Image analysis device and flood gate opening / closing system

Abstract

To solve the problems that a water level gauge and a flow direction gauge are likely to be washed away and even if they are out of order, it is difficult to notice.SOLUTION: A first image analysis device 1 in a flood gate opening / closing system comprises: a flow direction detection unit 17 that compares image data having imaging information from a first camera 11 with flow direction information learning data for indicating flow direction information set by image data based on multiple images, and obtains flow direction determination information; and a water level detection unit 15 that compares image data having the imaging information from the first camera 11 with water level information learning data for indicating water level information set by image data based on multiple images, and obtains water level determination information.SELECTED DRAWING: Figure 7

Description

[Technical field]

[0001] The present disclosure relates to an image analysis device and a water gate opening and closing system that obtain flow direction determination information and water level determination information from imaging information from a camera. [Background technology]

[0002] Patent Document 1 shows an automatic flood gate control system that determines whether or not an emergency signal is received from a water level gauge and a flow direction gauge, and automatically closes the flood gates of a water gate installed to prevent the backflow of seawater or rivers due to tsunamis or high tides. If an emergency signal is received, the system closes automatically. [Prior art documents] [Patent documents]

[0003] [Patent Document 1] JP 2019-120036 A Summary of the Invention [Problem to be solved by the invention]

[0004] The automatic flood gate control system shown in Patent Document 1 measures the water level using a water level gauge installed near the floodgate, and measures the water flow directly below the floodgate door using a flow direction gauge installed near the floodgate. An urgency determination unit receives the measurement value from the water level gauge and the occurrence of backflow from the flow direction gauge and determines whether or not it corresponds to an emergency signal to close the floodgate. The automatic flood gate control system disclosed in Patent Document 1 had the problem that the water level gauge and flow direction gauge were at risk of being washed away, and that it was difficult to notice if the water level gauge and flow direction gauge were malfunctioning. In addition, there was also the issue that only limited, partial information could be obtained from the installation of water level gauges and current direction gauges, resulting in cases where accurate information was lacking.

[0005] The present disclosure has been made in consideration of the above-mentioned points, and aims to provide an image analysis device that can accurately obtain flow direction information and water level information for closing floodgates to prevent backflow or overflow of rivers caused by floods or tsunamis. [Means for solving the problem]

[0006] The video analysis device of the present disclosure includes a camera, a water level detection unit that compares video data based on imaging information from the camera with water level information learning data for indicating water level information set by video data based on multiple images, and obtains water level determination information, and a flow direction detection unit that compares video data based on imaging information from the camera with flow direction information learning data for indicating flow direction information set by video data based on multiple images, and obtains flow direction determination information. Effect of the Invention

[0007] According to the present disclosure, flow direction determination information and water level determination information can be obtained with high accuracy from imaging information from a camera, and as a result, the floodgates can be automatically closed efficiently and accurately. [Brief description of the drawings]

[0008] [Figure 1] 1 is a block diagram showing a configuration example of a watergate opening and closing system according to a first embodiment. [Diagram 2] 1 is a schematic diagram showing the arrangement of a first image analysis device and a second image analysis device in a floodgate opening and closing system according to embodiment 1. FIG. [Diagram 3] FIG. 2 is a diagram showing a schematic view of an image based on flow direction and flow speed determination image information in the floodgate opening and closing system according to the first embodiment. [Figure 4] FIG. 2 is a diagram showing a schematic view of an image based on water level judgment image information in the watergate opening and closing system according to the first embodiment. [Diagram 5] FIG. 2 is a diagram showing a schematic view of an image based on obstacle determination image information in the floodgate opening and closing system according to the first embodiment. [Figure 6]FIG. 2 is a diagram showing a schematic diagram of an image based on human detection image information in the floodgate opening and closing system according to the first embodiment. [Figure 7] FIG. 2 is a block diagram mainly showing a first video analysis device and a second video analysis device in the floodgate opening and closing system according to the first embodiment. [Figure 8] FIG. 1 is a block diagram mainly showing a water gate control system in a water gate opening and closing system according to a first embodiment. [Figure 9] 6 is a flowchart showing the operation of a flow direction detection unit of the first image analysis device in the floodgate opening and closing system according to the first embodiment. [Figure 10] 6 is a flowchart showing the operation of a water level detection section of the first image analysis device in the floodgate opening and closing system according to the first embodiment. [Figure 11] 11 is a flowchart showing the operation of a flow velocity detection unit and an obstacle detection unit of the first image analysis device in the floodgate opening and closing system according to the first embodiment. [Figure 12] 13 is a flowchart showing the operation of a human detection section of a second video analysis device in the floodgate opening and closing system according to the first embodiment. [Figure 13] 5 is a flowchart showing the operation of a signal determination unit of the watergate control system in the watergate opening and closing system according to the first embodiment. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0009] Embodiment 1 The water gate opening and closing system according to the first embodiment will be described with reference to Figs. First, an example of the overall configuration of a water gate opening / closing system according to the first embodiment will be described with reference to FIG. The floodgate opening and closing system of embodiment 1 comprises a first video analysis device 1, a second video analysis device 2, a floodgate control system 3, a video recording device 4, an alarm device 5, a data distribution switch (L3 switch) 6, routers 7 and 8, and a management operation device 9.

[0010] The first image analysis device 1 is a first AI camera for closing a water gate to prevent backflow or overflow of a river caused by a flood or a tsunami. Hereinafter, the first image analysis device 1 will be described as a first AI camera 1. The first AI camera 1 is installed near the floodgate 100, as shown in FIG. The first camera 11 in the first AI camera 1 is a network camera for capturing images of the river and its surroundings in real time and monitoring the water gate. In the first embodiment, the first camera 11 captures an image of the river surface downstream from the water gate 100, including the water gate 100, and the bank of the river.

[0011] In FIG. 2, the first AI camera 1 is installed near the floodgate 100, but the first camera 11 may be installed near the floodgate 100, and components other than the first camera 11 may be arranged separately. Furthermore, first camera 11 is not limited to one network camera, but a plurality of network cameras may be arranged so that their angles of view do not overlap, and images captured by the plurality of network cameras may be synthesized to produce imaging information. In the first embodiment, the downstream side of the water gate 100 in FIG. 2 is the sea.

[0012] The first AI camera 1 obtains water level determination information, flow direction determination information, and flow velocity determination information from the imaging information from the first camera 11 periodically according to a set analysis cycle, and upon request from the management operation device 9. In addition, the first AI camera 1 generates flow direction and flow speed determination result image information and water level determination result image information based on the imaging information from the first camera 11.

[0013] The image based on the flow direction and flow speed determination result image information is an image in which the flow direction and flow speed are added by arrows based on the flow direction information and flow speed information to a schematic base map, as shown in FIG. The direction of the arrows in FIG. 3 indicates the flow direction, and the length of the arrow indicates the flow velocity. As shown in FIG. 4, the image based on the water level determination result image information is an image in which a line representing the water level is added to the schematic base diagram relative to the reference overflow line, i.e., the overflow water level in FIG. 4.

[0014] The first AI camera 1 further obtains obstacle determination information for automatic closing of the floodgate based on the imaging information from the first camera 11. In addition, the first AI camera 1 generates obstacle determination result image information based on the imaging information from the first camera 11. The image based on the obstacle determination result image information is an image in which a diagrammatic representation of an obstacle is added to a schematic base diagram, as shown in FIG.

[0015] As described above, the first AI camera 1 receives imaging information from the first camera 11 and operates a flow direction information acquisition program to obtain flow direction information, a water level information acquisition program to obtain water level information, a flow velocity information acquisition program to obtain flow velocity information, and an obstacle information acquisition program to obtain obstacle information to obtain information for monitoring the floodgate.

[0016] The second video analysis device 2 is a second AI camera that checks for the presence or absence of humans around the floodgate when the floodgate is opened or closed. Hereinafter, the second video analysis device 2 will be described as the second AI camera 2. The second AI camera 2 is installed near the floodgate 100, as shown in FIG. The second camera 21 in the second AI camera 2 is a network camera for capturing images of the surroundings of the river in real time and monitoring the presence or absence of people within a set range near the floodgate 100.

