JP2021072475A - Monitoring system and monitoring system setting program - Google Patents

Abstract

To allow an observer to immediately check a specific status of a monitoring event related to an alarm issuance when the occurrence of the monitoring event is detected and the alarm is issued immediately after the alarm is issued, or even in a case in which the chack is delayed from the alarm issuance.SOLUTION: In a case in which an image analysis server 3 detects the occurrence of a predetermined monitoring event on the basis of a real-time camera image taken by a camera in a monitoring area, and when the monitoring event meets a predetermined alarm issuance condition, the image analysis server issues an alarm, transmits event detection information including position information of an object related to the detected monitoring event to a camera management server 2 at the same timing as the alarm issuance, and the camera management server 2 adds a mark image indicating the object related to the monitoring event to a temporarily stored camera image to generate a check image.SELECTED DRAWING: Figure 1

Description

The present invention is a monitoring system that detects the occurrence of a predetermined monitoring event and issues an alarm based on a real-time camera image of the monitoring area, and a monitoring system setting that causes an information processing device to make settings related to the monitoring system. It's about the program.

In recent years, a monitoring system in which a camera is installed at a monitoring point and a real-time camera image taken by the camera is displayed on a monitoring screen of a monitor has become widespread. Further, a monitoring system that issues an alarm to notify the observer of the occurrence of a predetermined monitoring event when the occurrence of a predetermined monitoring event is detected by image analysis of the camera image has also been put into practical use.

In such a monitoring system, when the observer recognizes the occurrence of the monitoring event by issuing an alarm, the specific status of the monitoring event can be confirmed by the camera image on the monitoring screen, but the confirmation work by the camera image is performed. If the alarm is delayed, the real-time camera image may no longer show the monitoring event, so there is a problem that the specific status of the monitoring event cannot be known.

Therefore, as a solution to such a problem, conventionally, when it is detected that a person has entered a predetermined intrusion prohibited area, the detected person and the intrusion prohibited area in which the person has entered are displayed on the monitor screen of the monitor. There is known a technique for highlighting and continuing to highlight the intrusion prohibited area even after a person leaves the intrusion prohibited area (see Patent Document 1). With this technology, even if the confirmation on the monitoring screen is delayed from the alarm issuance, the intrusion prohibited area is still highlighted, so the observer confirms that a monitoring event (entry into the intrusion prohibited area) has occurred. be able to.

Japanese Unexamined Patent Publication No. 2016-146527

However, in the conventional technology, the observer can confirm that the monitoring event has occurred by continuously highlighting the intrusion prohibited area on the monitoring screen, but on the monitoring screen, the camera image at the time of the occurrence of the monitoring event is displayed. However, since the real-time camera image is displayed, there is a problem that the specific status of the monitoring event cannot be confirmed if the confirmation work is delayed from the alarm issuance. Further, if the recorded camera image is displayed on the recorder, the specific status of the monitoring event can be confirmed. However, in the conventional technology, the camera image at the time when the monitoring event occurs is not configured to be displayed immediately. There is a problem that the troublesome work of finding the camera image at the time of the occurrence of the monitoring event from the recorder is required.

Therefore, in the present invention, when the occurrence of a monitoring event is detected and an alarm is issued, the observer can determine the specific status of the monitoring event related to the alarm issuance immediately after the alarm is issued. The main purpose is to provide a monitoring system and a monitoring system setting program that can be confirmed immediately even if the alarm is delayed.

The monitoring system of the present invention is a monitoring system that detects the occurrence of a predetermined monitoring event and issues an alarm based on a real-time camera image of the monitoring area, and at least one of the monitoring areas is photographed. A camera, an image storage unit that stores the camera image taken by the camera, and an image analysis of the camera image to detect the occurrence of the monitoring event and notify the user of the occurrence of the monitoring event. An image analysis unit that issues an alarm and generates event detection information including the position information of the detected object related to the monitoring event, and a mark image indicating the object related to the monitoring event based on the event detection information. A confirmation image generation unit that generates a confirmation image by adding the confirmation image to the temporarily stored camera image, and an image distribution unit that distributes the confirmation image to the user device at the same timing as the alarm issuance. The configuration is provided.

Further, the monitoring system setting program of the present invention sets the setting related to the monitoring system that detects the occurrence of a predetermined monitoring event and issues an alarm based on the real-time camera image taken by the camera of the monitoring area in the information processing device. It is a monitoring system setting program to be performed, a setting screen is displayed on a user device, setting information related to the monitoring system is acquired according to a screen operation of the user on the setting screen, and the setting screen is the monitoring. A camera selection unit that selects the camera to be set from a plurality of the cameras provided in the system, and a plurality of types of monitoring events that can be detected for the camera selected by the camera selection unit. The monitoring is performed by a monitoring event selection unit that selects at least one of them, an image storage unit that stores the camera image corresponding to the camera selected by the camera selection unit, and image analysis of the camera image. A setting information display unit that displays setting information related to an image analysis unit that detects the occurrence of an event and issues an alarm to notify the user of the occurrence of the monitoring event, and the camera selected by the camera selection unit. The configuration is provided with an image display unit that displays a real-time camera image taken by the camera.

According to the present invention, when the occurrence of a monitoring event is detected and an alarm is issued, a confirmation image (camera image for alarm confirmation) is displayed substantially at the same time as the alarm is issued. The observer can immediately confirm the specific status of the monitoring event immediately after the alarm is issued or even if the alarm is delayed.

Overall configuration diagram of the monitoring system according to this embodiment Sequence diagram showing the operation procedure of this monitoring system Timing diagram showing the operating status of the image analysis server 3 A block diagram showing a schematic configuration of the camera management server 2 and the image analysis server 3. Block diagram showing the functional configuration of the image analysis server 3 Explanatory drawing which shows the outline of the take-out detection process performed by the image analysis server 3. Explanatory drawing which shows the outline of the fall detection processing performed by the image analysis server 3. Explanatory drawing which shows the outline of the retention detection processing performed by the image analysis server 3. Explanatory drawing showing the outline of the intrusion detection process performed by the image analysis server 3. Explanatory diagram showing the operation status confirmation screen displayed on the maintenance terminal 7. Explanatory drawing which shows the setting screen displayed on maintenance terminal 7. Explanatory drawing which shows the main part of the setting screen when the take-out setting button 122 and the fall setting button 123 are operated. Explanatory drawing which shows the main part of the setting screen when the stay setting button 124 and the intrusion setting button 125 are operated. Explanatory drawing showing the main part of the setting screen in the area setting mode Explanatory drawing which shows the monitoring screen displayed on the monitoring terminal 6. A block diagram showing a schematic configuration of the camera management server 2 and the image analysis server 3 according to the second embodiment.

