Damage identification device, damage identification method, and program

Abstract

To provide a damage specification device, a damage specification method, and a program capable of specifying damage occurrence places in a river basin.SOLUTION: The damage specification device comprises: an acquisition unit which acquires respective images from a plurality of cameras installed in a river basin having a river extending therethrough; a detection unit which detects the difference between the images; and a specification unit which specifies the position of damage based on the detected difference.SELECTED DRAWING: Figure 2

Description

[Technical Field]

[0001] This disclosure relates to a device for identifying damage, a method for identifying damage, and a program. [Background technology]

[0002] For example, Patent Document 1 discloses a river monitoring device that automatically determines the presence or absence of floating objects using the brightness value of the video signal from an imaging unit that monitors rivers. According to this river monitoring device, it is not necessary for river monitors to patrol and monitor rivers to determine the presence or absence of floating objects. [Prior art documents] [Patent Documents]

[0003] [Patent Document 1] Japanese Patent Application Publication No. 7-85385 [Overview of the project] [Problems that the invention aims to solve]

[0004] Incidentally, when heavy rains occur, landslides and other damage can occur in river basins. In such cases, floating debris (driftwood, etc.) can accumulate in the river, starting from the location of the damage. The technology described in Patent Document 1 above only determines whether or not floating debris is present in the river, so it has the problem of being difficult to pinpoint the location of damage in the river basin.

[0005] This disclosure was made to solve the above-mentioned problems and aims to provide a damage identification device, a damage identification method, and a program that can identify the locations of damage in river basins. [Means for solving the problem]

[0006] In order to solve the above problems, the damage identification device relating to this disclosure is The system comprises an acquisition unit that acquires images of each of the rivers, including the river, from multiple cameras installed in the river basin over which the river extends; a detection unit that detects differences between the images; and an identification unit that identifies the location of damage in the river basin based on the detected differences. Each camera is installed in the river basin to capture one section of the river, which is divided into multiple sections, and each camera captures a different section from the others. The detection unit detects differences in the water level of the river in the images as differences between the images. The identification unit identifies the location of the overflow or levee breach that occurred in the river basin as the location of the damage based on the detected differences. The identification unit identifies the land facing the section as the location of the damage where the overflow or levee breach occurred, when the water level of the river in one section is lower than the water level of the river in a section adjacent to that section from the upstream side.

[0007] The damage identification method according to the present disclosure is The system performs the steps of: acquiring images of each image including the river from multiple cameras installed in a river basin over which the river extends; detecting differences between each of the images; and identifying the location of damage in the river basin based on the detected differences. Each camera is installed in the river basin to capture one section of the river which is divided into multiple sections, and each camera captures a different section from each other. In the step of detecting differences between the images, differences in the water level of the river in the images are detected as differences between the images. In the step of identifying the location of damage, the location of the overflow or breach that occurred in the river basin is identified as the location of the damage based on the detected differences. In the step of identifying the location of damage, if the water level of the river in one section is lower than the water level of the river in a section adjacent to that section from the upstream side, the land facing that section is identified as the location of the damage where the overflow or breach occurred.

[0008] The program according to the present disclosure is The computer is made to perform the following steps: acquire images of each of the rivers from multiple cameras installed in the river basin over which the river extends; detect differences between each of the images; and identify the location of the damage in the river basin based on the detected differences. Each camera is installed in the river basin to capture one section of the river which is divided into multiple sections, and each camera captures a different section from each other. In the step of detecting differences between the images, differences in the water level of the river in the images are detected as differences between the images. In the step of identifying the location of the damage, the location of the overflow or breach that occurred in the river basin is identified as the location of the damage based on the detected differences. In the step of identifying the location of the damage, if the water level of the river in one section is lower than the water level of the river in a section adjacent to that section from the upstream side, the land facing that section is identified as the location of the damage where the overflow or breach occurred. The damage identification device according to this disclosure comprises: an acquisition unit that acquires images including the river from multiple cameras installed in a river basin over which the river extends; a detection unit that detects differences between the images; and an identification unit that identifies the location of damage in the river basin based on the detected differences. Each camera is installed in the river basin to capture one section of the river, which is divided into multiple sections, and each camera captures a different section from the others. The detection unit detects the difference between the amount of debris flowing in one section of the river and the amount of debris flowing in other sections upstream or downstream of that section as a difference between the images. The detection unit then uses pre-acquired data for each section. Using a pre-trained model that has been repeatedly machine-learned with a dataset in which images and features at the time of image acquisition are input data and classifications indicating the amount of driftwood in the images are output data, the difference between the amount of driftwood flowing in one section and the amount of driftwood flowing in other sections is detected as a difference between each of the images. The features at the time of image acquisition include time data, seasonal data, water level data, and water temperature data at the time of image acquisition. The detection unit inputs the images and features received from the acquisition unit into the pre-trained model, and as an output of the pre-trained model, it obtains the amount of driftwood and detects the difference between each of the images. The damage identification method relating to this disclosure includes the steps of: acquiring images of the river from multiple cameras installed in the river basin over which the river extends; detecting differences between the images; and identifying the location of the damage in the river basin based on the detected differences. Each camera is installed in the river basin to capture one section of the river, which is divided into multiple sections, and each camera captures a different section from the others. In the step of detecting differences between the images, the previously acquired images of each section and the feature quantities at the time of acquisition of the images are used as input data to indicate the amount of debris in the images. Using a pre-trained model that has been repeatedly machine-learned on a dataset whose output data is classification, the difference between the amount of driftwood flowing in one section and the amount of driftwood flowing in the other section is detected as a difference between each of the images. The features at the time of image acquisition include the time data, seasonal data, water level data, and water temperature data at the time of image acquisition. In the step of detecting the difference between each of the images, the images and features received in the acquisition step are input to the pre-trained model, and the amount of driftwood is obtained as the output of the pre-trained model, and the difference between each of the images is detected. The program relating to this disclosure causes a computer to perform the following steps: acquire images of the river from multiple cameras installed in a river basin over which the river extends; detect differences between the images; and identify the location of damage in the river basin based on the detected differences. Each camera is installed in the river basin to capture one section of the river, which is divided into multiple sections, and each camera captures a different section from the others. In the step of detecting differences between the images, a trained model that has been repeatedly trained on a dataset in which previously acquired images of each section and feature quantities at the time of acquisition of the images are input data, and classifications indicating the amount of debris in the images are output data, is used to detect the difference between the amount of debris flowing in one section and the amount of debris flowing in other sections as differences between the images. The feature quantities at the time of acquisition of the images are the time data at the time of image capture, the seasonal data at the time of image capture, the water level data at the time of image capture, and the water temperature data at the time of image capture. include In the step of detecting differences between each of the aforementioned images, the images and features received in the acquisition step are input to the trained model, and the amount of driftwood is obtained as the output of the trained model, and differences between each of the aforementioned images are detected. <000006%>

