JP2011186727A - Monitoring apparatus and program - Google Patents

Abstract

<P>PROBLEM TO BE SOLVED: To provide a technique for detecting probability of the occurrence of an accident or a case in a predetermined monitoring area. <P>SOLUTION: A monitoring apparatus 1 detects a plurality of moving objects in a monitoring area from a video of a camera 2, detects the probability of the occurrence of an accident or a case based on mutual relationship of positions of the plurality of moving objects and movement information, and records the video in a recording medium while reflecting the detection results. Further, the monitoring apparatus 1 discriminates the height of the probability of the occurrence of a case based on the positions of the plurality of moving objects and the movement information. Further, the monitoring apparatus 1 detects a further detailed accident or case based on mutual relationship of the positions of the moving objects and the movement information in a certain frame in addition to the moving object information in a succeeding frame. <P>COPYRIGHT: (C)2011,JPO&INPIT

Description

  The present invention relates to a monitoring device for monitoring a predetermined monitoring area, a monitoring method, and a program for causing a computer to execute the monitoring method.

For the purpose of monitoring a predetermined place, for example, a monitoring system using a monitoring camera has become widespread.
Such a monitoring system is configured to include a camera installed to image a predetermined place (monitoring area) and a monitoring device connected to the camera via a line or the like. The monitoring device is generally realized by a computer such as a personal computer or a workstation, or a video monitor.
According to such a monitoring system, the monitoring person can check the video of the monitoring area through the monitoring device.
As an example of such a conventional monitoring device, there are those described in Patent Documents 1 and 2.

  However, in order for the supervisor to find out that an accident or incident actually occurred in the surveillance area, the supervisor had to always watch the video, which was a heavy burden. Therefore, it is conceivable that the video of the camera is stored in a recording medium such as a hard disk recorder, and the supervisor looks back on the stored video. The burden is the same because you have to look at it.

  Therefore, Patent Document 3 discloses a technique for detecting and reporting an abnormal behavior of a moving body (for example, a walking human) in a camera image.

JP 2004-266633 A JP 2003-259349 A JP 2005-353016 A

  However, the technique disclosed in Patent Document 3 focuses on the way of movement of one moving body, and detects the way of moving that is considered abnormal. For example, when the moving body is a human, it detects an abnormal way of walking. It is difficult to detect behaviors that occur between multiple mobile objects, such as accidents and incidents where one person harms another person.

  An object of the present invention is to provide a technique capable of detecting the possibility that a predetermined action has occurred between a plurality of moving bodies such as accidents and incidents in a predetermined monitoring area in the monitoring device as described above. To do.

  The monitoring apparatus according to the present invention includes a moving body detection unit that detects a moving body within a predetermined monitoring region, and a position of the moving body detected by the moving body detection unit on coordinates corresponding to the monitoring region. When a plurality of moving bodies are detected by the position deriving means for deriving and the moving body detecting means, and the positions of the respective moving bodies are derived by the position deriving means, Detecting means for detecting a predetermined behavior among a plurality of moving bodies.

  The program according to the present invention is a program for causing a computer to execute a process of monitoring a predetermined monitoring area, and the process includes a moving object detection step of detecting a moving object in the monitoring area; A position deriving step for deriving the position of the moving body detected in the moving body detecting step, and a plurality of moving bodies are detected in the moving body detecting step, and the position of each moving body is derived in the position deriving step; And a detecting step of detecting a predetermined behavior among the plurality of moving bodies based on the mutual relationship between the positions of the moving bodies.

  According to the present invention, it is possible to detect the possibility that a predetermined action has occurred between a plurality of moving bodies such as accidents and incidents within a predetermined monitoring area.