[0017] In addition, in FIG. 2, the second AI camera 2 is installed near the floodgate 100, but the second camera 21 may be installed near the floodgate 100, and components other than the second camera 21 may be arranged separately. Further, the second camera 21 is not limited to one network camera, but a plurality of network cameras may be arranged so that their angles of view do not overlap, and images captured by the plurality of network cameras may be synthesized to generate imaging information.

[0018] The second AI camera 2 obtains human detection determination information around the water gate from the imaging information from the second camera 21 periodically according to a set analysis cycle and upon request from the management operation device 9. In addition, the second AI camera 2 generates image information of the person detection determination result around the water gate based on the imaging information from the second camera 21.

[0019] An image based on the human detection determination result image information is an image in which a diagrammatic representation of a person is added to a schematic base image, as shown in FIG. The second AI camera 2 receives imaging information from the second camera 21, and operates a human detection information acquisition program for obtaining human detection information to obtain information for detecting the presence of people in the vicinity of the water gate.

[0020] The flow direction determination information, water level determination information, obstacle determination information, and flow speed determination information, which are the analysis and judgment results based on the imaging information from the first camera 11 in the first AI camera 1, are transmitted as data to the floodgate control system 3 and the video recording device 4 via the data distribution switch 6. The flow direction and flow speed determination result video information, water level determination result video information, and obstacle determination result video information generated from the imaging information from the first camera 11 in the first AI camera 1 are transmitted as data to the video recording device 4 via the data distribution switch 6.

[0021] The human detection determination information, which is the analysis result based on the imaging information from the second camera 21 in the second AI camera 2, is transmitted as data to the floodgate control system 3 and the video recording device 4 via the data distribution switch 6. The human detection determination result video information generated from the imaging information from the second camera 21 in the second AI camera 2 is transmitted to the video recorder 4 via the data distribution switch 6 . The data distribution switch 6 is a commonly known data distribution switch connected to a network, such as an L3 switch.

[0022] The floodgate control system 3 analyzes and judges the flow direction determination information and water level determination information received from the first AI camera 1 via the data distribution switch 6, and controls the opening and closing operations of the floodgates. The floodgate control system 3 analyzes and judges the obstacle determination information and flow velocity determination information received from the first AI camera 1 via the data distribution switch 6, and controls the opening and closing operations of the floodgates.

[0023] The video recording device 4 records flow direction determination information, water level determination information, obstacle determination information, and flow speed determination information from the first AI camera 1 received via the data distribution switch 6, and also records person detection determination information from the second AI camera 2 received via the data distribution switch 6. The video recording device 4 records the video information of the flow direction and flow speed determination results, the video information of the water level determination result, and the video information of the obstacle determination result from the first AI camera 1 received via the data distribution switch 6, and also records the video information of the person detection determination result from the second AI camera 2 received via the data distribution switch 6.

[0024] In addition, the video recording device 4 may record imaging information from the first camera 11 linked to flow direction determination information, water level determination information, flow speed determination information, and obstacle determination information from the first AI camera 1 periodically and in emergencies, and may record imaging information from the second camera 21 linked to person detection determination information from the second AI camera 2.

[0025] The alarm device 5 receives the warning instruction information transmitted based on the results of analysis and judgment by the floodgate control system 3 via the data distribution switch 6, and generates a warning sound to warn people around the floodgate. The warning sound is a sound prepared in advance by the user. The alarm device 5 also has a function of issuing voice instructions from the administrator remotely via the administrative operation device 9.

[0026] Routers 7 and 8 are used to remotely change the equipment settings of the first AI camera 1, the second AI camera 2, the floodgate control system 3, the video recording device 4, and the alarm 5 from the management operation device 9, and to confirm the analysis and judgment results. The administration operation device 9 is an administrator's PC (Personal Computer). The administration operation device 9 is hereinafter referred to as the administrator's PC 9. The administrator PC 9 can obtain imaging information from the second camera 21 of the second AI camera 2 in real time via the routers 7 and 8 and the data distribution switch 6 in response to user operation.

[0027] The administrator PC 9 can be operated by the user to access the first AI camera 1 and the second AI camera 2 via routers 7, 8 and the data distribution switch 6, and allow the user to select which of the following programs to run: a flow direction information acquisition program for obtaining flow direction information from the first AI camera 1, a water level information acquisition program for obtaining water level information, a flow velocity information acquisition program for obtaining flow velocity information, an obstacle information acquisition program for obtaining obstacle information, and a person detection information acquisition program for obtaining person detection information from the second AI camera 2.

[0028] By operating the administrator PC 9, the user can access the first AI camera 1 and the second AI camera 2 via routers 7, 8 and the data distribution switch 6, and check the water level information, flow direction information, flow speed information, and obstacle information from the first AI camera 1, as well as the person detection information from the second AI camera 2. By operating the administrator PC 9, the user can access the first AI camera 1 and the second AI camera 2 via routers 7, 8 and the data distribution switch 6, and read out the video information of the water level determination result, video information of the flow direction and flow speed determination result, and video information of the obstacle determination result by the first AI camera 1, as well as the video information of the person detection determination result by the second AI camera 2, and confirm them as the images shown in Figures 3 to 6.

[0029] That is, the administrator's PC 9 has an operation unit that outputs operation instruction signals to selectively operate the water level detection unit 15, flow direction detection unit 14, obstacle detection unit 17, and flow speed detection unit 16 in the first AI camera 1 described below, and the person detection unit 22 in the second AI camera 2. In addition, the administrator's PC 9 has a display unit to which water level determination result video information generated by the water level detection unit 15 is input and an image based on the input water level determination result video information is displayed, to which flow direction and flow speed determination result video information generated from the flow direction information by the flow direction detection unit 14 and the flow speed information by the flow speed detection unit 16 is input and an image based on the input flow direction and flow speed determination result video information is displayed, and to which person detection determination result video information generated by the person detection unit 22 is input and an image based on the input person detection determination result video information is displayed.

[0030] The administrator PC 9 can access the water gate control system 3 via the routers 7, 8 and the data distribution switch 6 through user operation, thereby allowing the water gate control system 3 to control the opening and closing operations of the water gates. By operating the user, the administrator PC 9 can access the video recording device 4 via routers 7, 8 and the data distribution switch 6, and check the past flow direction determination information, water level determination information, flow speed determination information, and obstacle determination information from the first AI camera 1, as well as the person detection determination information from the second AI camera 2, which have been recorded in the video recording device 4.

[0031] By operating the user, the administrator PC 9 can access the video recording device 4 via routers 7, 8 and the data distribution switch 6, and read out past video information recorded in the video recording device 4 of the flow direction and flow speed determination results, water level determination results, and obstacle determination results from the first AI camera 1, as well as video information of the person detection determination results from the second AI camera 2, and review them as the images shown in Figures 3 to 6.

[0032] By operating the user, the administrator PC 9 can obtain image information from the second camera 21 via the routers 7, 8 and the data distribution switch 6, and by accessing the alarm 5, can directly issue a warning to the person identified from the image information from the second camera 21 via the alarm 5. That is, the operation section of the administrator PC 9 outputs to the alarm unit 5 alarm generation information to be issued toward a person in the image based on the person detection determination result image information displayed on the display section of the administrator PC 9.

[0033] The first AI camera 1 makes judgments using analysis technology based on artificial intelligence (AI) using a learning model, and is capable of switching between trained analysis S / W (program) using publicly known technology and detection S / W (program) (untrained analysis S / W), and it is also possible to use both S / W simultaneously. The learning model is created using a so-called known method by previously learning video data based on multiple images for analyzing and determining flow direction information, multiple images for analyzing and determining water level information, multiple images for analyzing and determining flow velocity information, and multiple images for analyzing and determining obstacle information as learning data, and the learning model is used to determine flow direction and water level, as well as flow velocity and obstacles.

[0034] Determination based on flow direction information involves setting video data based on multiple images as learning data in advance to indicate the flow direction, i.e., information on forward flow and reverse flow, and learning this as flow direction information learning data. Video data from the image information from the first camera 11 is compared with the flow direction information learning data, and an artificial intelligence detection technique is used to detect the video data in the flow direction information learning data for the video data from the image information from the first camera 11, and a determination is made as to the flow direction, indicating forward flow, reverse flow, or stagnation. The information on forward flow and reverse flow is angle information that indicates the flow angle of water in a river.