Effect of the Invention

[0009] According to the present disclosure, it is possible to provide a damage identification device, a damage identification method, and a program capable of identifying the location where damage occurs in a river basin.

Brief Description of the Drawings

[0010] [Figure 1] It is a diagram showing a schematic configuration of a river basin according to an embodiment of the present disclosure. [Figure 2] It is a functional block diagram of a damage identification device according to a first embodiment of the present disclosure. [Figure 3] It is a flowchart showing the operation of a damage identification device according to a first embodiment of the present disclosure. [Figure 4] It is a flowchart showing the operation of a damage identification device according to a second embodiment of the present disclosure. [Figure 5] It is a flowchart showing the operation of a damage identification device according to a third embodiment of the present disclosure. [Figure 6] It is a functional block diagram of a damage identification device according to a fourth embodiment of the present disclosure. [Figure 7] It is a flowchart showing the operation of a damage identification device according to a fourth embodiment of the present disclosure. [Figure 8] It is a hardware configuration diagram showing the configuration of a computer according to an embodiment of the present disclosure. [Figure 9] It is a flowchart showing the operation of a damage identification device according to another embodiment of the present disclosure.

Modes for Carrying Out the Invention

[0011] <First Embodiment> Hereinafter, a form for implementing the damage identification system according to the present disclosure will be described with reference to the accompanying drawings. The damage identification system in this embodiment is a system for identifying the locations where damage occurs in a river basin. First, the river basin Rv will be described using FIG. 1.

[0012] The river basin Rv is the range where rain, snow, etc. that fall on the ground flow into the river R. The river R extends meandering from the mountainous area on the upstream side towards the sea on the downstream side. The river R in this embodiment is sandwiched by land, and, for example, levees B are provided on both banks of the river R. Also, a plurality of bridges Br are spanned across the river R. In FIG. 1, only one bridge Br is illustrated.

[0013] The river R is divided into a plurality of sections in the direction in which the river R extends. In this embodiment, the case where the river R is divided into seven sections will be described as an example. The divided sections of the river R may be set, for example, for each evacuation unit. Here, the “evacuation unit” means a regional aggregation (such as a large village or a school district) that is the target of evacuation instructions, and is designated, for example, by the administrative area where the river basin Rv exists.

[0014] Hereinafter, for convenience of explanation, each section of the river R divided into seven sections will be referred to in order from the downstream side as the “first section I1”, “second section I2”, “third section I3”, “fourth section I4”, “fifth section I5”, “sixth section I6”, and “seventh section I7”.

[0015] The damage identification system 1 is introduced into the river basin Rv. The damage identification system 1 in this embodiment includes a camera 20 and a damage identification device 10.

[0016] (Camera) Multiple cameras 20 are installed in the river basin Rv and are capable of acquiring visible images. The multiple cameras 20 are arranged in a line along the direction in which the river R extends. The number of cameras 20 installed in the river basin Rv is the same as the number of divisions in the river R. Therefore, in this embodiment, a configuration in which seven cameras 20 are arranged in a line along the direction in which the river R extends will be described as an example. Each camera 20 is installed in the river basin Rv so as to capture one section of the river R which is divided into multiple sections, and so as to each camera 20 so as to capture different sections from each other.

[0017] For the sake of explanation, the camera 20 that captures the entirety of the first section I1 of river R will be referred to as "First Camera 21". The camera 20 that captures the entirety of the second section I2 will be referred to as "Second Camera 22". The camera 20 that captures the entirety of the third section I3 will be referred to as "Third Camera 23". The camera 20 that captures the entirety of the fourth section I4 will be referred to as "Fourth Camera 24". The camera 20 that captures the entirety of the fifth section I5 will be referred to as "Fifth Camera 25". The camera 20 that captures the entirety of the sixth section I6 will be referred to as "Sixth Camera 26". The camera 20 that captures the entirety of the seventh section I7 will be referred to as "Seventh Camera 27".