The first invention made to solve the above-mentioned problems is a monitoring system that detects the occurrence of a predetermined monitoring event and issues an alarm based on a real-time camera image of a monitoring area. The occurrence of the monitoring event is detected by at least one camera that captures the monitoring area, an image storage unit that stores the camera image captured by the camera, and image analysis of the camera image, and the monitoring event is detected. An image analysis unit that issues an alarm to notify the user of the occurrence and generates event detection information including the position information of the object related to the detected monitoring event, and the monitoring based on the event detection information. A confirmation image generator that generates a confirmation image by adding a mark image pointing to an object related to an event to the temporarily accumulated camera image, and the confirmation image to a user device at the same timing as the alarm issuance. It is configured to include an image distribution unit for distribution.

According to this, when the occurrence of a monitoring event is detected and an alarm is issued, a confirmation image (camera image for alarm confirmation) is displayed almost at the same time as the alarm is issued, so monitoring related to the alarm issuance. The specific situation of the event can be confirmed by the observer immediately after the alarm is issued or immediately after the alarm is issued.

Further, in the second invention, the image analysis unit attaches the camera to the camera based on the setting information in which the type selected in advance for each camera is registered from the plurality of types that can be detected as the monitoring event. The configuration is such that the occurrence of the corresponding type of monitoring event is detected.

According to this, since the type of the monitoring event to be detected can be selected for each camera, appropriate monitoring can be performed according to the characteristics of the monitoring area and the like.

Further, in the third invention, the image analysis unit has a configuration in which the user selects a plurality of types smaller than the maximum number of types that can be detected as the monitoring event, and registers the setting information.

According to this, it is possible to reduce the load on the behavior analysis in the server device, detect the occurrence of multiple types of monitoring events for one camera, and issue alarms for different types of monitoring events that occur in one monitoring area. You can make a report.

Further, in the fourth invention, the confirmation image generation unit uses the frame image surrounding the object related to the monitoring event as the mark image in the camera image based on the event detection information acquired from the image analysis unit. In addition, the confirmation image is generated.

According to this, the object related to the monitoring event can be confirmed immediately. In addition to the frame image, the mark image may be an image of characters, symbols, and figures (stars, arrows, etc.).

Further, in the fifth invention, the image analysis unit and the confirmation image generation unit are configured by one server device.

According to this, the configuration of the monitoring system can be simplified.

Further, the sixth invention is a monitoring in which an information processing apparatus is made to set a monitoring system for detecting the occurrence of a predetermined monitoring event and issuing an alarm based on a real-time camera image taken by a camera in the monitoring area. A system setting program that displays a setting screen on a user device, acquires setting information related to the monitoring system according to a user's screen operation on the setting screen, and provides the setting screen in the monitoring system. A camera selection unit that selects the camera to be set from the plurality of the cameras selected, and at least a plurality of types of monitoring events that can be detected for the camera selected by the camera selection unit. Occurrence of the monitoring event by a monitoring event selection unit that selects one, an image storage unit that stores the camera image corresponding to the camera selected by the camera selection unit, and image analysis of the camera image. Is taken by the setting information display unit that displays the setting information related to the image analysis unit that detects the occurrence of the monitoring event and issues an alarm to notify the user of the occurrence of the monitoring event, and the camera selected by the camera selection unit. The configuration is provided with an image display unit that displays a real-time camera image.

According to this, the settings related to the surveillance system can be efficiently made with reference to the camera.

Further, according to the seventh aspect of the present invention, the setting screen is provided with a detection area designation unit for designating a range of detection areas for which an alarm is issued when an object related to a monitoring event is located.

According to this, by setting a detection area in a part of the shooting area, it is possible to avoid unnecessary alarm issuance.

Further, according to the eighth invention, the setting screen is provided with a mask area designation unit for designating a range of mask areas to be excluded from the target of alarm issuance when an object related to a monitoring event is located.

According to this, it is possible to suppress false alarms by excluding areas where false alarms are likely to occur as mask areas from the target of alarm issuance.

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

(First Embodiment)
FIG. 1 is an overall configuration diagram of a monitoring system according to the present embodiment.

The monitoring system notifies the observer of the occurrence of a predetermined monitoring event such as an abnormal state of the user in the monitoring area in the facility such as a resort facility or a commercial facility, and the camera 1 and the camera management server 2 ( Image storage unit, confirmation image generation unit, image distribution unit), image analysis server 3 (image analysis unit), monitoring server 4, warning light 5, monitoring terminal 6 (user device), and maintenance terminal 7 (user). The device) and. These devices are connected to each other via an appropriate communication path such as a premises network.

The cameras 1 are installed at a plurality of monitoring points in the facility targeted by this monitoring system, and photograph the surroundings of the monitoring points.

The camera management server 2 stores real-time camera images (moving images) received from the camera 1. Further, the camera management server 2 distributes a real-time camera image to the monitoring terminal 6 for live use. Further, the camera management server 2 transmits a real-time camera image to the image analysis server 3.

The image analysis server 3 receives a camera image from the camera management server 2, and detects the occurrence of a monitoring event from the camera image by image analysis (behavior analysis) of the camera image. In the present embodiment, it is possible to detect a plurality of types of monitoring events, specifically, the occurrence of a maximum of eight types of abandoned, taken away, overturned, stagnant, invaded, crowd, quarrel, and lost property, and the camera 1 It is possible to select and register the type of monitoring event to be detected for each camera 1 according to the characteristics of the monitoring point which is the shooting area of the camera. In the following description, an example of selecting two types for each camera 1 from a maximum of four types is described. In this way, as the behavior analysis of the image analysis server 3, the types to be executed are limited (8 types to 4 types), or a plurality of types less than the maximum number of types that can be analyzed by the image analysis server 3 for each camera 1. By making it possible to select the type of camera 1, the load on the behavior analysis on the image analysis server 3 can be significantly reduced, and as a result, the number of cameras 1 connected to each image analysis server 3 is increased. It becomes possible to make it. The maximum number of types to be executed by the image analysis server 3 may be limited, and a plurality of types less than the maximum number may be selected for each camera 1. In any case, in the image analysis server 3 or the camera 1, it is possible to omit the analysis process for monitoring events that are considered unnecessary or infrequently used. Therefore, the larger the scale of the monitoring system, the more efficient it is. Behavioral analysis can be performed.

Here, the types of a plurality of monitoring events that can be analyzed by the image analysis server 3 will be described. Take-away is a state in which items such as equipment have been taken away. A fall is a state in which a person has fallen. Staying is a state in which a person stays in a predetermined area for a long time. An intrusion is a state in which a person has entered a predetermined area where it is necessary to pay attention to the entry of the user.

Further, in the present embodiment, the alarm issuance condition is set as necessary, and when the occurrence of the monitoring event is detected and the alarm issuance condition is satisfied, the image analysis server 3 detects the detection start information and the alarm issuance. The information is transmitted to the camera management server 2. Further, the image analysis server 3 transmits the detection result of the monitoring event, that is, the event detection information including the information regarding the position of the object (including the person) related to the monitoring event on the camera image to the camera management server 2.