[0035] Judgment based on water level information is made by setting video data based on multiple images to show water level information in advance as learning data, learning this as water level information learning data, and comparing the video data from the imaging information from the first camera 11 with the water level information learning data, and using artificial intelligence detection technology to detect the video data in the water level information learning data relative to the video data from the imaging information from the first camera 11, and judging whether the water level indicates normal or abnormal. The water level information indicates the water area defined by the overflow line.

[0036] Judgment based on flow velocity information involves setting video data based on multiple images to indicate the speed of the river flow as learning data in advance, learning this as flow velocity information learning data, and comparing the video data from the imaging information from the first camera 11 with the flow velocity information learning data, and using artificial intelligence detection technology to detect the video data in the flow velocity information learning data for the video data from the imaging information from the first camera 11, and judging whether the flow velocity indicates whether the river is fast, slow, or stagnant.

[0037] To determine whether or not an obstacle is present, video data based on multiple images showing an obstacle is set in advance as learning data, which is then learned as obstacle information learning data. Video data based on image information from the first camera 11 is then compared with the obstacle information learning data, and an object detection technique using artificial intelligence is used to detect obstacles based on the video data in the obstacle information learning data in the video data based on image information from the first camera 11, thereby determining whether or not an obstacle is present.

[0038] As shown in Figure 7, the first AI camera 1 includes a first camera 11, a video receiving unit 12, a switch 13, a flow direction detection unit 14, a water level detection unit 15, a flow speed detection unit 16, an obstacle detection unit 17, a signal input unit 18, and a signal output unit 19. The flow direction detection unit 14, the water level detection unit 15, the flow velocity detection unit 16 and the obstacle detection unit 17 are realized by a hardware configuration using a computer. The signal input unit 18 is an input interface unit in the computer, and the signal output unit 19 is an output interface unit in the computer.

[0039] The image receiving unit 12 receives image data based on imaging information from the first camera 11 that captures an image of the river surface downstream from the water gate 100, including the water gate 100, and the bank of the river. The switch 13 selectively outputs the video data from the video receiving unit 12 to the flow direction detecting unit 14 , the water level detecting unit 15 , the flow speed detecting unit 16 and the obstacle detecting unit 17 via the signal input unit 18 . That is, the switch 13 outputs to a computer (artificial intelligence) having the functions of a flow direction detection unit 14, a water level detection unit 15, a flow speed detection unit 16, and an obstacle detection unit 17.

[0040] The flow direction detection unit 14 compares the angle information in the video data from the imaging information from the first camera 11 with pre-set angle information indicating forward flow, reverse flow, and stagnation, and obtains flow direction determination information for forward flow, reverse flow, and stagnation. The flow direction detection unit 14 compares the video data from the imaging information from the first camera 11 with flow direction information learning data for indicating flow direction information set by video data based on multiple images, and obtains flow direction determination information indicating forward flow, reverse flow, and stagnation.

[0041] The flow direction detection unit 14 obtains flow direction determination information at regular intervals, and the obtained flow direction determination information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted to the water gate control system 3 and video recording device 4 via the data distribution switch 6. The flow direction detection unit 14 obtains flow direction determination information through the user's operation of the administrator PC 9, and the obtained flow direction determination information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted to the administrator PC 9 via the data distribution switch 6 and routers 7 and 8.

[0042] Each of the multiple image data in the flow direction information learning data is data in which information indicating a flow angle and angle information are linked to image information for determining the flow angle relative to the river flow or angle information indicating forward flow, reverse flow, and stagnation. The flow direction information learning data is stored in a storage unit in the first AI camera 1, that is, in a memory. Each image data in the flow direction information learning data may be linked to either information indicating a flow angle or angle information.

[0043] The information indicating the flow angle is counterclockwise, with the upstream direction being 0°. As an example, the angle information may be first angle information indicating 330° to 30°, second angle information indicating 150° to 210°, and third angle information indicating 30° to 150° and 210° to 330° other than the first angle information and the second angle information. When the first angle information is displayed, it indicates a reverse flow, when the second angle information is displayed, it indicates a forward flow, and when the third angle information is displayed, it indicates a stagnation.

[0044] The flow direction detection unit 14 includes a flow direction analysis unit 14a and a flow direction determination unit 14b. The flow direction analysis unit 14a compares the video data based on the imaging information from the first camera 11 with each video data in the flow direction information learning data, and obtains from the flow direction information learning data video data having the same or closest video information to the video data based on the imaging information from the first camera 11. The flow direction determination unit 14b obtains information indicating the angle information or flow angle of the video data based on the imaging information from the first camera 11 from the video data in the flow direction information learning data obtained by the flow direction analysis unit 14a, and obtains flow direction determination information.

[0045] The flow direction determination unit 14b determines that there is a reverse flow when the angle information linked to the video data in the flow direction information learning data obtained by the flow direction analysis unit 14a is the first angle information or the information indicating the flow angle is within the range of 330° to 30°, determines that there is a forward flow when the angle information is the second angle information or the information indicating the flow angle is within the range of 150° to 210°, and determines that there is a stagnation when the angle information is the third angle information or the information indicating the flow angle is within the ranges of 30° to 150° and 210° to 330°.

[0046] The flow direction detection unit 14 executes the analysis and judgment process multiple times within a set period, and when the flow direction judgment unit 14b makes multiple judgments and the judgment result shows that the majority of the judgments are for backflow, it obtains flow direction judgment information indicating backflow, when the majority of the judgments are for forward flow, it obtains flow direction judgment information indicating forward flow, and when the majority of the judgments are for stagnation, it obtains flow direction judgment information indicating stagnation. By performing the analysis and determination process multiple times, it is possible to obtain highly accurate flow direction determination information. It is also possible to obtain flow direction determination information by performing the analysis and determination process once.

[0047] The flow direction detection unit 14 generates flow direction determination result image information based on flow direction determination information obtained by analyzing the image capture information from the first camera 11 and performing a determination process. The flow direction determination result image information is information indicating the direction of an arrow indicating the direction of the flow based on the flow direction determination information to be added to the schematic base map.

[0048] The water level detection unit 15 compares the video data from the imaging information from the first camera 11 with water level information learning data for indicating water level information set from video data based on multiple images, and obtains water level determination information indicating abnormal water areas and normal water areas. The water level detection unit 15 obtains water level determination information at regular intervals, and the obtained water level determination information is output to the data distribution switch 6 via the signal output unit 19. The data is then transmitted via the data distribution switch 6 to the water gate control system 3, video recording device 4, and alarm device 5. The water level detection unit 15 obtains water level determination information through the user's operation of the administrator PC 9, and the obtained water level determination information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted to the administrator PC 9 via the data distribution switch 6 and routers 7 and 8.

[0049] Each of the multiple image data in the water level information learning data is data in which information indicating the water level and information indicating an abnormal water area or a normal water area are linked to image information for determining a water area. The water level information learning data is stored in a storage unit in the first AI camera 1, that is, in a memory. Each image data in the water level information learning data may be linked to either information indicating the water level or information indicating an abnormal water area or a normal water area.

[0050] The information indicating the water level is, for example, the water level expressed by ± with respect to a preset overflow line. The information indicating an abnormal water area or a normal water area is as follows: if the water level is above a preset overflow line, it is considered abnormal; if the water level is below the overflow line, it is considered normal. The water level detection unit 15 includes a water level analysis unit 15a and a water level determination unit 15b.

[0051] The water level analysis unit 15a compares the video data from the imaging information from the first camera 11 with each video data in the water level information learning data, and obtains from the water level information learning data the video data having the same or closest video information to the video data from the imaging information from the first camera 11. The water level determination unit 15b obtains information indicating the water level of the image data based on the imaging information from the first camera 11 or information indicating an abnormal water area or a normal water area from the image data in the water level information learning data obtained by the water level analysis unit 15a, and obtains water level determination information.

[0052] The water level determination unit 15b determines that the information indicating the water level linked to the video data in the water level information learning data obtained by the water level analysis unit 15a is abnormal if it is above the overflow line, or if the information indicating an abnormal water area or a steady water area is abnormal, and determines that the information indicating the water level is steady if it is below the overflow line, or if the information indicating an abnormal water area or a steady water area is steady.