FIG. 1 is a diagram illustrating a monitoring system. (Embodiment 1) FIG. 2 is a diagram illustrating a processing unit of the monitoring apparatus. (Embodiment 1) FIG. 3 is a diagram illustrating the position and movement information of the moving body. (Embodiment 1) FIG. 4 is a diagram showing the occurrence of an accident or incident. (Embodiment 1) FIG. 5 is a flowchart showing processing of the monitoring apparatus. (Embodiment 1) FIG. 6 is a diagram showing the recorded contents of the recording unit 12. (Embodiment 1) FIG. 7 is a diagram illustrating a display example of the display unit 11. (Embodiment 1) FIG. 8 is a diagram illustrating a processing unit of the monitoring apparatus. (Embodiment 2) FIG. 9 is a diagram illustrating discrimination between the approach degree and the approximation degree. (Embodiment 2) FIG. 10 is a diagram showing the recorded contents of the recording unit 12. (Embodiment 2) FIG. 11 is a diagram illustrating a display example of the display unit 11. (Embodiment 2) FIG. 12 is a diagram illustrating a processing unit of the monitoring apparatus. (Embodiment 3) FIG. 13 is a diagram illustrating the occurrence of a collision accident. (Embodiment 3) FIG. 14 is a diagram illustrating the occurrence of a flank accident. (Embodiment 3) FIG. 15 is a diagram illustrating the occurrence of a drag accident. (Embodiment 3) FIG. 16 is a diagram illustrating a processing unit of the monitoring apparatus. (Embodiment 4) FIG. 17 is a diagram illustrating the position and movement information of the moving body. (Embodiment 4) FIG. 18 is a diagram illustrating a monitoring system. (Supplement) FIG. 19 is a diagram illustrating a monitoring system. (Supplement) FIG. 20 is a diagram illustrating a monitoring system. (Supplement) FIG. 21 is a diagram illustrating a monitoring system. (Supplement)

<Embodiment 1>
The monitoring apparatus according to the first embodiment detects a plurality of moving objects in the monitoring area from the video of the camera, and detects the possibility of occurrence of an accident or an incident based on the correlation between the positions of the plurality of moving objects and movement information. Then, the video is recorded on the recording medium reflecting the detection result.

  As shown in FIG. 1, a monitoring device 1 is connected to a camera 2 via a network (NW), and constitutes a monitoring system as a whole.

  The camera 2 is installed in a predetermined monitoring area such as a street or near the entrance of a building. The camera 2 captures an image of the monitoring area and transmits the obtained image to the monitoring apparatus 1 via the network.

  The monitoring device 1 receives video from the camera 2 via the network, and executes predetermined processing, video display, video recording, and the like, which will be described later. Specifically, the monitoring device 1 is realized by a computer such as a personal computer or a workstation. The monitoring device 1 roughly includes a display unit 10, a processing unit 11, and a recording unit 12. For example, the processing unit 10 reads a program written in a ROM (not shown) and processes the video of the camera 2. The display unit 11 displays the video processed by the processing unit 10, and is specifically realized by a video monitor or the like. The recording unit 12 records the video processed by the processing unit 10, and is specifically realized by a hard disk recorder or the like.

  The network includes various types such as a communication line such as a LAN and an Internet network, a coaxial line, and the like.

  FIG. 2 is a block diagram illustrating functions of the processing unit 10 of the monitoring device 1.

  As shown in FIG. 2, the processing unit 10 first performs display processing for each frame on the video transmitted from the camera 2 (100). Here, for example, if the video is digitally encoded and transmitted from the camera 2, the digitally encoded video is decoded frame by frame so that it can be reproduced as a video. The display-processed video is sent to the display unit 11.

Further, the processing unit 10 detects a moving body for each frame of the display-processed video (101), and derives the position of the moving body in units of frames (102). Here, first, for each frame, a moving body existing in the frame is detected. The position of the detected moving body on the coordinates corresponding to the real world including the monitoring area where the camera 1 is actually installed is derived by the calibration technique.
Here, when the moving part detection is started in a certain frame when the processing unit 10 performs the moving body detection over successive frames, the detection is identified as one detection until the detection of the detected moving body is completed. For example, when the moving object “a” has entered the monitoring area, detection of the moving object “a” is started in the Nth frame (N is a natural number), and the detection of the moving object “a” continues until the (N + 100) th frame. , Detection of the moving object a over N to (N + 100) frames is regarded as one detection. Specifically, identification information such as an ID may be added to the detection of a group.
Similarly, when the processing unit 10 derives the position of the moving object over successive frames, the position of the moving object in a plurality of frames identified as one detection as a lump when detecting the moving object is taken as a group. Identify as a position group. For example, if the detection of the moving object a over N to (N + 100) frames is identified as a group as described above, the continuous position of the moving object a over N to (N + 100) frames is represented by the position group. Identify as. Specifically, identification information such as an ID may be attached to a group of positions.