Now, the camera management server 2 transfers the real-time camera image to the image analysis server 3. Further, the camera management server 2 temporarily stores real-time camera images, and when it receives event detection information from the image analysis server 3, it points to an object related to the monitoring event reflected in the camera image based on the event detection information. An analysis result image (confirmation image) is generated by superimposing the detection frame (mark image) on the camera image.

When the monitoring server 4 receives the alarm issuing information from the camera management server 2, the monitoring server 4 instructs the alarm light 5 to perform an alarm output operation. Further, the monitoring server 4 stores the alarm issuing information received from the camera management server 2.

The alarm light 5 performs an alarm output operation (lamp lighting or blinking operation) in response to an instruction from the monitoring server 4. As a result, it is possible to notify the observer of the occurrence of the monitoring event.

The monitoring terminal 6 is installed in the monitoring center and operated by an observer. The monitoring terminal 6 displays a real-time camera image transmitted from the camera 1 and received via the camera management server 2. Further, the monitoring terminal 6 receives and displays the analysis result image (confirmation image) generated by the camera management server 2.

The maintenance terminal 7 is operated by a maintenance worker. The maintenance terminal 7 performs maintenance operations related to this monitoring system. For example, it is possible to access the image analysis server 3 from the maintenance terminal 7 and check whether or not the camera 1 associated with the image analysis server 3 is operating normally.

In the present embodiment, the alarm light 5 is provided as an alarm device for notifying the observer of the occurrence of a monitoring event in response to the alarm issuance, but an alarm device that outputs an alarm sound may be provided. Further, the monitoring terminal 6 may display an alarm image in addition to the analysis result image (confirmation image) to notify the observer of the occurrence of the event of interest.

Next, the operation procedure of this monitoring system will be described. FIG. 2 is a sequence diagram showing an operation procedure of this monitoring system.

First, the camera 1 photographs the monitoring area and transmits a real-time camera image to the camera management server 2.

Next, when the camera management server 2 receives the real-time camera image from the camera 1, the live image is transferred to the image analysis server 3 as it is. Further, the camera management server 2 transmits a real-time camera image to the monitoring terminal 6 for live performance. The monitoring terminal 6 displays a real-time camera image.

Here, the camera management server 2 stores the real-time camera image as a buffer image (moving image) in its own device for a predetermined time (for example, 30 seconds or 1 minute).

Next, when the image analysis server 3 receives the real-time camera image from the camera management server 2, the image analysis (behavioral analysis) of the real-time camera image causes each monitoring event of take-away, fall, stay, and intrusion. Is detected. Then, when the alarm issuing condition is satisfied, the image analysis server 3 transmits the alarm information to the camera management server 2. This alarm information includes information on the type of detected event (take-away, fall, stay, intrusion).

Next, when the camera management server 2 receives the alarm information from the image analysis server 3, the camera management server 2 transmits the alarm information to the monitoring server 4.

Next, when the monitoring server 4 receives the alarm information from the camera management server 2, it transmits an alarm output instruction to the alarm light 5 and causes the alarm light 5 to perform an alarm output operation. Further, the monitoring server 4 stores the alarm information received from the camera management server 2 in its own device.

Further, the image analysis server 3 transmits the event detection information (metadata) to the camera management server 2 as a result of the image analysis (behavior analysis). This event detection information includes position information of an object (including a person) related to the monitoring event and time information regarding the time when the monitoring event is detected.

Further, when the camera management server 2 receives the event detection information from the image analysis server 3, the camera management server 2 receives a mark image indicating an object related to the monitoring event on the buffer image stored in the own device based on the event detection information. Specifically, a process of superimposing and drawing a rectangular detection frame surrounding an object related to a monitoring event (frame frame processing) is performed, and an analysis result image (confirmation image) is generated.

Next, the camera management server 2 transmits the analysis result image to the monitoring terminal 6. The analysis result image (confirmation image) is displayed on the monitoring terminal 6.

Next, the operating status of the image analysis server 3 will be described. FIG. 3 is a timing diagram showing the operating status of the image analysis server 3.

When the image analysis server 3 receives a real-time camera image from the camera management server 2, it detects the occurrence of a predetermined monitoring event (take-away, fall, stay, intrusion) by image analysis (behavioral analysis) of the camera image. .. Then, when a predetermined alarm issuing condition is satisfied for the detected monitoring event, the detection start information and the alarm issuing information are transmitted to the camera management server 2.

Here, the information transmitted to the camera management server 2 differs depending on the type of monitoring event. Specifically, as shown in FIG. 3A, in the case of taking away and staying, these information are transmitted to the camera management server 2 in the order of alarm issuance information and detection start issuance information. On the other hand, as shown in FIG. 3B, in the case of a fall or intrusion, only the alarm issuance information is transmitted to the camera management server 2.

Further, when the image analysis server 3 detects the occurrence of a monitoring event (take-away, fall, stay, intrusion), the image analysis server 3 transmits the event detection information to the camera management server 2.

Here, in the case of taking away and staying, event detection information regarding the detection time (the period from event detection to alarm issuance) is transmitted to the camera management server 2, and the camera management server 2 issues an alarm from the detection start time. The analysis result image is generated from the camera image in the time until the information time. On the other hand, in the case of a fall or intrusion, the event detection information regarding the camera image at the alarm issuance time is transmitted to the camera management server 2, and the analysis result image is generated. As the analysis result image, a still image of the alarm issuing time and a moving image starting from or in the middle of the alarm issuing time are generated.

Next, a schematic configuration of the camera management server 2 and the image analysis server 3 according to the first embodiment will be described. FIG. 4 is a block diagram showing a schematic configuration of the camera management server 2 and the image analysis server 3.

The camera management server 2 includes a communication unit 21, a memory 22, and a processor 23.

The communication unit 21 communicates with the camera 1, the monitoring server 4, and the monitoring terminal 6. Specifically, a real-time camera image is received from the camera 1, the real-time camera image is transferred to the image analysis server 3, and the real-time camera image is transferred to the monitoring terminal 6 for live use. Further, the alarm issuance information is received from the image analysis server 3 and the alarm issuance information is transferred to the monitoring server 4. Further, the analysis result image is received from the image analysis server 3 and the analysis result image is transferred to the monitoring terminal 6.

The memory 22 stores a program or the like executed by the processor 23. Further, the memory 22 stores the camera image received from the camera 1. Further, the memory 22 temporarily stores the real-time camera image received from the camera 1 as a buffer image. As a result, even if the processing performed by the image analysis server 3 is delayed to some extent, it is possible to appropriately perform the analysis result image generation processing or the like performed based on the processing result of the image analysis server 3.

The processor 23 performs various processes related to information collection by executing the program stored in the memory 22. In the present embodiment, the processor 23 controls the transfer of the camera image, the accumulation control of the camera image, the transfer of the alarm issuance information, the framing process, the control related to the transfer of the analysis result image, and the like. I do.