[0018] Each camera 20 is connected to the damage identification device 10 by wired or wireless connection. Each camera 20 transmits a signal indicating the visible image acquired by capturing an image to the damage identification device 10. For the sake of explanation, the visible image acquired by the camera 20 will be simply referred to as "image". The signal indicating the image transmitted by the camera 20 to the damage identification device 10 includes the time data when the image was acquired.

[0019] (Damage identification device) The damage identification device 10 acquires images transmitted from each camera 20 and identifies the location of damage X occurring in the river basin Rv based on the acquired images. The damage identification device 10 is installed, for example, at the river management office R.

[0020] As shown in Figure 2, the damage identification device 10 in this embodiment includes an acquisition unit 11, a detection unit 12, a identification unit 13, and a notification unit 14.

[0021] (Acquisition Department) The acquisition unit 11 receives signals indicating images transmitted from each camera 20. The acquisition unit 11 sends the received image to the detection unit 12.

[0022] (Detection unit) The detection unit 12 detects the differences between each image based on the images received from the acquisition unit 11. In this embodiment, the detection unit 12 detects the presence or absence of driftwood Dw in each image by, for example, using AI (artificial intelligence).

[0023] The detection unit 12 detects differences between images of adjacent sections by comparing the presence or absence of driftwood Dw in each image. Here, "difference" means that driftwood Dw is present in one section, while driftwood Dw is not present in the section adjacent to this section from the upstream side.

[0024] When the detection unit 12 detects a difference between two images, it sends camera information corresponding to this difference to the identification unit 13. The "camera information" here includes information about which two of the multiple cameras 20 acquired the two images.

[0025] (Specific part) The identification unit 13 identifies the location of damage X based on camera information received from the detection unit 12. Specifically, the identification unit 13 identifies the land facing the downstream section of two adjacent sections determined from the camera information as the location where damage X occurred. For example, if the identification unit 13 receives camera information indicating the first camera 21 and the second camera 22 from the detection unit 12, it identifies the land facing the first section I1 as the location where damage X occurred. Damage X in this embodiment could be, for example, a landslide that causes driftwood Dw to form. The identification unit 13 sends the location of the identified damage X to the notification unit 14.

[0026] (News Department) The notification unit 14 transmits the location of the damage X, which it received from the identification unit 13, to a monitor used by, for example, a river R monitor. In other words, the notification unit 14 notifies the river R monitor that damage X has occurred at the location where it occurred by displaying the location of damage X on the monitor.

[0027] (Operation of the damage identification device) Next, the operation of the damage identification device 10 in this embodiment will be described with reference to Figure 3. First, the acquisition unit 11 of the damage identification device 10 acquires images from multiple cameras 20 (step S1).

[0028] Next, the detection unit 12 of the damage identification device 10 detects the differences between the multiple images acquired in step S1 (step S2). Next, the identification unit 13 of the damage identification device 10 identifies the location of the landslide as damage X based on the differences between the images detected in step S2 (step S3).

[0029] Next, the notification unit 14 of the damage identification device 10 informs the river R monitor that damage X has occurred at the location identified in step S3 (step S4). The damage identification device 10 processes described above, from step S1 to step S4, are repeatedly executed while the river R is being monitored.

[0030] (Effects and Benefits) Due to heavy rainfall and other factors, damage such as landslides may occur in the river basin (Rv). According to the above, the location of damage X within the river basin Rv is identified based on the differences between each detected image. Specifically, by using the presence or absence of driftwood Dw in each image as the difference between the images, the location of the landslide that generated the driftwood Dw is identified as the location of damage X. Therefore, when the location of damage X is identified by the damage identification device 10, for example, a river monitor can quickly and accurately issue evacuation orders to residents living in areas close to the location of damage X.

[0031] <Second Embodiment> Next, the damage identification device 10 according to the second embodiment of this disclosure will be described. The same reference numerals will be used to denote the same parts as in the first embodiment, and redundant explanations will be omitted.

[0032] (Damage identification device) The damage identification device 10 in this embodiment includes an acquisition unit 11, a detection unit 12, a identification unit 13, and a notification unit 14.

[0033] (Acquisition Department) The acquisition unit 11 receives signals indicating images transmitted from each camera 20. The acquisition unit 11 sends the received images to the detection unit 12.

[0034] (Detection unit) The detection unit 12 detects the differences between each image based on the images received from the acquisition unit 11. In this embodiment, the detection unit 12 detects differences between images of adjacent sections by comparing the amount of driftwood Dw in each image. Here, "difference" means that there is less driftwood Dw in a section adjacent to a section downstream from a given section than there is in a given section. When the detection unit 12 detects a difference between images, it sends camera information corresponding to this difference to the identification unit 13.

[0035] (Specific part) The identification unit 13 identifies the location of damage X based on camera information received from the detection unit 12. Specifically, the identification unit 13 identifies the location of damage X as the land facing the upstream section of two adjacent sections determined by the camera information. For example, if the identification unit 13 receives camera information indicating only the fourth camera 24 and the fifth camera 25 from the detection unit 12, the identification unit 13 identifies that damage X has occurred in the fourth section I4. In this case, the identification unit 13 considers the bridge Br in the fourth section I4 as the bridge Br where driftwood Dw has accumulated, and identifies the location of this bridge Br as the location of damage X. Damage X in this embodiment could include, for example, the clogging of bridge Br with driftwood Dw. The identification unit 13 sends the location of the identified damage X to the notification unit 14.