Next, the processing unit 10 derives movement information such as a moving direction and a moving speed of the moving body by taking the history of the position of the moving body derived for each frame over a plurality of continuous frames (103). ). Here, the moving direction on the coordinates can be obtained from the difference between the position of the moving body on a certain frame and the position on the subsequent frame. Further, the moving speed can be obtained from the position difference between frames and the frame transition time.
There are various methods for detecting a moving body, such as a method based on a change in luminance between frames. Similarly, there are various calibration methods such as a projective transformation method, but they are well known and will not be described in detail. These techniques are also described in the following patent publications that have been filed and published by the present applicant.
-JP 2004-110280 A-JP 2005-003377 A-JP 2006-245795 A-JP 2009-210331 A

FIG. 3 is a diagram visualizing an example of the position and movement information created by the processing unit 10, and a map showing an example of the position and movement information when two moving bodies a and b are detected M is a map on coordinates corresponding to the real world including the monitoring area where the camera 2 is installed, and the monitoring area of the camera 2 is indicated by S. Here, the position Pa of the moving object a and the position Pb of the moving object b in the current frame derived by the processing unit 10 are plotted. Further, in order to visualize the position history of the moving object a and the moving object b, a movement history Ha is created by connecting the positions of the moving object a derived in the past plural frames, and similarly the position of the moving object b To create a movement history Hb. Further, for example, the moving directions of the moving bodies a and b derived (103) as movement information are indicated by arrows, and the moving speeds of the moving bodies a and b are indicated by numerical values.
In this figure, both the moving bodies a and b are moving in substantially the same direction (upward in the drawing), and are moving at 2 m / s and 3 m / s, respectively. .

  When the positions of the plurality of moving objects are derived within the frame, the processing unit 10 compares the positions of the plurality of moving objects and the movement information for each frame (104), and based on the comparison result. Then, an accident or incident caused by any of the plurality of moving objects is detected (105).

  The processing unit 10 sets identification information (flag) indicating that an accident or incident may have occurred for a frame in which an accident or incident is detected, and records the video in the recording unit 12.

Note that the processing unit 10 records a frame in which no accident or incident has been detected in the recording unit 12 without setting a flag.
The processing unit 10 can read the video recorded in the recording unit 12 and display the video on the display unit 11.

FIG. 4 is a diagram for explaining the detection (105) of an accident or incident based on the relative relationship between the position of the moving body and the moving information, and shows an example in which two moving bodies a and b are detected.
As shown in FIG. 4A, in a certain frame, when the position Pa of the moving object a and the position Pb of the moving object b are separated from each other by a predetermined distance or more, both the moving object a and the moving object b On the other hand, no accidents or incidents that cause harm are not detected, and no accidents or incidents are detected.
As shown in FIG. 4B, the position Pa of the moving object a and the Pb of the moving object b are within a predetermined distance, and the moving direction of the moving object a is different from the moving direction of the moving object b. Detect accidents. This is because there may be an accident in which the moving body a and the moving body b approach and collide.
As shown in FIG. 4 (c), the position Pa of the moving body a and the position Pb of the moving body b are within a predetermined distance, and the moving body a and the moving body b are in the same moving direction. When there is a predetermined difference in moving speed between the moving body a and the moving body b, the incident is detected. This is because there is a possibility that a “snacking” incident has occurred for the one with the fast moving speed to the other with the slow moving speed.

FIG. 5 is a flowchart illustrating a process in which the processing unit 10 detects an accident or an incident based on the mutual relationship between the moving objects.
When a plurality of moving objects are detected, the processing unit 10 determines whether or not the distance between the moving objects is within a predetermined distance (S10).
If the distance between the moving bodies is a predetermined distance or more (S10: N), no accident or incident is detected. If it is within the predetermined distance (S10: Y), the processing unit 10 further determines the moving directions of both moving bodies (S11).
If the moving directions of the two moving bodies are different (S11: N), the processing unit 10 detects an accident that there is a possibility of collision between the moving bodies (S12).
If the moving directions of both moving bodies are substantially the same (S11: Y), the processing unit 10 further determines the moving speed difference between the both moving bodies (S13).
If there is no difference between the moving speeds of the two moving bodies (S13: N), the processing unit 10 does not detect an accident or incident, but if there is a difference greater than the predetermined (S13: Y), the faster moving body. A case is detected as a possibility that a slow moving body has been snatched (S14).

  In this way, the processing unit 10 can detect the possibility that an accident or incident has occurred between the moving bodies based on the mutual relationships (mutual distance, moving direction, moving speed) of the moving bodies.