In the framing process, the processor 23 superimposes and draws a frame image (mark image) indicating an object related to the monitoring event on the camera image based on the event detection information received from the image analysis server 3, and analyzes the analysis result image ( Video) is generated. At this time, a camera image at a corresponding time is acquired from the camera image (buffer image) stored in the memory 22, and the camera image is framed.

The image analysis server 3 includes a communication unit 31, a memory 32, and a processor 33.

The communication unit 31 communicates with the camera management server 2 and the maintenance terminal 7. Specifically, the communication unit 31 receives the camera image from the camera management server 2. Further, the communication unit 31 transmits the alarm issuing information to the camera management server 2. Further, the communication unit 31 transmits the event detection information generated by the processor 33 to the camera management server 2. Further, the communication unit 31 transmits / receives information for a setting operation performed by the maintenance terminal 7.

The memory 32 stores a program or the like executed by the processor 33.

The processor 33 performs various processes related to image analysis (behavioral analysis) by executing a program stored in the memory 32. In the present embodiment, the processor 33 performs setting processing, monitoring event detection processing, notification necessity determination processing, learning update processing, and the like.

In the setting process, the processor 33 sets a detection area to be the target of the monitoring event detection process and a mask area to be excluded from the target of the alarm issuance according to the input operation of the administrator. In addition, threshold values used in various determination processes are set according to the input operation of the administrator. The mask area is an area within the detection area where there is a high possibility that a false alarm will partially occur.

In the monitoring event detection process, the processor 33 detects the occurrence of a monitoring event of a type selected and set in advance for each camera 1. In the present embodiment, for each camera image, the occurrence of two types of monitoring events selected from the take-away, fall, stay, and intrusion events are detected as monitoring events. In this way, by selecting the type for each camera 1, it is possible to perform more efficient image analysis (behavior analysis) by limiting the type to the type having a high priority at the installation location of the camera 1. .. Here, when the occurrence of a monitoring event is detected, the detection result, that is, an event including the type of the detected monitoring event (take-away, fall, stay, intrusion), and the position information of the object (equipment, person) related to the monitoring event. The detection information is transmitted to the camera management server 2.

In the alarm issuance necessity determination process, the processor 33 determines the necessity of alarm issuance based on the detection result of the monitoring event detection process according to the determination result of whether or not the predetermined alarm issuance condition is satisfied. To do. Here, when the alarm issuance condition is satisfied and it is determined that the alarm issuance is necessary, the alarm issuance information is transmitted to the camera management server 2.

In the learning update process, the processor 33 updates the machine learning model for fall detection, the machine learning model for retention detection, and the machine learning model for intrusion detection. The user (administrator) visually determines whether the alarm is a correct report or a false report based on the camera image and the analysis result image for each alarm issued in the past, and registers the result in the alarm history database. The processor 33 relearns the machine learning model using the camera image related to the alarm determined to be an erroneous report as training data, and updates the machine learning model.

Next, the functional configuration of the image analysis server 3 will be described. FIG. 5 is a block diagram showing a functional configuration of the image analysis server 3.

The image analysis server 3 is provided with a plurality of image analysis programs (for example, 12) that detect monitoring events by image analysis (behavioral analysis). This image analysis program becomes one channel that can independently process camera images. In addition, each image analysis program is configured to be able to detect only one of all types of monitoring events (take-away, fall, stay, intrusion), and one for one type of monitoring event detection process. Assigned one by one. Further, in the present embodiment, a maximum of two types of detection processing are performed on one camera 1. Therefore, a maximum of two image analysis programs are assigned to one camera 1. It should be noted that it is also possible to select and set only one type of monitoring event detection for one camera 1. In this case, one image analysis program is assigned to one camera 1.

On the other hand, if this monitoring system is a large-scale one, the number of cameras 1 far exceeding the number of image analysis programs (for example, 12) provided in one image analysis server 3 is installed. Therefore, a plurality of image analysis servers 3 are provided to share the processing of camera images.

As described above, in the present embodiment, the image analysis server 3 is configured to be able to detect a plurality of types of monitoring events, and the type of monitoring event detection to be performed is set for each camera 1. For this reason, in a large-scale monitoring system, the scale has gradually increased due to the expansion, and even when various types of cameras 1 are mixed, appropriate monitoring event detection is performed according to the characteristics of the shooting point of the camera 1. At the same time, unified control can be performed for all the cameras 1.

Next, the take-away detection process performed by the image analysis server 3 will be described. FIG. 6 is an explanatory diagram showing an outline of the take-away detection process performed by the image analysis server 3.

In the take-out detection process, a disappeared object (other than a person) is detected from a camera image as an object related to a monitoring event (object detection). Specifically, the camera image is compared with the background image acquired in advance, and the disappeared object is detected. Specifically, when the difference between the feature amount of the background and the object is equal to or larger than a predetermined value (feature amount difference threshold value), it is determined that the object has disappeared.

At this time, an object detection frame surrounding the disappeared object is set on the camera image (frame).

Next, in the take-away detection process, as a positional condition for issuing an alarm, it is determined whether or not the disappeared object is located in an effective area other than the mask area in the detection area. The detection area and mask area for take-out detection can be specified by a rectangle. The detection area and the mask area may be designated by polygons defined by a predetermined number or less of vertices.

Next, in the take-away detection process, as a time condition for issuing an alarm, a predetermined time (for example, a take-away detection time is in the range of 3 to 30 minutes) within the effective area after the disappeared object disappears. It is determined whether or not the set time value) or more has passed. Here, when the elapsed time after the disappearance of the object is taken away and the detection time is reached, an alarm is issued.

Here, a person is detected from the camera image, and if the article exists at the leg of the person based on the positional relationship between the article and the surrounding person, it is regarded as the property of the person and the alarm is issued. Exclude from the target.

In addition, if the state of being concealed by an object (moving object) such as another person continues for a predetermined time (concealment limit time, for example, 1 minute) or more before the removal detection time elapses after detecting the disappeared object. , Exclude from the target of alarm issuance. If the device returns to the non-concealed state before the concealment limit time elapses due to temporary concealment, it will not be excluded from the alarm issuance target, and if the take-away detection time elapses, it will be determined to be taken away. ..

Next, the fall detection process performed by the image analysis server 3 will be described. FIG. 7 is an explanatory diagram showing an outline of the fall detection process performed by the image analysis server 3.

In the fall detection process, a fallen person is detected from a camera image as an object related to a monitoring event (falling person detection). In the present embodiment, a person in a fall state is detected from a camera image (frame) by using a machine learning model for fall detection by deep learning or the like. Specifically, a camera image is input to the machine learning model, and a fall score (likelihood indicating the high possibility of a fall) output from the machine learning model is acquired. Then, when the fall score is equal to or higher than a predetermined value (fall score threshold value), the fall state is determined.

At this time, a fall detection frame surrounding the fallen person is set on the camera image (frame).