[0036] (News Department) The notification unit 14 transmits the location of the damage X, which it received from the identification unit 13, to a monitor used by, for example, a river R monitor. In other words, the notification unit 14 notifies the river R monitor that damage X has occurred at the location where it occurred by displaying the location of damage X on the monitor.

[0037] (Operation of the damage identification device) Next, the operation of the damage identification device 10 in this embodiment will be described with reference to Figure 4. First, the acquisition unit 11 of the damage identification device 10 acquires images from multiple cameras 20 (step S1').

[0038] Next, the detection unit 12 of the damage identification device 10 detects the differences between the multiple images acquired in step S1' (step S2'). Next, the identification unit 13 of the damage identification device 10 identifies the location of the bridge Br as the damaged object X based on the differences between the images detected in step S2' (step S3').

[0039] Next, the notification unit 14 of the damage identification device 10 informs the river R monitor that damage X has occurred at the location identified in step S3' (step S4'). The damage identification device 10 processes described above from step S1' to step S4' are repeatedly executed while the river R is being monitored.

[0040] (Effects and Benefits) According to the above, the location of damage X is identified based on the differences between each detected image. Specifically, by using the amount of driftwood Dw in each image as the difference between the images, the location of the bridge Br where the driftwood Dw is clogged is identified as the location of damage X. Therefore, when the location of damage X is identified by the damage identification device 10, for example, a river monitor R can determine that there is a possibility of overflow or levee breach in the section where a bridge Br with accumulated driftwood Dw exists. As a result, evacuation orders can be issued quickly and accurately to residents living in areas close to the location of damage X.

[0041] <Third Embodiment> Next, the damage identification device 10 according to the third embodiment of this disclosure will be described. The same reference numerals will be used to denote the same parts as in the first embodiment, and redundant explanations will be omitted. The damage identification device in this embodiment comprises an acquisition unit 11, a detection unit 12, a identification unit 13, and a notification unit 14.

[0042] (Acquisition Department) The acquisition unit 11 receives signals indicating images transmitted from each camera 20. The acquisition unit 11 also acquires feature quantities associated with the received images. Here, "feature quantities" refer to, for example, time data at the time of image capture, seasonal data at the time of image capture, water level data at the time of image capture, water temperature data at the time of image capture, etc.

[0043] Specifically, the acquisition unit 11 acquires time data of the time of shooting from the signal indicating the received image. The acquisition unit 11 also acquires seasonal data of the time of shooting based on the date (for example, four classifications of spring, summer, autumn, and winter) from a computer 1100 or the like on which the damage identification device 10 is implemented. The acquisition unit 11 also acquires water temperature data of the time of shooting by receiving it from a temperature sensor (not shown) located in the river R and capable of measuring the temperature of the water flowing in the river R. The acquisition unit 11 also acquires water level data of the time of shooting by receiving it from a level sensor (not shown) located in the river R and capable of measuring the water level of the water flowing in the river R. The acquisition unit 11 may also acquire the temperature data and water level data from the image on which the data was acquired.

[0044] The acquisition unit 11 sends the acquired image and feature quantities to the detection unit 12.

[0045] (Detection unit) The detection unit 12 detects the differences between each image based on the images and features received from the acquisition unit 11. In this embodiment, the detection unit 12 uses a trained model that has undergone machine learning to detect the amount of driftwood Dw in each image.

[0046] The trained model is generated based on information that correlates multiple images of the target section of river R, which have been acquired in advance, with features such as time data, seasonal data, water temperature data, and water level data from when those images were taken. Specifically, the trained model is generated by repeatedly performing machine learning on a dataset (supervised data) in which these images and features are used as input data and the classification results described below are used as output data.

[0047] Here, the time of day when the image is taken affects, for example, how shadows are formed in the image. The season affects, for example, the color of the image. Water temperature affects, for example, the color of the water in the image. Water level affects, for example, the flow velocity in the target section of river R. If it is difficult to generate a trained model for the target river R, a trained model generated for a river with similar width and other characteristics to river R may be used.

[0048] In the trained model, combinations of images and associated features are classified (clustered) according to the amount of driftwood Dw, based on a predetermined criterion indicating the amount of driftwood Dw. For example, the trained model classifies the amount of driftwood Dw into four stages: "none," "few," "medium," and "many." These "none," "few," "medium," and "many" represent the order in which the amount of driftwood Dw in the image increases. "None" means that there is no driftwood Dw in the image.

[0049] When a pre-trained model is input with a combination of image and features, it outputs a classification (one of the four classifications above) that contains the combination closest to the input. In other words, when the above combination is input, the pre-trained model outputs a classification result indicating the abundance of driftwood Dw.

[0050] The detection unit 12 inputs the image and features received from the acquisition unit into a trained model, thereby obtaining the amount of driftwood Dw as the output of the trained model. The detection unit 12 detects differences between images in adjacent sections by comparing the amount of driftwood Dw in each image. Here, "difference" means that there is less driftwood Dw in a section adjacent to a section downstream from a given section than there is in a given section. When the detection unit 12 detects a difference between images, it sends camera information corresponding to this difference to the identification unit 13.