[0053] The water level detection unit 15 executes the process multiple times within a set period, and when the water level determination unit 15b makes multiple judgments and determines that there are more abnormalities than normal, it obtains water level determination information indicating an abnormal water area, and when there are more normal areas than abnormal, it obtains water level determination information indicating a normal water area. By performing the determination multiple times, highly accurate water level determination information can be obtained. Note that the water level determination information may be obtained by performing the determination once.

[0054] The water level detection unit 15 generates water level judgment result image information based on water level judgment information obtained by analyzing the image pickup information from the first camera 11 and performing judgment processing. The water level determination result image information is information in which a water level line indicating the water level based on the water level determination information is added to the schematic base diagram. The generated water level determination result video information is output to the data distribution switch 6 via the signal output unit 19 , and is then transmitted as data to the video recording device 4 via the data distribution switch 6 .

[0055] The flow velocity detection unit 16 compares the video data from the imaging information from the first camera 11 with flow velocity information learning data for indicating flow velocity information set by the video data based on multiple images, and obtains flow velocity determination information indicating whether the river flow is fast, slow, or stagnant. The flow velocity detection unit 16 obtains flow velocity determination information at regular intervals, and the obtained flow velocity determination information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted via the data distribution switch 6 to the floodgate control system 3, video recording device 4, and alarm device 5. The flow velocity detection unit 16 obtains flow velocity determination information through the user's operation of the administrator PC 9, and the obtained flow velocity determination information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted to the administrator PC 9 via the data distribution switch 6 and routers 7 and 8.

[0056] Each of the multiple image data in the flow velocity information learning data is data in which information indicating the flow velocity and information indicating whether the flow of the river is fast, slow, or stagnant are linked to image information for determining the flow velocity. The flow velocity information learning data is stored in a storage section in the first AI camera 1, that is, in a memory. Each image data in the flow velocity information learning data may be linked to either information indicating the flow velocity or information indicating whether the flow of the river is fast, slow, or stagnant.

[0057] Information indicating whether a river is fast, slow, or stagnant may, for example, be defined as fast if the upstream flow speed is 3 m / s or more, slow if it is between 1 m / s and 3 m / s, and stagnant if it is less than 1 m / s. The flow velocity detection unit 16 includes a flow velocity analysis unit 16a and a flow velocity determination unit 16b.

[0058] The flow velocity analysis unit 16a compares the video data based on the imaging information from the first camera 11 with each video data in the flow velocity information learning data, and obtains from the flow velocity information learning data video data having the same or closest video information to the video data based on the imaging information from the first camera 11. The flow velocity determination unit 16b obtains information indicating the flow velocity of the video data from the imaging information from the first camera 11 or information indicating whether the river flow is fast, slow, or stagnant from the video data in the flow velocity information learning data obtained by the flow velocity analysis unit 16a, and obtains flow velocity determination information.

[0059] Each of the video data based on the imaging information from the first camera 11 is data based on imaging information from a plurality of still images taken within a set time. Similarly, each of the video data in the flow direction information learning data is also data based on imaging information from a plurality of still images taken within a set time. In other words, the flow velocity of a river can be derived from at least two still images taken within a set period of time. The video data based on the imaging information from the first camera 11 and each video data in the flow direction information learning data may be a moving image captured at a set time.

[0060] The flow velocity determination unit 16b determines that the river flow is fast if the information indicating fast, slow, or stagnant linked to the video data in the flow velocity information learning data obtained by the flow velocity analysis unit 16a is fast, or the information indicating the flow velocity is 3 m / s or more, determines that the river flow is slow if the information indicating fast, slow, or stagnant is slow, or the information indicating the flow velocity is in the range of 1 m / s or more and less than 3 m / s, and determines that the river flow is stagnant if the information indicating fast, slow, or stagnant is stagnant, or the information indicating the flow velocity is less than 1 m / s.

[0061] The flow velocity detection unit 16 executes the process multiple times within a set period, and when the flow velocity determination unit 16b makes multiple judgments and the majority of the judgment results indicate that the flow is fast, it obtains flow velocity determination information indicating that the flow is fast; when the majority of the judgment results indicate that the flow is slow, it obtains flow velocity determination information indicating that the flow is slow; and when the majority of the judgment results indicate that the flow is stagnant, it obtains flow velocity determination information indicating that the flow is stagnant. By performing the determination multiple times, it is possible to obtain highly accurate flow velocity determination information. Note that it is also possible to obtain flow velocity determination information by performing the determination once.

[0062] The flow velocity detection unit 16 generates flow velocity judgment result image information based on flow velocity judgment information obtained by analyzing and performing judgment processing on the image information from the first camera 11. The flow velocity determination result image information is information indicating the length of an arrow indicating the flow velocity based on the flow velocity determination information to be added to the schematic base image. Based on the flow direction determination result video information indicating the direction of the arrow and the flow speed determination result video information indicating the length of the arrow, flow direction and flow speed determination result video information is generated in which arrows indicating the flow direction and flow speed are added to the schematic base diagram, and the generated flow direction and flow speed determination result video information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted to the video recording device 4 via the data distribution switch 6.

[0063] The obstacle detection unit 17 compares the video data based on the imaging information from the first camera 11 with obstacle information learning data for indicating obstacles set by video data based on multiple images, and obtains obstacle determination information indicating the presence or absence of an obstacle. The obstacle detection unit 17 obtains obstacle determination information at regular intervals, and the obtained obstacle determination information is output to the data distribution switch 6 via the signal output unit 19. The data is then transmitted via the data distribution switch 6 to the floodgate control system 3, video recording device 4, and alarm 5. The obstacle detection unit 17 obtains obstacle determination information through the user's operation of the administrator PC 9, and the obtained obstacle determination information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted to the administrator PC 9 via the data distribution switch 6 and routers 7 and 8.

[0064] Each of the plurality of image data in the obstacle information learning data is image information data showing an obstacle for use in determining the obstacle. The obstacle information learning data is stored in a storage section in the first AI camera 1, that is, in a memory. The obstacle detection unit 17 includes an obstacle analysis unit 17a and an obstacle determination unit 17b.

[0065] The obstacle analysis unit 17a compares the video data based on the imaging information from the first camera 11 with each piece of video data in the obstacle information learning data, and detects an obstacle from the video data based on the imaging information from the first camera 11 using the video data showing an obstacle in the obstacle information learning data. The obstacle determination unit 17b counts the number of obstacles detected in the video data based on the image pickup information from the first camera 11 obtained by the obstacle analysis unit 17a, and if the number of obstacles is one or more, it determines that an obstacle is present and obtains obstacle determination information indicating the presence of an obstacle, and if the number of obstacles is zero, it determines that no obstacle is present and obtains obstacle determination information indicating the absence of an obstacle.

[0066] The obstacle detection unit 17 receives the flow velocity determination information obtained by the flow velocity detection unit 16, and if the flow velocity determination information indicates that the flow velocity is fast, the obstacle detection unit 17 reduces the fixed interval for obtaining the obstacle determination information to 1 / 2 the interval, that is, the number of seconds in the analysis period to 1 / 2. When the river flows fast, the obstacle detection section 17 obtains obstacle determination information at a shorter period of time compared to when the river flows slowly or is stagnant. When the river flow is fast, the floodgate 100 is often closed, and obstacles are detected at short intervals to prevent the floodgate 100 from interfering with its closing operation, such as an obstacle coming into contact with the floodgate 100 during the closing operation.

[0067] The obstacle detection unit 17 generates obstacle determination result image information based on obstacle determination information obtained by analyzing the image information from the first camera 11 and performing a determination process. The obstacle determination result image information is information in which an obstacle based on the obstacle determination information is added to a schematic base image. The generated obstacle determination result image information is outputted to the data distribution switch 6 via the signal output unit 19 , and is then transmitted as data to the image recording device 4 via the data distribution switch 6 .

[0068] The second AI camera 2 makes judgments using artificial intelligence (AI) analysis technology that uses a learning model, and is capable of switching between trained analysis S / W (program) using publicly known technology and detection S / W (program) (untrained analysis S / W), and it is also possible to use both S / W simultaneously. The learning model is created using known techniques by previously learning video data based on multiple images for analyzing (detecting) and determining the presence of a person, and the learning model is then used to determine the presence of a person.