As illustrated in FIG. 6, the processing unit 10 sets a flag indicating that the frame in which an accident or incident has been detected, and associates the frame date, time, and the video (target video) including the frames before and after the frame. To the recording unit 12. The target video is a video composed of a plurality of frames that are adjacent to a frame in which the occurrence of an accident or incident is detected and from which movement information is derived. More specifically, it is an image made up of a plurality of frames identified as a group of positions by the processing unit 10 as a group at the time of position derivation (102).
For example, in this figure, a frame in which occurrence of a collision is detected in a frame at 07:00:15 on February 15, 2010 is identified as one position group, from 07:00:02 to 07:01:18. Are associated with a video composed of a plurality of frames.
In this figure, a flag indicating the possibility of an accident (collision) is set to 1, and a flag indicating the possibility of an incident (spotting) is set to 2.
FIG. 7 is a diagram illustrating a display example when the processing unit 10 reads out the video recorded in the recording unit 12 and displays the video on the display unit 11.
On the display unit 11, a list L of frames with flags of accidents and incidents is displayed, and a window W for displaying an image is displayed next to the list.
Since the flagged frames are displayed in a list, the supervisor can quickly find a video part where an accident or an incident may have occurred without looking at the video from the beginning to the end.
When the monitor performs an operation of selecting a desired frame from the frame list L, the processing unit 10 reads out and reproduces the target video including the selected frame from the recording unit 12 in response to the selection operation, and the window W To display.
In this way, when the monitor selects a desired frame where an accident or incident may have occurred, the moving object included in the desired frame (the moving object that may have caused the accident or incident) The behavior in the monitoring area can be displayed, and the supervisor can quickly clarify the accident or incident. In other words, the target video is a video from when the moving object resulting from the occurrence of the accident or incident appears in the monitoring area until it disappears, so the supervisor can confirm the situation in which the accident or incident has occurred.

<Embodiment 2>
In the monitoring apparatus according to the first embodiment, the occurrence of an accident or an incident is determined based on the position and movement information of a plurality of moving bodies. In the second embodiment, the probability of the occurrence of an incident is determined based on the position and the movement information. Determine the height.

  The monitoring device according to the second embodiment is obtained by changing the processing unit 10 of the monitoring device according to the first embodiment to a processing unit 20. Since there is no particular difference in the configuration other than the processing unit 20, the processing unit 20 will be described in detail here.

The processing unit 20 performs frame display processing (200), mobile object detection (201), position derivation (202), mobile object information derivation (203), and comparison of position and movement information (204), and then an incident occurs. Is detected (205), and identification information (flag) is set in the frame according to the detection result (206).
The processing unit 10 according to the first embodiment executes frame display processing (200), moving object detection (201), position derivation (202), derivation of moving object information (203), and comparison of position and movement information (204). Frame display processing (100), moving object detection (101), position derivation (102), moving object information derivation (103), and position and movement information comparison (104).

  In the detection of the probability of occurrence of an incident (205), the processing unit 20 moves based on the degree of proximity between the moving bodies based on the derived positions of the plurality of moving bodies and the movement information on the plurality of derived moving bodies. The probability of occurrence of the incident is determined according to the moving direction between the bodies and the approximate degree of moving speed.

FIG. 9 shows an example of a table for the processing unit 20 to determine the degree of approach and the degree of approximation.
As illustrated in FIG. 9, the processing unit 20 assigns points based on the degree of approach and the degree of approximation, and determines the probability of occurrence of an incident according to the overall point height.
In the example shown in FIG. 9, regarding the degree of approximation of the distance between a plurality of moving objects, 2 points are given if they are within 1 m of each other, 1 point is given if they are within 1 m to 3 m, and 0 points are given if they are 3 m or more. In addition, regarding the degree of approximation of the moving direction between a plurality of moving bodies, 2 points if within ± 30 ° of each other, 1 point if within ± 30 ° to ± 45 °, 0 points if over ± 45 °. Give. Moreover, about the approximation of the moving speed between several moving bodies, if it is 2.5 m / h or more mutually, 2 points will be provided, and if it is less than 2.5 m / h, 1 point will be provided.
After giving points, the processing unit 20 sets the probability to 0 (zero), for example, when there is 0 point in either one of the proximity of the distance and the approximation of the moving direction, and does not determine that the incident has occurred. On the other hand, if there is no 0 point, the total of all points is considered as a high probability. The processing unit 20 assigns the level of probability as identification information (flag) to the frame from which the probability of occurrence of the event is derived in this way (206), and records the video in the recording unit 12.