[0051] (Specific part) The identification unit 13 identifies the location of the damage X based on the camera information received from the detection unit 12. Specifically, the identification unit 13 identifies the location of the damage X as the land facing the upstream section of two adjacent sections determined by the camera information. For example, if the identification unit 13 receives only camera information indicating the fourth camera 24 and the fifth camera 25 from the detection unit 12, the identification unit 13 identifies that the damage X occurred in the fourth section I4. In this embodiment, damage X could be, for example, a landslide that causes driftwood Dw to form. The identification unit 13 sends the location of the identified damage X to the notification unit 14.

[0052] (Operation of the damage identification device) Next, the operation of the damage identification device 10 in this embodiment will be described with reference to Figure 5. First, the acquisition unit 11 of the damage identification device 10 acquires images from multiple cameras 20 and also acquires feature quantities associated with each image (step S10).

[0053] Next, the detection unit 12 of the damage identification device 10 detects the differences between each image based on the amount of driftwood Dw output by inputting the combination of multiple images acquired in step S10 and the features associated with each image into a trained model (step S11). Next, the identification unit 13 of the damage identification device 10 identifies the location of the landslide as damage X based on the differences between each image detected in step S11 (step S12).

[0054] Next, the notification unit 14 of the damage identification device 10 informs the river R monitor that damage X has occurred at the location identified in step S12 (step S13). The damage identification device 10 processes described above from step S10 to step S13 are repeatedly executed while the river R is being monitored.

[0055] (Effects and Benefits) According to the above, the location of damage X is identified based on the differences between each detected image. Specifically, the location of the landslide is identified as the location of damage X by using the difference in the amount of driftwood Dw in each image obtained using the trained model as the difference between the images. Therefore, the location of damage X can be identified with greater precision.

[0056] <Fourth Embodiment> Next, a damage identification device 10 according to the fourth embodiment of this disclosure will be described. The same reference numerals will be used to denote the same parts as in the first embodiment, and redundant explanations will be omitted. As shown in FIG. 6, the damage identification device in the present embodiment includes an acquisition unit 11, a detection unit 12, an identification unit 13, a notification unit 14, and an estimation unit 15.

[0057] (Acquisition Unit) The acquisition unit 11 receives a signal indicating an image transmitted from each camera 20. Further, the acquisition unit 11 acquires the water level in each section by receiving water level data transmitted from a level sensor (not shown) that is arranged in each section of the river R and can measure the water level of the river R. Note that the acquisition unit 11 may acquire the water level data from the acquired image. The acquisition unit 11 transmits a signal indicating the acquired image to the detection unit 12 and sends a signal indicating the acquired water level to the estimation unit 15.

[0058] (Estimation Unit) The estimation unit 15 estimates the time it takes for the floating log Dw to flow down between two sections of the river R based on the water level in each section received from the acquisition unit 11. Hereinafter, the time it takes for the floating log Dw to flow down between two sections is referred to as the "arrival time".

[0059] Specifically, the estimation unit 15 substitutes the water level (D i [[ID=> i : i is any one of 1 to 7) into a predetermined function (W = W(D [[ID=> i [[ID=> i i )) to calculate the width (W [[ID=> i [[ID=> i i : i is any one of 1 to 7) of the water area of the target section. The estimation unit 15 substitutes the acquired water level (D [[ID=> i [[ID=> i i ) and the width of the water area (W [[ID=> i [[ID=> i i ) into a predetermined function (V [[ID=> i [[ID=> i i = V(D [[ID=><000 / / 0007> [[ID=> i i , W [[ID=> i [[ID=> i i )) to calculate the flow velocity (V [[ID=> i [[ID=> i i : i is any one of 1 to 7) in each section. Note that this function may take, for example, the gradient obtained by associating the elevations and distances between a plurality of points in each section with each other or the radius of curvature in each section as arguments.

[0060] [[ID=> The estimation unit 15 calculates the flow velocity (V [[ID=> i [[ID=><C i iBased on this, the average flow velocity between sections (V a Specifically, the detection unit 12 calculates the average flow velocity (V) between one section (for example, the fifth section I5) and another section downstream of this section (for example, the third section I3). a The estimation unit 15 calculates the average flow velocity (V5, V4, V3) by summing the flow velocities (V5, V4, V3) of each section (from the fifth section I5 to the third section I3) and dividing by the number of sections (3). a Calculate ).

[0061] Furthermore, the estimation unit 15 divides the distance between one section and the other section (L), which is pre-stored in the estimation unit 15, by the calculated average flow velocity (L / V a ), estimates the arrival time of driftwood Dw from one section to another. The estimation unit 15 calculates the estimated arrival time (L / V). a The data is sent to the detection unit 12. The damage identification device 10 may also include a storage unit which stores the distance (L) between the two sections in advance, and the estimation unit 15 obtains the distance (L) by referring to the storage unit.

[0062] (Detection unit) The detection unit 12 detects differences between images based on the images received from the acquisition unit 11 and the arrival times received from the estimation unit 15. The detection unit 12 detects differences between images in adjacent sections by comparing the amount of driftwood Dw in each image. Hereinafter, one of these two sections will be referred to as "Section 1," and the section adjacent to Section 1 from the downstream side will be referred to as "the other section."