[0069] The presence of a person is determined by setting video data based on multiple images showing a person as learning data in advance to indicate the presence or absence of a person, learning this as person information learning data, and comparing the video data from the imaging information from the second camera 21 with the person information learning data, and using object detection technology using artificial intelligence to detect a person based on the video data in the person information learning data in the video data from the imaging information from the second camera 21, and determining whether or not a person is present.

[0070] As shown in FIG. 7, the second AI camera 2 includes a second camera 21, a video receiving unit 12, a switch 13, a human detection unit 22, a signal input unit 18, and a signal output unit 19. 7, the video receiving unit 12, the switch 13, the signal input unit 18, and the signal output unit 19 are configured to be shared with the first AI camera 1, but they may be configured separately. In this case, the switch 13 is not necessary. The human detection unit 22 is realized by a hardware configuration of a computer.

[0071] The human detection unit 22 receives image data based on imaging information from the second camera 21 that captures an image of a set range around the river near the water gate 100 via the image receiving unit 12 and the switch 13 . The human detection unit 22 compares the video data from the imaging information from the second camera 21 with human information learning data for indicating a person set by video data based on multiple images, and obtains human detection determination information indicating whether a person is present or not.

[0072] The human detection unit 22 obtains human detection judgment information at regular intervals, and the obtained human detection judgment information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted to the floodgate control system 3 and the video recording device 4 via the data distribution switch 6. The human detection unit 22 obtains human detection judgment information through the user's operation of the administrator PC 9, and the obtained human detection judgment information is output to the data distribution switch 6 via the signal output unit 19, and the data is transmitted to the administrator PC 9 via the data distribution switch 6 and routers 7 and 8.

[0073] Each of the multiple video data in the person information learning data is video information data showing a person for determining the presence of the person. The person information learning data is stored in a storage unit in the second AI camera 2, that is, in a memory. The human detection unit 22 includes a human analysis unit 22a and a human determination unit 22b.

[0074] The person analysis unit 22a compares the video data based on the imaging information from the second camera 21 with each video data in the person information learning data, and detects people from the video data based on the imaging information from the second camera 21 using the video data in the person information learning data. The person determination unit 22b counts the number of people detected in the video data based on the imaging information from the second camera 21 obtained by the person analysis unit 22a, and if there is one or more people, it determines that a person is present and obtains person detection determination information indicating the presence of a person, and if there is zero people, it determines that no people are present and obtains person detection determination information indicating the absence of a person.

[0075] The person analysis unit 22a generates person detection result video information based on person detection determination information obtained by analyzing the image pickup information from the second camera 21 and performing a determination process. The person detection result image information is information in which a person based on the person detection determination information is added to a schematic base image. The generated human detection result video information is output to the data distribution switch 6 via the signal output unit 19 , and is then transmitted to the video recording device 4 via the data distribution switch 6 .

[0076] Although the second AI camera 2 is configured separately from the first AI camera 1, the flow direction detection unit 14, water level detection unit 15, flow speed detection unit 16, and obstacle detection unit 17 in the first AI camera 1 and the person detection unit 22 in the second AI camera 2 can be realized by a hardware configuration using the same computer, and the first camera 11 and the second camera 21 can each be a network camera, and the first camera 11 and the second camera 21 can be an AI camera (video analysis device) connected to the same computer (artificial intelligence).

[0077] The floodgate control system 3 closes the floodgate 100 when the water level determination information received from the first AI camera 1 via the data distribution switch 6 indicates a steady water area and the flow direction determination information indicates a reverse flow. The floodgate control system 3 closes the floodgate 100 on the condition that the obstacle determination information from the first AI camera 1 indicates that there is no obstacle.

[0078] When the floodgate control system 3 closes the floodgate 100, if the human detection determination information from the second AI camera 2 indicates the presence of a person, it outputs alarm instruction information to the alarm 5 via the data distribution switch 6. The closing operation of the floodgate 100 refers to the period from when the floodgate 100 starts the closing operation to when it ends.

[0079] As shown in FIG. 8, the floodgate control system 3 has a signal receiving unit 31, a signal determining unit 32, an opening / closing signal output unit 33, an automatic control unit 34, a floodgate power unit 35, an abnormality detection unit 36, and an alarm signal output unit 37. The signal determination unit 32 is realized by a hardware configuration using a computer. The signal receiving unit 31 is an input interface unit in the computer, and the open / close signal output unit 33 and the alarm signal output unit 37 are output interfaces in the computer.

[0080] The signal receiving unit 31 receives flow direction determination information, water level determination information, flow speed determination information, and obstacle determination information from the first AI camera 1, and human detection determination information from the second AI camera 2 via the data distribution switch 6, and provides them to the signal determination unit 32. When the floodgate 100 is open, the signal determination unit 32 outputs a closing operation signal to the automatic control unit 34 via the opening / closing signal output unit 33 when the obstacle determination information indicates no obstacle, the water level determination information indicates a steady water area, and the flow direction determination information indicates a backflow.

[0081] When the floodgate 100 is open, the signal determination unit 32 outputs a maintain signal to maintain the open state to the automatic control unit 34 via the opening / closing signal output unit 33 if the obstacle determination information indicates the presence of an obstacle, the water level determination information indicates an abnormal water area, or the flow direction determination information indicates either stagnation or forward flow. The maintaining signal for maintaining the open state means that the closing operation signal is not output, and includes not outputting the signal. When the obstacle determination information indicates that there is no obstacle, the water level determination information indicates a steady water area, the flow direction determination information indicates a reverse flow, and the person detection determination information indicates the presence of a person, the signal determination unit 32 outputs an alarm instruction signal via the alarm signal output unit 37 and through the data distribution switch 6 to the alarm device 5.

[0082] As described above, the signal determination unit 32 is a function that receives flow direction determination information, water level determination information, flow speed determination information, and obstacle determination information from the first AI camera 1, and operates an opening / closing operation information acquisition program for obtaining an opening / closing operation signal, and further receives human detection determination information from the second AI camera 2, and operates an alarm information acquisition program for obtaining an alarm instruction signal.

[0083] In addition, the first camera 11 of the first AI camera 1 and the second camera 21 of the second AI camera 2 may be network cameras, and the flow direction detection unit 14, water level detection unit 15, flow speed detection unit 16, obstacle detection unit 17 in the first AI camera 1, the person detection unit 22 in the second AI camera 2, and the signal judgment unit 32 in the floodgate control system 3 may be realized by a hardware configuration using the same computer, and each program may be stored in the same ROM or other recording medium.

[0084] The automatic control unit 34 controls the opening and closing operation of the water gate power unit 35 . The automatic control unit 34 receives a closing operation signal from the signal determination unit 32 and controls the water gate power unit 35 to close the water gate 100. Furthermore, when the floodgate 100 is open, the automatic control unit 34 controls the floodgate power unit 35 to close the floodgate 100 in response to a request from the manager's PC 9 to close the floodgate 100 . When the floodgate 100 is closed, the automatic control unit 34 performs control for the floodgate power unit 35 to open the floodgate 100 in response to a request from the manager's PC 9 to open the floodgate 100 .

[0085] The abnormality detection unit 36 ​​monitors the water gate power unit 35, and if an error occurs in the opening / closing operation of the water gate 100 by the water gate power unit 35, it outputs a stop signal to the automatic control unit 34 to stop the opening / closing operation of the water gate 100. The automatic control unit 34, the floodgate power unit 35, and the abnormality detection unit 36 ​​are generally known units such as a power unit such as a motor that opens and closes the floodgate 100, a control unit that controls the motor, and an abnormality detection unit that monitors the motor for abnormalities, so detailed explanations will be omitted.

[0086] Next, the operation of the watergate opening and closing system according to the first embodiment will be described for each component. First, the operation of obtaining information required for opening and closing the floodgate 100, mainly the operation of obtaining information required for opening and closing the floodgate 100 using the first AI camera 1 and the second AI camera 2, will be described. First, the operation of flow direction detection unit 14 in first AI camera 1 will be described with reference to FIG. Since the flow direction detection unit 14 is realized by a hardware configuration of a computer, the following operations are realized by a program stored in the ROM of the computer, that is, a flow direction information acquisition program.