FIG. 10 is a diagram illustrating an example in which the probability of occurrence of an incident is assigned as identification information.
In the frame at time 07:00:15, the probability of occurrence of the incident is determined to be 4. According to the table of FIG. 9, for example, the distance between moving bodies is within 1 m to 3 m (1 point), and the difference in moving direction is within ± 30 ° to ± 45 ° (1 point). This corresponds to a case where the speed difference is 2.5 m or more (2 points).
In the frame at time 07:22:09, the probability of occurrence of the event is determined to be 6. According to the table of FIG. 9, for example, the distance between moving bodies is 1 m or more (2 points), the difference in moving direction is within ± 30 ° (2 points), and the moving speed difference is 2.5 m. This corresponds to the case where there are (2 points).
Note that when the processing unit 20 adds a flag and records the video in the recording unit 12, the processing unit 20 records the video in association with the video including the frames before and after the flagged frame (target video) in the same manner as in the first embodiment. is there.

  In this way, the processing unit 20 can determine the probability of occurrence of an incident based on the degree of proximity of the position between a plurality of moving objects and the degree of approximation of the moving direction and moving speed, and the supervisor can determine whether the incident has occurred. It is possible to efficiently search for a scene with the possibility of

FIG. 11 is a diagram illustrating a display example when the processing unit 20 reads out the video with the flag in this way from the recording unit 12 and displays the video on the display unit 11.
Although the frame list L and the window W are displayed on the display unit 11 and the target video associated with the frame selected from the frame list L is displayed on the window W, as in the first embodiment, As shown in the figure, in the frame list L, a list is displayed in order from frames having a high probability of occurrence of an incident.
By doing so, the observer can check the video from a scene with a high probability of occurrence of the incident, so that the criminal investigation and the like can be performed more efficiently.

<Embodiment 3>
Although the monitoring device according to the first and second embodiments detects an accident or an incident based on the correlation between the position and movement information between moving bodies in a certain frame, in the third embodiment, the monitoring apparatus between the moving bodies in a certain frame In addition to the correlation between position and movement information, more detailed accidents or incidents are detected based on moving body information in subsequent frames.

FIG. 12 is a diagram illustrating a configuration of the processing unit 30 according to the third embodiment. The frame display process (300), the moving object detection (301), the position derivation (302), the moving object information derivation (303), and the comparison between the position and the movement information (304) of the processing unit 30 are the processing units according to the first embodiment. 10 display processing (100), moving object detection (101), position derivation (102), moving object information derivation (103), comparison of position and movement information (104), but the subsequent processing is different. Yes.
The processing unit 30 compares the derived positions of a plurality of moving objects and movement information (movement direction, movement speed) for a certain frame (304), and if there is a possibility that an accident or an accident occurs as a result of the comparison, Then, movement information in a plurality of subsequent frames is derived (305), and a detailed accident or incident is detected according to the derived result (306).

  13-15 is a figure for demonstrating the detection (304-305) of an incident or an accident, and has shown the example in which the two mobile bodies a and the mobile bodies b were detected.

1. Detection of Collision Accident As shown in FIG. 13, first, if the position Pa of the moving body a and the position Pb of the moving body b are within a predetermined distance in a certain frame, there is a possibility that a collision accident has occurred in this frame. Yes (304). In this case, the positions and movement information of the moving bodies a and b are derived in the following frame in the same manner as 300 to 303 (305), and the moving speeds of both the moving bodies a and b are remarkably 0 (zero). If it is close to, it is detected that a collision accident has occurred between the moving body a and the moving body b (306). For example, when it is detected that a collision accident may have occurred in the Nth frame (N is a natural number), the moving speeds of both moving bodies are determined for the (N + 1) to (N + M) (M is a natural number) frames. If it is detected that the moving speeds of both moving bodies are close to 0, it is detected that a collision accident has occurred in the Nth frame, and a flag indicating that fact is set in the Nth frame (307).
The reason for this determination is that if a collision accident has occurred, there is a high probability that both moving bodies that have moved up to that point will stop.