[0063] In this process, the detection unit 12 detects the difference between an image of one section at the observation time and an image of another section at a time after the arrival time. Here, "difference" means that there is less driftwood Dw in the other section than there is driftwood Dw in the first section. When the detection unit 12 detects a difference between the images, it sends camera information corresponding to this difference to the identification unit 13.

[0064] Furthermore, the detection unit 12 detects the flow state of driftwood Dw over time in each section based on one image for each section received from the acquisition unit 11 and an image acquired at a time prior to the time when the first image was acquired. Hereinafter, an image acquired at a time prior to the time when the first image was acquired will be referred to as the "previous time image".

[0065] Specifically, the detection unit 12 detects the drifting state by detecting the difference between the amount of driftwood Dw in one image for each section and the amount of driftwood Dw in the image from the previous time point. If the detection unit 12 detects that there is no difference between the amount of driftwood Dw in one image and the amount of driftwood Dw in the previous time image, it sends a signal to that effect to the identification unit 13 along with camera information. The "camera information" here refers to information including which of the multiple cameras 20 acquired the two images (one image and the previous time image).

[0066] Specifically, if there is no driftwood Dw in one image or the previous time frame image, the detection unit 12 detects that the drifting state is "no driftwood" and sends a signal indicating this to the identification unit 13 along with camera information. On the other hand, if there is driftwood Dw in one image or the previous time frame image, and the amount is the same, the detection unit 12 detects that the drifting state is "driftwood present" and sends a signal indicating this to the identification unit 13 along with camera information.

[0067] Furthermore, if the detection unit 12 detects a difference between the amount of driftwood Dw in one image and the amount of driftwood Dw in the previous image, it sends a signal to the identification unit 13 indicating this. Specifically, if the amount of driftwood Dw in one image is greater than the amount of driftwood Dw in the previous image, the detection unit 12 detects that the drifting state is "increased driftwood" and sends a signal to the identification unit 13 along with camera information. On the other hand, if the amount of driftwood Dw in one image is less than the amount of driftwood Dw in the previous image, the detection unit 12 detects that the drifting state is "decreased driftwood" and sends a signal to the identification unit 13 along with camera information indicating this.

[0068] (Specific part) The identification unit 13 identifies the location of the damage X based on the camera information received from the detection unit 12. Specifically, the identification unit 13 identifies the location of the damage X as the land facing the upstream section of two adjacent sections determined by the camera information. For example, if the identification unit 13 receives only camera information indicating the fourth camera 24 and the fifth camera 25 from the detection unit 12, the identification unit 13 identifies that the damage X occurred in the fifth section I5. In this embodiment, damage X could be, for example, a landslide that causes driftwood Dw to form.

[0069] Furthermore, the identification unit 13 identifies the location of the damaged area X based on the camera information received from the detection unit 12 and the state of the driftwood Dw's movement. Specifically, the identifying unit 13 identifies that there is no driftwood Dw in a section if the flow condition in one section is "no driftwood" and the flow condition in a section downstream of this section is also "no driftwood".

[0070] Furthermore, the identifying unit 13 identifies that damage X has occurred on the land facing one section or a section upstream of one section if the flow condition in one section (other than "no driftwood") and the flow condition in a section downstream of this section (other than "no driftwood") are the same. The identification unit 13 sends the location of the identified damage X to the notification unit 14.

[0071] (Operation of the damage identification device) Next, the operation of the damage identification device 10 in this embodiment will be described with reference to Figure 7. First, the acquisition unit 11 of the damage identification device 10 acquires images in each section and also acquires the water level in each section (step S20).

[0072] Next, the estimation unit 15 of the damage identification device 10 estimates the arrival time based on the water level obtained in step S20 (step S21).

[0073] Next, the detection unit 12 of the damage identification device 10 detects differences between the multiple images acquired in step S20 and the arrival time estimated in step S21 (step S22). In step S22, the detection unit 12 may also detect the flow state of the driftwood Dw in each section based on one image and an image from the previous time. Next, the identification unit 13 of the damage identification device 10 identifies the location of the landslide as damage X based on the differences between the images detected in step S22 (step S23). In step S23, the identification unit 13 may also identify the location of damage X based on the flow state of driftwood Dw in each section detected in step S22.

[0074] Next, the notification unit 14 of the damage identification device 10 informs the river R monitor that damage X has occurred at the location identified in step S23 (step S24). The damage identification device 10 processes described above from step S20 to step S24 are repeatedly executed while the river R is being monitored.

[0075] (Effects and Benefits) According to the above, the location of damage X is identified based on the differences between each detected image. In this process, the difference between one section and other sections after the arrival time is detected, so the location of damage X can be identified while suppressing the influence of differences in flow velocity between sections. Therefore, the location of damage X can be identified with greater precision.

[0076] (Other embodiments) Although embodiments of this disclosure have been described in detail above with reference to the drawings, the specific configurations are not limited to those of each embodiment, and additions, omissions, substitutions, and other modifications to the configurations are possible without departing from the gist of this disclosure.

[0077] Figure 8 is a hardware configuration diagram showing the configuration of the computer 1100 according to this embodiment. The computer 1100 includes a processor 1110, main memory 1120, storage 1130, and interface 1140.