[0087] In step ST11, the flow direction analysis unit 14a compares the video data based on the imaging information from the first camera 11 (hereinafter simply referred to as the video data from the first camera 11) with each video data in the flow direction information learning data, obtains video data from the flow direction information learning data that has the same or closest video information as the video data from the first camera 11, and detects the flow angle in the video data from the first camera 11, that is, the flow direction angle from 0 degrees to 360 degrees. Step ST11 is a flow angle detection step, which is a flow angle detection procedure in the flow direction information acquisition program.

[0088] In step ST12, the flow direction determining unit 14b determines whether or not the flow angle in the video data from the first camera 11 is within the range of 330° to 30°. If it is within the range, it is determined that there is a backflow and the process proceeds to step ST13, and if it is outside the range, the process proceeds to step ST14. Step ST12 is a backflow determination step, which is a backflow determination procedure in the flow direction information acquisition program.

[0089] In step ST14, the flow direction determining unit 14b determines whether or not the flow angle in the video data from the first camera 11 is within the range of 150° to 210°. If it is within the range, it is determined to be a forward flow and the process proceeds to step ST15, whereas if it is outside the range, it is determined to be a stagnation and the process proceeds to step ST16. Step ST14 is a forward flow / stay determination step, which is a procedure for determining forward flow / stay in the flow direction information acquisition program. Steps ST12 to ST16 are flow direction determination steps, which are the flow direction determination procedures in the flow direction information acquisition program.

[0090] Steps ST11 to ST16 are executed multiple times within a set period, and in step ST17, the most common determination result among the backflow determination result shown in step ST13, the forward flow determination result shown in step ST15, and the stagnation determination result shown in step ST16 is determined to be the flow direction of the river, and flow direction determination information indicating either backflow, forward flow, or stagnation is obtained. Step ST17 is a flow direction calculation step, which is a flow direction calculation procedure in the flow direction information acquisition program.

[0091] The operation of the water level detection unit 15 in the first AI camera 1 will be described with reference to FIG. Since the water level detection unit 15 is realized by a hardware configuration of a computer, the following operations are realized by a program stored in the ROM of the computer, that is, a water level information acquisition program.

[0092] In step ST21, the water level analysis unit 15a compares the image data from the first camera 11 with each image data in the water level information learning data, obtains image data from the water level information learning data that has the same or closest image information to the image data from the first camera 11, and detects water areas in the image data from the first camera 11. Step ST21 is a water area detection step, which is a water area detection procedure in the water level information acquisition program.

[0093] In step ST22, the water level determination unit 15b determines whether the water area in the video data from the first camera 11 exceeds the overflow line, that is, whether it is an abnormal water area. If it does exceed the threshold, it is determined that the water area is abnormal and the process proceeds to step ST23, and if it does not exceed the threshold, it is determined that the water area is normal and the process proceeds to step ST24. Step ST22 is a water level determination step, which is a water level determination procedure in the water level information acquisition program.

[0094] The operation of the flow velocity detection unit 16 in the first AI camera 1 will be described with reference to FIG. Since the flow velocity detection unit 16 is realized by a hardware configuration of a computer, the following operations are realized by a program stored in the ROM of the computer, that is, a flow velocity information acquisition program.

[0095] In step ST31, the flow velocity analysis unit 16a compares the image data from the first camera 11 with each image data in the flow velocity information learning data, obtains image data from the flow velocity information learning data that has the same or closest image information as the image data from the first camera 11, and detects the flow velocity in the image data from the first camera 11, that is, the river flow velocity between 0 m / s and 10 m / s. Step ST31 is a flow velocity detection step, which is a flow velocity detection procedure in the flow velocity information acquisition program.

[0096] In step ST32, the flow velocity determining unit 16b determines whether or not the flow velocity in the video data from the first camera 11 is 3 m / s or more. If it is equal to or greater than this, it is determined to be fast and the process proceeds to step ST33, and if it is less than this, the process proceeds to step ST34. Step ST32 is a fast current determination step, which is a procedure for determining a fast current in the flow velocity information acquisition program.

[0097] In step ST34, the flow velocity determining unit 16b determines whether or not the flow velocity in the video data from the first camera 11 is within a range of 1 m / s or more and less than 3 m / s. If it is within the range, it is determined that the speed is slow, and the process proceeds to step ST35, whereas if it is outside the range, it is determined that the speed is stagnant, and the process proceeds to step ST36. Step ST34 is a slow flow / stagnation determination step, which is a procedure for determining slow flow / stagnation in the flow velocity information acquisition program. Steps ST32 to ST36 are flow velocity determination steps, which are the flow velocity determination procedures in the flow velocity information acquisition program.

[0098] Steps ST31 to ST36 are executed multiple times within a set period, and in step ST37, the most common judgment result among the judgment result of "fast" shown in step ST33, the judgment result of "slow" shown in step ST35, and the judgment result of "stagnant" shown in step ST36 is judged to be the speed of the river, and flow velocity judgment information indicating either fast, slow, or stagnant is obtained. Step ST37 is a flow velocity calculation step, which is a flow velocity calculation procedure in the flow velocity information acquisition program.

[0099] The operation of the obstacle detection unit 17 in the first AI camera 1 will be described with reference to FIG. Since the obstacle detection unit 17 is realized by a hardware configuration of a computer, the following operations are realized by a program stored in the ROM of the computer, that is, an obstacle information acquisition program.

[0100] In step ST41, the obstacle analysis unit 17a compares the image data from the first camera 11 with each image data in the obstacle information learning data, and detects an obstacle from the image data from the first camera 11 using the image data in the obstacle information learning data. The number of detected obstacles is set to 0 to 10. Step ST41 is an obstacle detection step, which is an obstacle detection procedure in the obstacle information acquisition program. When flow velocity determination information indicating fast is obtained in step ST37 in the flow velocity information acquisition step, step ST41 is executed at half the fixed interval for detecting obstacles compared to when flow velocity determination information indicating either slow or stagnant is obtained.

[0101] In step ST42, the obstacle determining unit 17b determines whether or not one or more obstacles are detected in the image data from the first camera 11. If one or more obstacles are detected, it is determined that there is an obstacle and the process proceeds to step ST43, whereas if no obstacle is detected, it is determined that there is no obstacle and the process proceeds to step ST44. Step ST42 is an obstacle determination step, which is an obstacle determination procedure in the obstacle information acquisition program.

[0102] The operation of the human detection unit 22 in the second AI camera 2 will be described with reference to FIG. Since the human detection unit 22 is realized by a hardware configuration of a computer, the following operations are realized by a program stored in the ROM of the computer, that is, a human detection information acquisition program.

[0103] In step ST51, the person analysis unit 22a compares the video data from the second camera 21 with each piece of video data in the person information learning data, and detects a person from the video data from the second camera 21 using the video data in the person information learning data. Step ST51 is a human detection step, which is a human detection procedure in the human detection information acquisition program.

[0104] In step ST52, the person determining section 22b determines whether or not one or more people are present in the video data from the second camera 21. If one or more people are detected, it is determined that a person is present and the process proceeds to step ST53, whereas if no person is detected, it is determined that a person is not present and the process proceeds to step ST54. Step ST52 is a person determination step, which is a person determination procedure in the person information acquisition program.

[0105] Next, we will explain the operations of the floodgate control system 3 using the information necessary for opening and closing the floodgate 100 obtained by the first AI camera 1 and the second AI camera 2, mainly the operations of closing the floodgate and issuing an alarm. First, the operation of the signal determination unit 32 will be described with reference to FIG. Since the signal determination unit 32 is realized by a hardware configuration of a computer, the following operations are realized by programs stored in the ROM of the computer, that is, an opening / closing operation information acquisition program and an alarm information acquisition program.

[0106] A case where the floodgate 100 is open will be described. In step ST61, the signal determination unit 32 determines the presence or absence of an obstacle in the river basin near the floodgate 100 based on the obstacle determination information received by the signal receiving unit 31 from the first AI camera 1. If the obstacle determination information indicates the presence of an obstacle, the process proceeds to step ST62, where an opening operation signal for maintaining the open state of the floodgate 100 is obtained, and the opening operation signal is output to the automatic control unit 34 via the opening / closing signal output unit 33.