2. As shown in FIG. 14, when a position Pa of the moving body a and a position Pb of the moving body b are within a predetermined distance in a certain frame, there is a possibility that a collision accident has occurred in this frame. Yes (304). In this case, the positions and movement information of the moving body a and the moving body b are derived in the following frame in the same manner as 300 to 303 (305), and the moving speed of the moving body a becomes remarkably close to 0 (zero). If the moving speed of the moving body b is other than 0, it is detected that the moving body b has escaped after hitting the moving body a (whether it has run away) (306). For example, when it is detected that a collision accident may have occurred in an Nth frame (N is a natural number), the moving speeds of both moving bodies are set for (N + 1) to (N + M) (M is a natural number) frames. If the moving speed of the moving body “a” is close to 0 and the moving speed of the moving body “b” is other than 0, it is detected that an accident has occurred in the Nth frame, and this is indicated in the Nth frame. Is set (307).
The reason for this decision is that in the case of a whispering accident, the victim is severely injured on the spot or ceases to move, but the perpetrator is likely to leave the spot.

3. Detection of dragging accident As shown in FIG. 15, first, in a certain frame, when the position Pa of the moving body a and the position Pb of the moving body b are within a predetermined distance, there is a possibility that a collision accident has occurred in this frame. Yes (304). In this case, the positions and movement information of the moving object a and the moving object b are derived in the following frame in the same manner as 300 to 303 (305), and the moving information of the moving object a is lost, that is, the moving object is not detected. When the movement information is obtained only for the moving body b and the moving speed is other than 0, it is detected that the moving body b has escaped while dragging the moving body a after colliding with the moving body a (306). For example, when it is detected that a collision accident may have occurred in the Nth (N is a natural number) frame, the movement information of both moving bodies is obtained for the (N + 1) to (N + M) (M is a natural number) frames. If the movement information of the moving object a is not detected and the moving speed of the moving object b is other than 0, it is detected that a drag accident has occurred in the Nth frame, and this is indicated in the Nth frame. A flag is set (307).
The reason for this determination is that in the case of a dragging accident, the victim is caught in the vehicle of the perpetrator and dragged on the spot, so there is a high probability that both moving bodies are detected as one.

As described above, the processing unit 30 can detect a more detailed accident or incident based on the moving object information in the subsequent frame in addition to the correlation between the position and moving information between moving objects in a certain frame. .
Also in this embodiment, the frame in which the flag indicating that an accident or an incident has occurred is a frame adjacent to the frame in which the occurrence of the accident or incident is detected in the recording unit 12, and the movement information is derived. It is the same as in the first and second embodiments that it is recorded in association with the target video composed of a plurality of frames.
<Embodiment 4>
Although the monitoring device according to the first and second embodiments detects an accident or an incident based on the correlation between the position of the moving body in a certain frame and the movement information, in the third embodiment, the accident or the incident in the frame is further detected. Detect detailed accidents or incidents based on the location where.

FIG. 16 is a diagram illustrating a configuration of the processing unit 40 according to the fourth embodiment. The frame display process (400), the moving object detection (401), the position derivation (402), the moving object information derivation (403), and the comparison between the position and the movement information (404) of the processing unit 40 are the processing units according to the first embodiment. 10 display processing (100), moving object detection (101), position derivation (102), moving object information derivation (103), comparison of position and movement information (104), but the subsequent processing is different. Yes.
The processing unit 40 compares the derived positions of a plurality of moving objects and movement information (movement direction, movement speed) for a certain frame (404), and if there is a possibility that an accident or an accident occurs as a result of the comparison, Then, the coordinates of the derived position of the moving body are compared (405), and the details of the accident or incident are detected based on the comparison result (406).

FIG. 17 shows a map on coordinates corresponding to the real world including the monitoring area where the camera 2 is installed. In the map, a broken line represents an area corresponding to a pedestrian crossing existing in the monitoring area. When a moving body is detected at coordinates in this area, the moving body is highly likely to be a pedestrian who passes a pedestrian crossing in the monitoring area. On the contrary, the mobile body detected by the coordinates outside the area indicated by the broken line is highly likely to be a vehicle that travels (roadway) other than the pedestrian crossing in the monitoring area.
Therefore, the monitoring device 1 stores in advance the real-world coordinates of an area (walking area) corresponding to a place where a pedestrian is likely to walk in the monitoring area. The storage location may be the recording unit 12, or storage means such as ROM or RAM may be provided in the monitoring apparatus 1 and stored therein.