Next, in the fall detection process, as a positional condition for issuing an alarm, it is determined whether or not the person in the fall state is located in an effective area other than the mask area in the detection area. At this time, it is determined whether or not the reference point of the person detection frame is located in the effective area. In the present embodiment, the center point of the lower side of the fall detection frame is used as the reference point. Further, the detection area and the mask area for fall detection can be designated by a polygon defined by a predetermined number or less (for example, 8 points) of vertices. The detection area and the mask area may be designated by a rectangle.

Next, in the fall detection process, as a temporal condition for issuing an alarm, when a person in a fall state is located in the effective area, the frame in which the person in the fall state is located in the effective area is the most recent. It is determined whether or not there are a predetermined number (the number of continuous fall frames) or more in the predetermined number (the number of the latest determination frames). When this time condition is satisfied, an alarm is issued.

Therefore, even if a person in a fall state is detected, if the reference point of the fall detection frame is outside the detection area, or if the reference point of the fall detection frame exists in the detection area does not satisfy the number of continuous fall frames. Is excluded from the target of alarm issuance.

In the present embodiment, the alarm issuing condition is limited to the case where the center point of the lower side of the fall detection frame enters the detection area. For example, the condition may be that the center point of the fall detection frame has entered the detection area, or the condition may be that the fall detection frame overlaps the detection area.

Next, the retention detection process performed by the image analysis server 3 will be described. FIG. 8 is an explanatory diagram showing an outline of the retention detection process performed by the image analysis server 3.

In the retention detection process, a person is detected from a camera image as an object related to a monitoring event (person detection). In this embodiment, a person is detected from a camera image (frame) by using a machine learning model for detecting a person by deep learning or the like (person detection). At this time, the camera image is input to the machine learning model, and the person score (likelihood indicating the high possibility of being a person) output from the machine learning model is acquired. Then, when the person score is equal to or higher than a predetermined value (person score threshold value), it is determined to be a person.

At this time, a person detection frame surrounding the person is set on the camera image (frame).

Further, in the retention detection process, a person detected from a camera image (frame) is tracked (person tracking). At this time, paying attention to a plurality of features such as the color and movement of the person, the person detected from each frame is associated with each other. In particular, the feature amounts of the persons detected from the current frame and the previous frame are compared, and when the similarity between the feature amounts of both is equal to or higher than a predetermined value (person similarity degree threshold value), the person is determined to be the same person. To do.

Next, in the retention detection process, as a positional condition for issuing an alarm, it is determined whether or not the person being tracked is staying in an effective area other than the mask area in the detection area. At this time, it is determined whether or not the reference point of the person detection frame is located in the effective area. In the present embodiment, the center point of the person detection frame is used as a reference point. Further, the detection area and the mask area for stay detection can be designated by a rectangle. The detection area and the mask area may be designated by a polygon defined by a predetermined number or less (for example, 8 points) of vertices.

Further, in the retention detection process, as a temporal condition regarding the alarm issuance, it is determined whether or not the person being tracked continuously stays in the effective area for a predetermined time (person staying time) or more. In this determination, if the staying time of the person being tracked exceeds the staying time threshold value, an alarm is issued.

Here, if the person being tracked leaves the detection area before the person staying time elapses, specifically, if the reference point of the person detection frame moves out of the detection area, the alarm is issued. exclude. In addition, if the upper body of the person being tracked is concealed by a concealed object such as a signboard or a pillar before the person's staying time elapses, the person is excluded from the alarm.

Next, the intrusion detection process performed by the image analysis server 3 will be described. FIG. 9 is an explanatory diagram showing an outline of the intrusion detection process performed by the image analysis server 3.

In the intrusion detection process, a person is detected from a camera image as an object related to a monitoring event (person detection). In this embodiment, a person is detected from a camera image (frame) by using a machine learning model for detecting a person by deep learning or the like. Specifically, a camera image is input to the machine learning model, and a person score (likelihood indicating the high possibility of being a person) output from the machine learning model is acquired. Then, when the person score is equal to or higher than a predetermined value (person score threshold value), it is determined to be a person.

At this time, a person detection frame surrounding the person is set on the camera image (frame).

Next, in the intrusion detection process, as a positional condition for issuing an alarm, it is determined whether or not the detected person is located in an effective area other than the mask area in the detection area. At this time, it is determined whether or not the reference point of the person detection frame is located in the effective area. In the present embodiment, the center point of the person detection frame is used as a reference point. Further, the detection area and the mask area for intrusion detection can be designated by a polygon defined by a predetermined number or less (for example, 8 points) of vertices. The detection area and the mask area may be designated by a rectangle. Further, the detection area is set to an area where it is necessary to pay attention to the entry of the user, for example, an area where the entry is prohibited except for the persons concerned with the facility.

Next, in the intrusion detection process, as a time condition for issuing an alarm, when the detected person is located in the effective area, the number of frames in which the person is located in the effective area is the latest predetermined number (most recent). It is determined whether or not there are a predetermined number (number of continuous intrusion frames) or more in the frames (number of determination frames). When this time condition is satisfied, an alarm is issued.

Next, the operation status confirmation screen displayed on the maintenance terminal 7 will be described. FIG. 10 is an explanatory diagram showing an operation status confirmation screen displayed on the maintenance terminal 7.

By accessing the image analysis server 3, the maintenance terminal 7 displays an operation status confirmation screen generated by the image analysis application executed by the image analysis server 3.

The operation status confirmation screen is provided with a camera list display unit 101, a setting information display unit 102, and a camera image display unit 103.

The camera list display unit 101 displays a list of cameras 1 associated with the image analysis server 3. In the camera list display unit 101, a maintenance worker can perform an operation of selecting the camera 1.

In the setting information display unit 102, as behavior analysis 1 and behavior analysis 2, two types of monitoring events being selected (taken away, taken away) detected by image analysis (behavior analysis) with respect to the camera 1 selected by the camera list display unit 101. Fall, stay, intrusion) is displayed.

The camera image display unit 103 displays a real-time camera image taken by the camera 1 selected by the camera list display unit 101.

Next, the setting screen displayed on the maintenance terminal 7 will be described. FIG. 11 is an explanatory diagram showing a setting screen when the main body setting button 121 displayed on the maintenance terminal 7 is operated. FIG. 12 is an explanatory diagram showing a main part of the setting screen when the take-away setting button 122 and the fall setting button 123 are operated. FIG. 13 is an explanatory diagram showing a main part of the setting screen when the residence setting button 124 and the intrusion setting button 125 are operated. FIG. 14 is an explanatory diagram showing a main part of the setting screen in the area setting mode.

As shown in FIG. 11, the setting screen includes an image analysis server setting unit 111, a camera selection unit 112, an item-specific setting input unit 113, a camera image display unit 114, and a guide message display unit 115. Has been done.