[0107] The opening operation signal may be a non-signal since it is sufficient to maintain the open state of the floodgate 100. Also, during the closing operation of the floodgate 100, if the obstacle determination information indicates the presence of an obstacle in step ST61, the opening operation signal is a signal to stop the closing operation of the floodgate 100 or a signal to change the operation of the floodgate 100 to an opening operation.

[0108] If the obstacle determination information indicates that there is no obstacle, the process proceeds to step ST63. Step ST62 is the first step of acquiring opening / closing operation information based on obstacle determination information, and is the first procedure based on the obstacle determination information in the opening / closing operation information acquisition program. In step ST63, the signal determination unit 32 determines, based on the water level determination information received by the signal receiving unit 31 from the first AI camera 1, whether or not the watershed in the river near the floodgate 100 is a water area beyond the overflow line.

[0109] If the water level determination information indicates an abnormal water area, the process proceeds to step ST62, where an opening operation signal for maintaining the open state of the water gate 100 is obtained, and the opening operation signal is output to the automatic control unit 34 via the opening / closing signal output unit 33. If the water level determination information indicates a steady water area, the process proceeds to step ST64. Step ST64 is a second step of acquiring opening / closing operation information based on water level determination information, and is a second procedure based on obstacle determination information in the opening / closing operation information acquisition program.

[0110] In step ST64, the signal determination unit 32 determines the river flow near the floodgate 100 in the river based on the flow direction determination information from the first AI camera 1 received by the signal receiving unit 31. If the flow direction determination information indicates a reverse flow, the process proceeds to step ST65, where a closing operation signal for closing the water gate 100 is obtained, and the closing operation signal is output to the automatic control unit 34 via the opening / closing signal output unit 33.

[0111] If the flow direction determination information indicates a forward flow or stagnation, the process proceeds to step ST62, where an opening operation signal is obtained to maintain the open state of the floodgate 100, and the opening operation signal is output to the automatic control unit 34 via the opening / closing signal output unit 33. Step ST65 is a third step of acquiring opening / closing operation information based on flow direction determination information, and is a third procedure based on obstacle determination information in the opening / closing operation information acquisition program.

[0112] The automatic control unit 34 controls the water gate power unit 35 using the opening operation signal and closing operation signal input via the opening / closing signal output unit 33, thereby controlling the opening and closing operation of the water gate 100. That is, when the water gate 100 is open and an opening operation signal is input, the automatic control unit 34 keeps the water gate power unit 35 stopped and keeps the water gate 100 open. On the other hand, when the floodgate 100 is open and a closing operation signal is input, the automatic control unit 34 drives the floodgate power unit 35 to close the floodgate 100 .

[0113] In addition, when the automatic control unit 34 receives an opening operation signal in step 61 while controlling the closing operation of the floodgate 100, it stops the floodgate power unit 35 or drives it to open the floodgate 100, thereby stopping or opening the floodgate 100. Furthermore, when an opening operation signal for opening the floodgate 100 is input from the manager's PC 9 while the floodgate 100 is closed, the automatic control unit 34 drives the floodgate power unit 35 to open the floodgate 100.

[0114] The operation of issuing an alarm in the floodgate control system 3 is as follows: in step ST71, the signal determination unit 32 determines whether or not a person is present around the floodgate 100 in the river based on the person detection determination information received by the signal receiving unit 31 from the second AI camera 2. If the person detection determination information indicates the presence of a person, the process proceeds to step ST72.

[0115] In step ST72, the signal judgment unit 32 determines that a closing operation signal for closing the floodgate 100 has been obtained in step ST64 and that the signal indicates the presence of a person in step ST72, and therefore determines to issue an alarm and obtains alarm instruction information, which is output to the alarm device 5 via the alarm signal output unit 37. Steps ST71 and ST72 are alarm information acquisition steps, which are procedures of an alarm information acquisition program.

[0116] As described above, the first image analysis device 1 in the floodgate opening and closing system of embodiment 1 is equipped with a water level detection unit 15 that compares the image data from the imaging information from the first camera 11 with water level information learning data for indicating water level information set by image data based on multiple images to obtain water level determination information, and a flow direction detection unit 14 that compares the image data from the imaging information from the first camera 11 with flow direction information learning data for indicating flow direction information set by image data based on multiple images to obtain flow direction determination information, so that the water level and flow direction in the water area near the floodgate 100 of the river can be obtained without using a water level gauge and a flow direction gauge, and since the flow direction determination information and water level determination information are obtained by image analysis of the water area near the floodgate 100 of the river, the water level and flow direction can be obtained with high accuracy.

[0117] In addition, the flow direction information learning data and water level information learning data can use data corresponding to the water area near the river gate 100, and flow direction determination information and water level determination information appropriate for the water area near the river gate 100 can be obtained.

[0118] The first image analysis device 1 in the floodgate opening and closing system of embodiment 1 is equipped with an obstacle detection unit 17 that compares the image data based on the imaging information from the first camera 11 with obstacle information learning data for indicating obstacles set by image data based on multiple images to obtain obstacle determination information, so that the presence or absence of an obstacle can be obtained with high accuracy, and the closing operation of the floodgate 100 can be performed more safely by using the obstacle determination information.

[0119] The first image analysis device 1 in the floodgate opening and closing system of embodiment 1 is equipped with a flow velocity detection unit 16 that compares video data based on imaging information from the first camera 11 with flow velocity information learning data for indicating flow velocity information set by video data based on multiple images, and obtains flow velocity judgment information.When the flow velocity judgment information indicates that the flow velocity is fast, the obstacle detection unit 17 sets the fixed interval for obtaining obstacle judgment information to 1 / 2 the fixed interval for obtaining obstacle judgment information when the flow velocity judgment information indicates that the flow velocity is slow or stagnant.This makes it possible to detect early on an obstacle flowing into the floodgate 100 during the closing operation of the floodgate 100 due to the fast flow of the river, and prevents the obstacle from coming into contact with the floodgate 100, or otherwise interfering with the closing operation of the floodgate 100.

[0120] In the floodgate opening and closing system of embodiment 1, the floodgate control system 3 determines the water level determination information and flow direction determination information from the first image analysis device 1 and controls the opening and closing operation of the floodgate, so that the floodgate control system can safely close the floodgate using highly accurate water levels and flow directions of the river. In the floodgate opening and closing system of embodiment 1, the floodgate control system 3 judges the person detection judgment information from the second image analysis device, and when it outputs a floodgate closing operation signal, if the person detection judgment information indicates the presence of a person, it outputs warning instruction information to an alarm, so that people present in the vicinity of the floodgate 100 can be appropriately notified of the dangerous condition of the river.

[0121] The floodgate opening and closing system of embodiment 1 is equipped with a management operation device having an operation unit which outputs an operation instruction signal for selectively operating the water level detection unit 15 and the flow direction detection unit 14 in the first image analysis device 1, the water level detection unit 15 generating water level determination result video information showing an image representing the water level based on the water level determination information, the flow direction detection unit 14 generating flow direction determination result video information showing an image representing the direction of the river flow based on the flow direction determination information, and a display unit which receives input of the water level determination result video information and displays an image based on the input water level determination result video information, and receives input of flow direction determination result video information and displays an image based on the input flow direction determination result video information, thereby allowing the water level and flow direction of the river to be confirmed as an image with high accuracy.

[0122] The floodgate opening and closing system of embodiment 1 is equipped with a management operation device having a human detection unit 22 in the second video analysis device 2 which generates human detection judgment result video information showing an image showing the presence of a person based on the human detection judgment information, an operation unit which outputs an operation instruction signal to operate the human detection unit 22 in the second video analysis device 2, and a display unit which receives input of the human detection judgment result video information and displays an image based on the input human detection judgment result video information, so that people present around the floodgate 100 can be confirmed as an image.

[0123] Furthermore, in the floodgate opening and closing system of embodiment 1, the management operating device can output alarm generation information to an alarm to be issued to people displayed on the display unit, so that people in the vicinity of the floodgate 100 can be appropriately notified of the dangerous state of the river, improving safety for people.