[0078] The damage identification device 10 described above is implemented in the computer 1100. The operation of each processing unit described above is stored in the storage 1130 in the form of a program. The processor 1110 reads the program from the storage 1130, loads it into the main memory 1120, and executes the above processing according to the program.

[0079] The program may be intended to implement some of the functions that the computer 1100 is to perform. For example, the program may perform its functions in combination with other programs already stored in the storage 1130, or in combination with other programs implemented in other devices.

[0080] Furthermore, in addition to the above configuration, or in place of the above configuration, the computer 1100 may also be equipped with a custom LSI (Large Scale Integrated Circuit) such as a PLD (Programmable Logic Device). Examples of PLDs include PAL (Programmable Array Logic), GAL (Generic Array Logic), CPLD (Complex Programmable Logic Device), and FPGA (Field Programmable Gate Array). In this case, some or all of the functions realized by the processor 1110 may be realized by the integrated circuit.

[0081] Examples of storage 1130 include magnetic disks, magneto-optical disks, and semiconductor memory. Storage 1130 may be an internal medium directly connected to the bus of computer 1100, or an external medium connected to computer 1100 via interface 1140 or a communication line. Furthermore, if this program is distributed to computer 1100 via a communication line, computer 1100 that receives the distribution may expand the program into main memory 1120 and execute the above processing. In the above embodiment, storage 1130 is a tangible storage medium that is not temporary.

[0082] Furthermore, the program may be intended to implement some of the functions described above. Furthermore, the program may be a so-called differential file (differential program) that implements the aforementioned functions in combination with other programs already stored in the storage 1130.

[0083] Furthermore, the detection unit 12 of the damage identification device 10 described in the first embodiment may detect differences between images by comparing the water level of the river R in the images. Here, "difference" means that the water level of the river R in one section is lower than the water level of the river R in a section adjacent to this section from the upstream side. When the detection unit 12 detects a difference between images, it sends camera information corresponding to this difference to the identification unit 13.

[0084] In this case, the identification unit 13 identifies the location of the damage X based on the camera information received from the detection unit 12. For example, if the identification unit 13 receives only camera information indicating the first camera 21 and the second camera 22 from the detection unit 12, it identifies the land facing the first section I1 as the location where the damage X occurred. Examples of this damage X include flooding and dike breaches. The identification unit 13 sends the identified location of the damage X to the notification unit 14.

[0085] Here, the operation of the damage identification device 10 in the above case will be explained with reference to Figure 9. First, the acquisition unit 11 of the damage identification device 10 acquires images from multiple cameras 20 (step S1a).

[0086] Next, the detection unit 12 of the damage identification device 10 detects the differences between the multiple images acquired in step S1a (step S2a). Next, the identification unit 13 of the damage identification device 10 identifies the location of the overflow or levee breach as damage X based on the differences between each image detected in step S2a (step S3a).

[0087] Next, the notification unit 14 of the damage identification device 10 informs the river R monitor that damage X has occurred at the location identified in step S3a (step S4a). The damage identification device 10 processes described above, from step S1a to step S4a, are repeatedly executed while the river R is being monitored.

[0088] Furthermore, the detection unit 12 of the damage identification device 10 described in the first embodiment may detect differences between images by comparing the turbidity of the river R in the images. Here, "difference" means that the turbidity of the river R in one section is higher than the turbidity of the river R in the section adjacent to this section from the upstream side. When the detection unit 12 detects a difference between images, it sends camera information corresponding to this difference to the identification unit 13.

[0089] In this case, the identification unit 13 identifies the location of the damage X based on the camera information received from the detection unit 12. Specifically, if the identification unit 13 receives only camera information indicating the first camera 21 and the second camera 22 from the detection unit 12, it identifies the land facing the first section I1 as the location where the damage X occurred. This damage X could be, for example, a landslide that causes turbidity in the river R. The identification unit 13 sends the identified location of the damage X to the notification unit 14. The operation of this damage identification device 10 is the same as the operation of the damage identification device 10 in the first embodiment described above, so its explanation is omitted.

[0090] Furthermore, the "difference" described in the first embodiment may mean that while driftwood Dw exists in multiple consecutive sections, driftwood Dw does not exist in sections adjacent to these multiple consecutive sections from the upstream side. In this case, the camera information that the detection unit 12 sends to the identification unit 13 only needs to include information from a camera 20 that captures the entire section located furthest upstream among the multiple consecutive sections where driftwood Dw exists, and information from a camera 20 that captures the entire section adjacent to this section from the upstream side.

[0091] Furthermore, the "difference" described in the second embodiment may mean that there is more driftwood Dw in the section adjacent to the first section from the upstream side than there is driftwood Dw in the first section. In this case, when the identification unit 13 identifies the location of the damage X based on the camera information received from the detection unit 12, it may identify that a new landslide has occurred on the land facing the section adjacent to the first section from the upstream side.

[0092] Furthermore, although the above embodiment described that the identification unit 13 of the damage identification device 10 detects differences between images in adjacent sections, it is not limited to this, and it may also detect differences between images in non-adjacent sections.

[0093] Furthermore, although the detection unit 12 described in the above embodiment detects the presence or absence and quantity (abundance or lack thereof) of driftwood Dw in the image as a difference between images, it is not necessarily limited to driftwood Dw, but any object that flows (advection) down the river R is acceptable.