The image analysis server setting unit 111 is provided with a server name display unit 126, a batch update button 127, a batch acquisition button 128, and a shared setting reflection button 129. The name of the image analysis server 3 to be set is displayed on the server name display unit 126. When the batch update button 127 is operated, a screen (not shown) for selecting a setting file in which setting information related to the image analysis server 3 is stored is displayed. When a setting file is selected here, the setting stored in the setting file is displayed. Based on the information, the setting contents of all the setting items are updated at once. By operating the batch acquisition button 128, related setting files can be saved in a batch. When the sharing setting reflection button 129 is operated, the setting information related to the target image analysis server 3 can be acquired from the setting file shared between the systems, and the setting contents of the setting information can be reflected.

In the camera selection unit 112, the user can select the camera 1 to be set.

The item-specific setting input unit 113 is provided with a main body setting button 121, a take-away setting button 122, a fall setting button 123, a retention setting button 124, and an intrusion setting button 125. The user can select the type of monitoring event to be set by operating the main body setting button 121, the take-out setting button 122, the fall setting button 123, the retention setting button 124, and the intrusion setting button 125.

The camera image display unit 114 displays a real-time camera image taken by the selected camera 1. In addition, by operating the take-away setting button 122, the fall setting button 123, the retention setting button 124, and the intrusion setting button 125, after selecting the type of monitoring event to be set, if a predetermined operation is performed, the area setting mode is set. In this area setting mode, the camera image display unit 114 can input each range of the detection area and the mask area by operating the mouse. When the detection area and the mask area are set, the frame image representing the range of the detection area and the frame image representing the range of the mask area are superimposed and drawn on the camera image.

In the area setting mode, the guide message display unit 115 displays a guide message regarding an area designation operation for designating the range of the detection area and the mask area.

Further, the setting screen is provided with an image acquisition button 116, an unregistered button 117, a setting update button 118, and an image read button 119. By operating the image acquisition button 116, the latest camera image can be acquired. By operating the unregistered button 117, the camera 1 can be unregistered. By operating the setting update button 118, it is possible to restart after saving the setting contents. By operating the image read button 119, the camera image saved in advance can be read, and even in a situation where the camera image of the selected camera 1 cannot be acquired, the camera image is displayed on the camera image display unit 114 and the area designation operation is performed. be able to.

As shown in FIG. 11, in the state where the main body setting button 121 is selected (initial state), the item-specific setting input unit 113 has the camera name display unit 131 and the detection event type selection as related to basic setting items. The unit 132 and the camera management server setting information display unit 133 are displayed.

The camera name display unit 131 displays the identification information of the camera 1, for example, the installation location of the camera 1.

The detection event type selection unit 132 can select the type of monitoring event detection (take-away, fall, stay, intrusion) to be performed on the camera image. In the present embodiment, two detection event type selection units 132 are provided, and the user selects two types of monitoring event detection targeting the camera image of one camera 1 and registers them as setting information. be able to.

The camera management server setting information display unit 133 displays the IP address, port number, and channel number of the camera management server 2 as setting items related to the camera management server 2.

In the present embodiment, two types can be selected from a maximum of four types for monitoring event detection for the camera image of one camera 1, but one type or three types can be selected. You may do so. Further, the maximum number of types that can be selected may be five or more, and a predetermined number of types may be selected from among them.

When "take-away" is selected in the detection event type selection unit 132 and the take-out setting button 122 is operated, as shown in FIG. 12 (A), the item-specific setting input unit 113 relates to the setting items related to the take-out detection. The notification necessity designation unit 141, the detection area designation unit 142, the mask area designation unit 143, and the processing condition designation unit 144 are displayed.

In the alarm issuance necessity designation unit 141, the user can specify the necessity of alarm issuance at the time of the removal detection by the radio button.

The detection area designation unit 142 is provided with an end point coordinate input unit 151 and a setting button 152. In the end point coordinate input unit 151, the user can input an end point that defines the range of the detection area. In take-away detection, the detection area can be specified as a rectangle. The xy coordinates of the two vertices (end points) in the diagonal direction in the rectangle are input to the end point coordinate input unit 151.

When the user operates the setting button 152, the mode shifts to the area setting mode. In this area setting mode, the operator can specify the detection area by operating the mouse. Specifically, the pointer can be moved on the camera image of the camera image display unit 114 by moving the mouse, and the positions of the two vertices (end points) in the diagonal direction in the rectangle are sequentially specified by the click operation. By doing so, the rectangle of the detection area is set.

Here, when the mode is changed to the area setting mode, the operation guide message related to the area setting is displayed on the guide message display unit 115 (see FIG. 14). When the rectangle of the detection area is set, a frame image representing the range of the detection area is drawn on the camera image display unit 114 by superimposing it on the camera image (see FIG. 14).

The mask area designation unit 143 is provided with an area number input unit 153, an end point coordinate input unit 154, an add button 155, a setting button 156, and a delete button 157. The area number input unit 153 allows the user to select the number of mask areas. In the end point coordinate input unit 154, the user can input an end point that defines the range of the mask area. In the take-out detection, the mask area can be specified by a rectangle. The xy coordinates of the two vertices (end points) in the diagonal direction in the rectangle are input to the end point coordinate input unit 154. The mask area can be set within the detection area.

The add button 155 adds a mask area. When the user operates the add button 155, the mode shifts to the area setting mode, and a new mask area range can be specified on the camera image of the camera image display unit 114 by operating the mouse. When the mask area is set, a frame image representing the range of the mask area is drawn on the camera image display unit 114 by superimposing it on the camera image (see FIG. 14). The setting button 156 redesignates the range of the mask area that has already been set. When the user operates the setting button 156, the mode shifts to the area setting mode, and the range of the mask area can be changed on the camera image of the camera image display unit 114 by operating the mouse. The delete button 157 deletes the mask area that has already been set. In the area setting mode, when the user selects the mask area superimposed and drawn on the camera image display unit 114, the user can operate the delete button 157, the setting information of the corresponding mask area is deleted, and the camera image is displayed. The frame image of the mask area of unit 114 is erased.

The processing condition designation unit 144 is provided with a take-away time input unit 158 and a feature amount difference threshold value input unit 159. In the take-out time input unit 158, the user can input the take-out time. The take-away time is a threshold value for the duration of the stationary state of the object in the effective area of the detection area. In the feature amount difference threshold input unit 159, the user can input the feature amount difference threshold value. The feature amount difference threshold value is a threshold value related to the feature amount difference when distinguishing a take-away object from the background.

When "fall" is selected in the detection event type selection unit 132 and the fall setting button 123 is operated, as shown in FIG. 12B, the item-specific setting input unit 113 is issued as a setting item related to the fall detection. The report necessity designation unit 141, the detection area designation section 142, the mask area designation section 143, and the processing condition designation section 144 are displayed.

In the alarm issuance necessity designation unit 141, the user can specify the necessity of alarm issuance at the time of fall detection by the radio button.