The processing unit 40 compares the position of the plurality of moving objects and the movement information (404), and determines that an accident has occurred, for example, if the positions are within a predetermined distance. Are compared with the stored coordinates (405). As a result of comparison, if it is included in the stored coordinates, it is highly likely that an accident occurred at a pedestrian crossing in the surveillance area, so it is judged that there is a high probability of a contact accident between a pedestrian and a vehicle, and this is detected. (406). On the other hand, if it is not included in the stored coordinates as a result of the comparison, it is highly probable that an accident has occurred on the roadway in the monitoring area, so it is determined that there is a high probability of a contact accident between vehicles, and this is detected. (406).
Based on the detection result, the processing unit 40 adds to the frame a flag indicating a contact accident between the pedestrian and the vehicle or a flag indicating a contact accident between the vehicles (407).
In this way, by determining the occurrence position of the accident or incident on the real-world coordinates corresponding to the monitoring area, it is possible to detect the occurrence of a more detailed accident or incident.
Also in this embodiment, the frame in which the flag indicating that an accident or an incident has occurred is a frame adjacent to the frame in which the occurrence of the accident or incident is detected in the recording unit 12, and the movement information is derived. It is the same as in the first to third embodiments that is recorded in association with a target video composed of a plurality of frames.

<Supplement>
As mentioned above, although the monitoring apparatus was demonstrated based on Embodiment 1-4, it supplements as follows about the structure of the monitoring apparatus of this invention.
1. In the derivation of the position of the moving object (102 to 402), the processing units 10 to 40 are very temporary or steep if the moving averaged position is derived based on the position detection result in the past frame. Can derive the position smoothly without being too sensitive to the movement of the moving body.
2. When the processing units 10 to 40 detect a moving body (101 to 401), person detection that focuses on a person walking image (silhouette) may be used. Such moving body detection is disclosed in, for example, Japanese Patent Application Laid-Open No. 2010-038835.
3. Although the monitoring device 1 includes the recording unit 12 and the processing units 10 to 40 record the video in the recording unit 12, the present invention is not limited thereto. For example, as shown in FIG. 18, the monitoring device 1 includes a notification unit 22, and when the processing units 10 to 40 detect an accident or an incident (105, 205, 306, 406), the notification unit 22 may be notified. For example, the reporting unit 22 is a driver that emits a warning sound such as an alarm, and may be any one that can cause a warning sound to sound on an external speaker.
4). The monitoring device 1 is connected to the camera 2 and the processing units 10 to 40 detect the moving body in the monitoring area based on the captured image of the camera 2 (101 to 401), but the present invention is not limited to this. For example, as shown in FIG. 19, the monitoring apparatus 1 is connected to a laser scanner (laser sensor) 3 in a monitoring area instead of a camera, and the processing units 10 to 40 are based on the scan result of the laser scanner 3. May be detected.
The laser scanner 3 is a device that irradiates laser light sequentially from one end to the other end in the scan area A, and detects that the laser light strikes an object in the scan area A. For example, when the laser scanner 3 detects an object, the processing units 10 to 40 detect this as a moving body (101 to 401), and derive the position of the moving body on the real world coordinates corresponding to the scan region A. (102-402).
When configured in this way, as described in Supplement 3 above, the monitoring device 1 is provided with a notification unit so that the processing units 10 to 40 are notified from the notification unit in response to the detection of an accident or incident. do it.
5. For the sake of convenience, the monitoring device 1 and the camera 2 are connected via a network. However, the monitoring device 1 and the camera 2 are not limited to a so-called Internet network or a communication line such as a LAN. Any means capable of exchanging signals between the monitoring apparatus 1 and the camera 2 such as a wired connection means such as a coaxial line or a wireless connection means such as Bluetooth (registered trademark) or infrared rays may be used. The connection with the laser scanner 3 shown in the above supplement is the same.
6). Although an example in which the monitoring device 1 is connected to the camera 2 via the network has been shown, a part or all of the configuration of the monitoring device 1 and the camera 2 are stored in the same housing, and the camera 2 has the functions of the monitoring device 1. Some or all of them may be executed. For example, as shown in FIGS. 20A and 20B, the camera 2 includes a processing unit 10, detects an accident or incident on the camera 2 side, and the monitoring device 1 displays and records depending on the detection result. Alternatively, the camera 2 may include the processing unit 10, the display unit 11, and the recording unit 12, and the camera 2 may completely function as a monitoring device.
In addition, as shown in FIG. 21, some functions of the processing unit may be executed by the camera 2 and the remaining functions may be executed by the monitoring device 1. In the example illustrated in FIG. 21, the camera 2 executes the moving body detection process 101 among the processes of the processing unit 10 and the monitoring apparatus 1 executes the remaining functions. However, the present invention is not limited to this.