[0124] In the first embodiment, any of the components may be modified or omitted. [Industrial Applicability]

[0125] The image analysis device in the floodgate opening and closing system of the present disclosure is suitable as a device for determining the water level and flow direction in the water area near a floodgate 100 on a river in order to automatically close the flood gates of the floodgate installed to prevent backflow of seawater or river water due to a tsunami or high tide. [Explanation of symbols]

[0126] 1 First image analysis device, 11 First camera, 14 Flow direction detection unit, 15 Water level detection unit, 16 Flow speed detection unit, 17 Obstacle detection unit, 2 Second image analysis device, 21 Second camera, 22 Person detection unit, 3 Floodgate control system, 31 Signal receiving unit, 32 Signal judgment unit, 33 Opening / closing signal output unit, 34 Automatic control unit, 35 Floodgate power unit, 36 Abnormality detection unit, 37 Alarm signal output unit, 5 Alarm, 9 Management operation device.

Claims

1. A camera and a water level detection unit that obtains water level determination information by comparing the image data based on the imaging information from the camera with water level information learning data for indicating water level information set by the image data based on a plurality of images; a flow direction detection unit that obtains flow direction determination information by comparing video data based on imaging information from the camera with flow direction information learning data for indicating flow direction information set by video data based on a plurality of images; A video analysis device comprising:

2. The water level determination information is information that indicates an abnormal water area when the image data from the imaging information from the camera indicates a water area above the overflow line as a result of comparison with the water level information learning data, and indicates a normal water area when the image data from the imaging information from the camera indicates a water area below the overflow line, The flow direction determination information is information indicating a reverse flow when the image data from the imaging information from the camera indicates first angle information as a result of comparison with the flow direction information learning data, a forward flow when the image data indicates second angle information different from the first angle information, and a stagnation when the image data indicates third angle information other than the first angle information and the second angle information. The video analysis device according to claim 1 .

3. The water level detection unit has a water level analysis unit which compares the image data based on the imaging information from the camera with each image data in the water level information learning data, and obtains image data from the water level information learning data having image information that is the same as or closest to the image data based on the imaging information from the camera, and a water level determination unit which obtains information indicating the water level of the image data based on the imaging information from the camera or information indicating an abnormal water area or a steady water area from the image data in the water level information learning data obtained by the water level analysis unit, and obtains water level determination information indicating an abnormal water area or a steady water area, The flow direction detection unit has a flow direction analysis unit that compares the image data based on the imaging information from the camera with each image data in the flow direction information learning data, and obtains image data from the flow direction information learning data having image information that is the same as or closest to the image data based on the imaging information from the camera, and a flow direction determination unit that obtains angle information or information indicating a flow angle of the image data based on the imaging information from the camera from the image data in the flow direction information learning data obtained by the flow direction analysis unit, and obtains the flow direction determination information indicating a reverse flow, a forward flow, and a stagnation. The video analysis device according to claim 1 .

4. an obstacle detection unit that obtains obstacle determination information by comparing image data based on the image pickup information from the camera with obstacle information learning data for indicating obstacles set based on image data based on a plurality of images; a flow velocity detection unit that obtains flow velocity determination information by comparing image data based on the imaging information from the camera with flow velocity information learning data for indicating flow velocity information set by image data based on a plurality of images; The video analysis device according to claim 1 .

5. The obstacle determination information is information indicating the presence of an obstacle when an obstacle is detected in the video data based on the image capture information from the camera as a result of a comparison with the obstacle information learning data, and indicating the absence of an obstacle when an obstacle is not detected, The flow velocity determination information is information indicating a flow velocity that is fast when the image data from the imaging information from the camera indicates a flow velocity that is equal to or greater than a first threshold value of the flow velocity as a result of comparison with the flow velocity information learning data, that is slow when the image data indicates a flow velocity that is less than a second threshold value that is slower than the first flow velocity, and that is stagnant when the image data indicates a flow velocity that is equal to or greater than the second threshold value and less than the first threshold value. The video analysis device according to claim 4.

6. The video analysis device according to claim 5, wherein the obstacle detection unit receives the flow velocity determination information obtained by the flow velocity detection unit, and when the flow velocity determination information indicates that the flow velocity is fast, sets the fixed interval for obtaining the obstacle determination information to 1 / 2 of the fixed interval for obtaining the obstacle determination information when the flow velocity determination information indicates that the flow velocity is slow or stagnant.

7. The obstacle detection unit includes an obstacle analysis unit which compares video data based on image capture information from the camera with each piece of video data in the obstacle information learning data, and detects an obstacle from the video data based on image capture information from the camera using the video data in the obstacle information learning data, and an obstacle determination unit which counts the number of obstacles detected in the video data based on image capture information from the camera obtained by the obstacle analysis unit, and if the number of obstacles is 1 or more, determines that an obstacle is present and obtains obstacle determination information indicating the presence of an obstacle, and if the number of obstacles is 0, determines that an obstacle is absent and obtains obstacle determination information indicating the absence of an obstacle, The flow velocity detection unit has a flow velocity analysis unit which compares the image data based on the imaging information from the camera with each image data in the flow velocity information learning data, and obtains image data having image information that is the same as or closest to the image data based on the imaging information from the camera from the flow velocity information learning data, and a flow velocity determination unit which obtains information indicating the flow velocity of the image data based on the imaging information from the camera or information indicating whether the flow of the river is fast, slow, or stagnant from the image data in the flow velocity information learning data obtained by the flow velocity analysis unit, and obtains flow velocity determination information indicating whether the flow of the river is fast, slow, or stagnant. The video analysis device according to claim 4.

8. The video analysis device according to claim 7, wherein the obstacle detection unit receives flow velocity determination information obtained by the flow velocity detection unit, and when the flow velocity determination information indicates that the flow velocity is fast, sets the fixed interval for obtaining the obstacle determination information to 1 / 2 of the fixed interval for obtaining the obstacle determination information when the flow velocity determination information indicates that the flow velocity is slow or stagnant.

9. A video analysis device according to any one of claims 1 to 3, A water gate control system that determines water level determination information and flow direction determination information from the image analysis device and controls the opening and closing operation of the water gate; A floodgate opening and closing system comprising:

10. The watergate opening / closing system according to claim 9 , wherein the watergate control system closes the watergate when the water level determination information indicates a steady water area and the flow direction determination information indicates a reverse flow.

11. the video analysis device includes an obstacle detection unit that obtains obstacle determination information by comparing video data based on imaging information from the camera with obstacle information learning data for indicating obstacles set by video data based on a plurality of videos; The floodgate control system closes the floodgate on condition that the obstacle determination information from the image analyzer indicates that there is no obstacle. The floodgate opening and closing system according to claim 10.

12. a second video analysis device having a second camera and a human detection unit that compares video data based on imaging information from the second camera with human information learning data for indicating a person set by video data based on a plurality of videos, and obtains human detection determination information indicating the presence or absence of a person; The water gate control system analyzes and determines the flow direction determination information and water level determination information from the video analysis device, and when the water gate is to be closed, if the human detection determination information from the second video analysis device indicates the presence of a person, outputs warning instruction information to an alarm. The floodgate opening and closing system according to claim 9.

13. The water level detection unit in the video analysis device generates water level determination result video information showing an image showing a water level according to the water level determination information, and the flow direction detection unit generates flow direction determination result video information showing an image showing a direction of a river flow according to the flow direction determination information, a management operation device having an operation unit that outputs an operation instruction signal for selectively operating a water level detection unit and a flow direction detection unit in the image analysis device, and a display unit to which the water level determination result image information is input and an image based on the input water level determination result image information is displayed, and to which the flow direction determination result image information is input and an image based on the input flow direction determination result image information is displayed, The floodgate opening and closing system according to claim 9.

14. the person detection unit in the second video analysis device generates person detection determination result video information showing a video showing the presence of a person based on the person detection determination information; a management operation device having an operation unit that outputs an operation instruction signal for operating the human detection unit in the second video analysis device, and a display unit to which the human detection determination result video information is input and on which a video based on the input human detection determination result video information is displayed; The floodgate opening and closing system according to claim 12.

15. The floodgate opening and closing system according to claim 14, wherein the operation unit in the management operation device outputs, to the alarm, alarm generation information to be issued to a person displayed on the display unit.

Images