The detection area designation unit 142 is provided with an area number input unit 161, an end point coordinate input unit 162, an add button 163, a setting button 164, and a delete button 165. The area number input unit 161 allows the user to select the number of detection areas. In the end point coordinate input unit 162, the user can input an end point that defines the range of the detection area. In fall detection, the detection area can be specified as a polygon. The xy coordinates of all the vertices (end points) of the polygon are input to the end point coordinate input unit 162.

The add button 163 adds a detection area. When the user operates the add button 163, the mode shifts to the area setting mode, and a new detection area range can be specified on the camera image of the camera image display unit 114 by operating the mouse. When the detection area is set, a frame image representing the range of the detection area is superimposed and drawn on the camera image on the camera image display unit 114. The setting button 156 redesignates the range of the detection area that has already been set. When the user operates the setting button 156, the mode shifts to the area setting mode, and the range of the detection area can be changed on the camera image of the camera image display unit 114 by operating the mouse. The delete button 165 deletes the detection area that has already been set. In the area setting mode, when the user selects the detection area superimposed and drawn on the camera image display unit 114, the user can operate the delete button 165, whereby the setting information of the corresponding detection area is deleted. The frame image in the mask area of the camera image display unit 114 is erased.

The mask area designation unit 143 is provided with an area number input unit 153, an end point coordinate input unit 154, an add button 155, a setting button 156, and a delete button 157. The area number input unit 153, the end point coordinate input unit 154, the add button 155, the setting button 156, and the delete button 157 are the same as in the case of the take-away detection shown in FIG. 12 (A). However, in the case of fall detection, the detection area can be specified by a polygon, and the xy coordinates of all the vertices (end points) of the polygon are input to the end point coordinate input unit 154.

The add button 155, the setting button 156, and the delete button 157 are the same as in the case of the take-away detection shown in FIG. 12 (A).

The processing condition designation unit 144 is provided with the latest determination frame number input unit 167, the fall continuation frame number input unit 168, and the fall score threshold value input unit 169. In the latest determination frame number input unit 167, the user can input the latest determination frame number. The latest determination frame is the total number of target frames (camera images) when determining the necessity of alarm issuance. In the fall continuation frame number input unit 168, the user can input the fall continuation frame number. The number of continuous fall frames is a threshold value related to the number of frames (camera images) in which a person in a fall state is located in the effective area, and among the frames of the number of most recent determination frames, a frame in which a person in a fall state is located in the effective area. If is greater than or equal to the number of continuous fall frames, it is determined that an alarm is required. In the fall score threshold input unit 169, the user can input the fall score threshold. The fall score threshold is a threshold related to the fall score (likelihood indicating the high possibility of falling) when determining the fall state.

When "retention" is selected in the detection event type selection unit 132 and the retention setting button 124 is operated, as shown in FIG. 13A, the item-specific setting input unit 113 is issued as a setting item related to retention detection. The report necessity designation unit 141, the detection area designation section 142, the mask area designation section 143, and the processing condition designation section 144 are displayed.

In the alarm issuance necessity designation unit 141, the user can specify the necessity of alarm issuance at the time of retention detection by the radio button.

The detection area designation unit 142 is provided with an end point coordinate input unit 151 and a setting button 152. The end point coordinate input unit 151 and the setting button 152 are the same as in the case of the take-away detection shown in FIG. 12 (A).

The mask area designation unit 143 is provided with an area number input unit 153, an end point coordinate input unit 154, an add button 155, a setting button 156, and a delete button 157. The area number input unit 153, the end point coordinate input unit 154, the add button 155, the setting button 156, and the delete button 157 are the same as in the case of the take-away detection shown in FIG. 12 (A).

The processing condition designation unit 144 is provided with a person stay time input unit 171, a person score threshold value input unit 172, and a person similarity threshold value input unit 173. The person staying time input unit 171 allows the user to input the person staying time. The person staying time is a threshold value regarding the person's staying time when determining that the person stays. The person score threshold input unit 172 allows the user to input the person score threshold. The person score threshold value is a threshold value related to a person score (likelihood indicating a high possibility of being a person) when determining a person. The person similarity threshold input unit 173 allows the user to input the person similarity threshold. The person similarity threshold is a threshold related to the similarity of the feature amount of a person when determining the person detected in the previous and next frames in the person tracking as the same person.

When "intrusion" is selected in the detection event type selection unit 132 and the intrusion setting button 125 is operated, as shown in FIG. 13B, the item-specific setting input unit 113 is issued as a setting item related to intrusion detection. The report necessity designation unit 141, the detection area designation section 142, the mask area designation section 143, and the processing condition designation section 144 are displayed.

In the alarm issuance necessity designation unit 141, the user can specify the necessity of alarm issuance at the time of intrusion detection by the radio button.

The detection area designation unit 142 is provided with an end point coordinate input unit 151 and a setting button 152. The end point coordinate input unit 151 and the setting button 152 are the same as in the case of the take-away detection shown in FIG. 12 (A). However, in the case of intrusion detection, the detection area can be specified as a polygon.

The mask area designation unit 143 is provided with an area number input unit 153, an end point coordinate input unit 154, an add button 155, a setting button 156, and a delete button 157. The area number input unit 153, the end point coordinate input unit 154, the add button 155, the setting button 156, and the delete button 157 are the same as in the case of the take-away detection shown in FIG. 12 (A). However, in the case of intrusion detection, the mask area can be specified as a polygon.

The processing condition designation unit 144 is provided with the latest determination frame number input unit 167, the intrusion continuation frame number input unit 175, and the person score threshold value input unit 172. In the latest determination frame number input unit 167, the user can input the latest determination frame number. The latest determination frame is the total number of target frames (camera images) when determining the necessity of alarm issuance. In the intrusion continuation frame number input unit 175, the user can input the intrusion continuation frame number. The number of continuous intrusion frames is a threshold value related to the number of frames (camera images) in which the person is located in the effective area, and among the frames of the most recent judgment frame number, the frame in which the person is located in the effective area is equal to or greater than the number of continuous intrusion frames. If, it is determined that an alarm is required. In the person score threshold input unit 172, the user can input the person score threshold. The person score threshold value is a threshold value related to a person score (likelihood indicating a high possibility of being a person) when determining a person.

Next, the monitoring screen displayed on the monitoring terminal 6 will be described. FIG. 15 is an explanatory diagram showing a monitoring screen displayed on the monitoring terminal 6.

The monitoring screen is provided with a live image display unit 181, an analysis result image display unit 182, a playback operation unit 183, and an event image display unit 184.

The live image display unit 181 displays a real-time camera image received from the camera management server 2. In the live image display unit 181, real-time camera images for each of the plurality of cameras 1 are displayed side by side. In addition, the display frame of the live image display unit 181 is highlighted in the selected state.