<Summary>
As described above, the monitoring device 1 according to the embodiment of the present invention can detect an accident or incident between a plurality of moving bodies that occur in the monitoring area.
In particular, since the detection is performed based on the position and movement information (movement direction difference, movement speed difference) between a plurality of moving bodies, it is necessary to create a database of movement types when the moving body has caused an accident or incident. The monitoring apparatus 1 can be configured with a simple configuration.

  The monitoring apparatus of the present invention can be widely applied to a monitoring apparatus that monitors a moving body in a predetermined monitoring area.

DESCRIPTION OF SYMBOLS 1 Monitoring apparatus 2 Camera 3 Laser scanner 10 Processing part 11 Display part 12 Recording part 100, 200, 300, 400 Frame display process 101, 201, 301, 401 Moving body detection process 102, 202, 302, 402 Position derivation process 103, 203, 303, 305, 403 Movement information derivation processing 104, 204, 304, 404 Position / movement information comparison processing 105, 306, 406 Accident / incident detection processing 106, 206, 307, 407 Flag assignment processing 205 Incident occurrence probability detection processing 405 Coordinate comparison process

Claims (8)

A moving body detecting means for detecting a moving body within a predetermined monitoring area;
Position deriving means for deriving a position of the moving body detected by the moving body detecting means on coordinates corresponding to the monitoring area;
When the plurality of moving bodies are detected by the moving body detecting means and the position of each moving body is derived by the position deriving means, the position between the moving bodies is determined based on the mutual relationship between the positions of the respective moving bodies. A monitoring device comprising: detecting means for detecting a predetermined action. It further comprises input means for inputting the video imaged by the camera that images the monitoring area,
The monitoring apparatus according to claim 1, wherein the moving body detection unit detects a moving body in an image input by the input unit. The position deriving means derives the position of the moving body for each frame constituting the video,
The monitoring apparatus according to claim 2, wherein the detection unit detects the predetermined action based on a distance between positions of the plurality of moving bodies derived by the position deriving unit. For each moving body, further comprising movement information deriving means for deriving at least one of the moving direction and moving speed of each moving body as movement information based on the position derived by the position deriving means,
The monitoring device according to claim 1, wherein the detection unit detects the predetermined action based on a mutual relationship between positions of the plurality of moving bodies and the movement information. For each moving body, further comprising movement information deriving means for deriving at least one of the moving direction and moving speed of each moving body as movement information based on the position derived by the position deriving means,
The detection means includes an approach degree of the distance between the plurality of moving bodies determined from the position derived by the position deriving means, and a moving direction between the plurality of moving bodies determined from the movement information derived by the movement information deriving means. The monitoring apparatus according to claim 1, wherein the detection is performed by changing the probability that the predetermined action has occurred based on at least one of the approximate degrees of movement speed. Video recording means for recording video captured by the camera;
The position deriving means derives a position for each frame constituting the video;
The detecting means detects the predetermined action based on the mutual relationship between the positions of the plurality of moving bodies derived by the position deriving means,
The monitoring apparatus according to claim 2, wherein the video recording unit records video by adding identification information for identifying the frame in which the predetermined action is detected by the detection unit. Video recording means for recording video captured by the camera;
The moving body detection means starts detecting the moving body when the moving body appears in the monitoring area, and ends the detection of the moving body when the moving body leaves the monitoring area. Identify from start to end as one detection,
The position deriving means derives a position for each frame constituting the video, and identifies the position of the moving body derived while the moving body detecting means performs one detection as one position group. And
The detecting means detects the predetermined action based on the mutual relationship between the positions of the plurality of moving bodies derived by the position deriving means,
The video recording means is composed of a frame in which the predetermined action is detected by the detection means, and a plurality of positions identified by the position deriving means as one position group. The monitoring apparatus according to claim 2, wherein the plurality of frames from which the is derived are recorded in association with each other. A program for causing a computer to execute a process for monitoring a predetermined monitoring area, wherein the process includes:
A moving body detecting step of detecting a moving body in the monitoring region;
A position deriving step for deriving the position of the moving body detected in the moving body detecting step;
When a plurality of moving bodies are detected in the moving body detecting step and the position of each moving body is derived in the position deriving step, a predetermined value between the plurality of moving bodies is determined based on the mutual relationship between the positions of the moving bodies. And a detection process for detecting the behavior of the person.

Images