, the installation position calculated last time is compared with the installation position calculated this time. A third process of determining that an abnormality has occurred in the camera that has captured the video and determining that an abnormality has occurred in the installation position of the camera to be photographed if both match,
An abnormality detection method for a surveillance camera system, comprising: In the second process, from the video position of the camera to be photographed on the video screen of the camera that photographed the camera to be photographed that has been detected in the first process and whose installation position is known, Calculate the orientation and posture,
In the third step, the previous direction / posture calculated in the second step and the current direction / posture of the camera that shot the target camera are compared. 6. The abnormality detection method for a surveillance camera system according to claim 5, wherein it is determined that an abnormality has occurred in the camera that has photographed the photographing camera.

Images