The analysis result image display unit 182 displays the analysis result image received from the camera management server 2. In the analysis result image, a detection frame (frame image) surrounding an object related to a monitoring event (take-away, fall, stay, and intrusion) is superimposed on a camera image (video) for a predetermined period based on the alarm issuance time. It is drawn. Specifically, in the case of taking away, the detection frame is displayed on an object other than the disappeared person. In the case of a fall, a detection frame is displayed on the person in the fall state. In the case of staying, a detection frame is displayed for a person who stays in a predetermined area for a long time. In the case of intrusion, a detection frame is displayed for the person who invaded the predetermined area. In the camera image (moving image) for a predetermined period, the detection frame is continuously displayed even after the alarm issuance time as long as it is in the detection state. If the monitoring event is taken away, the object disappears after the start of detection, and only the detection frame is displayed. Therefore, an analysis result image in which the detection frame is displayed on the object is generated from the image before the start of detection and disappears. Make it possible to confirm what the object is.

The reproduction operation unit 183 controls the reproduction of the analysis result image displayed on the analysis result image display unit 182, and can perform operations such as reproduction, fast forward, and fast rewind.

The event image is displayed on the event image display unit 184. The event image is a camera image (still image) at the time when the alarm is issued. For example, among the analysis result images, the frame image on which the detection frame is superimposed can be used as the event image for the first time.

On this monitoring screen, the analysis result image is reproducibly displayed on the analysis result image display unit 182 substantially at the same time as the alarm issuance, specifically, the timing when the alarm light 5 is turned on. As a result, the observer can immediately and in detail the actual situation of the monitoring event by viewing the analysis result image (confirmation image) showing the monitoring event related to the alarm issuance immediately after noticing the alarm issuance. Can be grasped.

(Second Embodiment)
Next, the second embodiment will be described. It should be noted that the points not particularly mentioned here are the same as those in the above-described embodiment. FIG. 16 is a block diagram showing a schematic configuration of the camera management server 2 and the image analysis server 3 according to the second embodiment.

In the first embodiment (see FIG. 4), a monitoring event detection process, that is, an image analysis server 3 that detects the occurrence of each monitoring event of take-away, fall, stay, and intrusion by image analysis (behavior analysis), and A frame processing, that is, a camera management server 2 for generating an analysis result image by adding a detection frame for notifying the observer of the occurrence of a monitoring event to the camera image is provided, but in the present embodiment, the camera management server is provided. The frame processing function of 2 is integrated into the image analysis server 3, and the image analysis server 3 performs the frame processing.

In this case, in the image analysis server 3, the processor 33 performs frame processing in addition to setting processing, monitoring event detection processing, notification necessity determination processing, and learning update processing. Further, the communication unit 31 receives the real-time camera image from the camera management server 2 and transmits the alarm issuance information and the analysis result image to the camera management server 2.

In the present embodiment, the frame processing function of the camera management server 2 is integrated into the image analysis server 3, but the camera management server 2 and the image analysis server 3 may be integrated into one server.

As described above, embodiments have been described as examples of the techniques disclosed in this application. However, the technique in the present disclosure is not limited to this, and can be applied to embodiments in which changes, replacements, additions, omissions, etc. have been made. It is also possible to combine the components described in the above embodiments to form a new embodiment.

In the monitoring system and the monitoring system setting program according to the present invention, when an alarm is issued by detecting the occurrence of a monitoring event, the observer gives an alarm about the specific status of the monitoring event related to the alarm issuance. It has the effect of being able to immediately confirm immediately after the alarm is issued, and even if it is delayed from the alarm issuance, and detects the occurrence of a predetermined monitoring event based on the real-time camera image of the monitoring area. It is useful as a monitoring system that issues an alarm, and a monitoring system setting program that allows an information processing device to make settings related to the monitoring system.

1 Camera 2 Camera management server (image storage unit, image distribution unit, confirmation image generation unit)
3 Image analysis server (image analysis unit)
4 Monitoring server 5 Alarm light 6 Monitoring terminal (user device)
7 Maintenance terminal (user device)
112 Camera selection unit 114 Camera image display unit 132 Detection event type selection unit (monitoring event selection unit)
133 Camera management server setting information display unit 142 Detection area designation unit 143 Mask area designation unit

Claims (8)

A monitoring system that detects the occurrence of a predetermined monitoring event and issues an alarm based on a real-time camera image of the monitoring area.
At least one camera that captures the surveillance area,
An image storage unit that stores the camera images taken by this camera, and
By image analysis of the camera image, the occurrence of the monitoring event is detected, an alarm is issued to notify the user of the occurrence of the monitoring event, and the position information of the object related to the detected monitoring event is included. An image analysis unit that generates event detection information,
A confirmation image generation unit that generates a confirmation image by adding a mark image pointing to an object related to the monitoring event to the temporarily accumulated camera image based on the event detection information.
A monitoring system including an image distribution unit that distributes the confirmation image to a user device at the same timing as the alarm issuance. The image analysis unit
It is characterized in that the occurrence of a monitoring event of a type corresponding to the camera is detected based on setting information in which a type selected in advance for each camera is registered from a plurality of types that can be detected as the monitoring event. The monitoring system according to claim 1. The image analysis unit
The monitoring system according to claim 2, wherein a plurality of types smaller than the maximum number of types that can be detected as the monitoring event are selected by the user and the setting information is registered. The confirmation image generation unit
A claim characterized in that a frame image surrounding an object related to the monitoring event is added to the camera image as the mark image based on the event detection information acquired from the image analysis unit to generate the confirmation image. Item 1. The monitoring system according to item 1. The monitoring system according to claim 1, wherein the image analysis unit and the confirmation image generation unit are composed of one server device. A monitoring system setting program that allows an information processing device to make settings related to a monitoring system that detects the occurrence of a predetermined monitoring event and issues an alarm based on a real-time camera image of the monitoring area taken by a camera.
A setting screen is displayed on the user device, and setting information related to the monitoring system is acquired according to the user's screen operation on the setting screen.
The setting screen is
A camera selection unit that selects the camera to be set from the plurality of cameras provided in the surveillance system, and a camera selection unit.
A monitoring event selection unit that selects at least one of a plurality of types of detectable monitoring events for the camera selected by the camera selection unit.
The occurrence of the monitoring event is detected by the image storage unit that stores the camera image corresponding to the camera selected by the camera selection unit and the image analysis of the camera image, and the occurrence of the monitoring event is detected. A setting information display unit that displays setting information related to the image analysis unit that issues an alarm to notify the user, and a setting information display unit.
An image display unit that displays a real-time camera image taken by the camera selected by the camera selection unit, and an image display unit.
A monitoring system setting program characterized by being equipped with. The setting screen is
The monitoring system setting program according to claim 6, further comprising a detection area designation unit that specifies a range of detection areas for which an alarm is issued when an object related to a monitoring event is located. The setting screen is
The monitoring system setting program according to claim 7, further comprising a mask area designation unit that specifies a range of mask areas to be excluded from the target of alarm issuance when an object related to a monitoring event is located.

Images