[0094] Furthermore, although the above embodiment describes the camera 20 as being capable of acquiring visible images, it is not limited to this. The camera 20 may be, for example, a camera capable of acquiring infrared images or near-infrared images. In this case, the acquisition unit 11 of the damage identification device 10 described above only needs to acquire infrared images or near-infrared images instead of visible images, and the detection unit 12 only needs to detect the presence or absence and amount of flowing material based on the temperature distribution (thermography) in the infrared images or near-infrared images as differences between the images.

[0095] Furthermore, the means for acquiring the images described in the above embodiment is not limited to the camera 20. For example, instead of the camera 20, multiple radars (Radio Detection and Ranging) may be placed in the river basin Rv to correspond to each section and capable of irradiating the entire area of ​​each section of the river R with radar waves (radio waves). In this case, the radar transmits the acquired radar images to the damage identification device 10 via wired or wireless connection. In the radar image, if there are objects such as floating debris in the target section, the size and position of the objects are displayed by the reflection of radar waves from these objects. The acquisition unit 11 of the damage identification device 10 described in the above embodiment only needs to acquire radar images instead of visible images, and the detection unit 12 only needs to detect the presence and amount of floating debris as differences between images based on plots showing the returned radio waves in the radar image.

[0096] Furthermore, the operation of the damage identification device 10 described in each of the above embodiments is not limited to independent configurations, and the operation of the damage identification device 10 may be configured by appropriately combining the operations described in each embodiment.

[0097] <Note> The damage identification device, damage identification method, and program described in the embodiment can be understood, for example, as follows.

[0098] (1) The damage identification device 10 according to the first embodiment comprises an acquisition unit 11 that acquires images from a plurality of cameras 20 installed in a river basin Rv over which the river R extends, a detection unit 12 that detects differences between the respective images, and an identification unit 13 that identifies the location of the damage X based on the detected differences.

[0099] As a result, by using the differences between images acquired by each of the multiple cameras 20, the location of damage X in the river basin Rv can be identified. For example, a river R monitor can quickly and accurately issue evacuation orders to residents living in areas close to the location of damage X.

[0100] (2) The damage identification device 10 according to the second embodiment is the damage identification device 10 of (1), wherein each camera 20 may be installed in the river basin Rv such that it captures one section of the river R which is divided into multiple sections, and each camera 20 captures different sections from each other.

[0101] This allows the above effects to be achieved with greater precision.

[0102] (3) The damage identification device 10 according to the third embodiment is the damage identification device 10 of (2), wherein the detection unit 12 may detect the difference between the amount of driftwood (driftwood Dw) flowing in one section of the river R and the amount of driftwood flowing in other sections upstream or downstream of the one section as a difference between the respective images.

[0103] This allows for the detection of differences between images as differences in the amount of debris flowing, making it possible to pinpoint the location of damage such as a landslide.

[0104] (4) The damage identification device 10 according to the fourth embodiment is the damage identification device 10 of (3), wherein the identification unit 13 may identify the location of the bridge Br over the river R where the debris has accumulated as the location of the damage X, based on the detected difference.

[0105] This allows the location of the bridge Br where the debris is clogged to be identified as the location of the damage X, so for example, a river monitor can determine that there is a possibility of overtopping or levee breach in the section where this bridge Br is located.

[0106] (5) The damage identification device 10 according to the fifth embodiment is the damage identification device 10 of (1) or (2), wherein the detection unit 12 may detect differences in the degree of turbidity of the river R in the images as differences between each of the images.

[0107] This allows for the detection of differences between images as differences in the turbidity of the river R, making it possible to pinpoint, for example, the location of a landslide.

[0108] (6) In the sixth embodiment of the damage identification device 10, the detection unit 12 may detect differences in the water level of the river R in the images as differences between each of the images, and the identification unit 13 may identify the location of the overtopping or levee breach that occurred in the river basin Rv as the location of the damage X based on the detected differences.

[0109] This allows, for example, a river monitor R to issue evacuation orders quickly and accurately to residents living in areas close to the location of the damage X.

[0110] (7) The damage identification device 10 according to the seventh embodiment is the damage identification device 10 of (3) or (4), wherein the detection unit 12 may use a trained model that has been repeatedly trained on a dataset in which images of each of the previously acquired sections and the feature quantities at the time of acquisition of the images are input data, and classifications indicating the amount of debris flowing in the images are output data, to detect the difference between the amount of debris flowing in one section and the amount of debris flowing in the other sections as a difference between the respective images.

[0111] (8) The damage identification device 10 according to the eighth embodiment is the damage identification device 10 of (3) or (4), further comprising an estimation unit 15 that estimates the time it takes for the debris to flow between two sections in the river R, and the detection unit 12 may determine the upstream section of the two sections as the first section and the downstream section as the other section, and detect the difference between the amount of debris flowing in the first section and the amount of debris flowing in the other section after the time it has flowed as a difference between the respective images.

[0112] (9) The damage identification method according to the ninth aspect includes an acquisition step of acquiring images from multiple cameras 20 installed in a river basin Rv over which the river R extends, a detection step of detecting differences between the images, and an identification step of identifying the location of the damage X based on the detected differences.

[0113] (10) The program according to the eleventh embodiment causes the computer 1100 to perform the following steps: acquire images from multiple cameras 20 installed in a river basin Rv over which the river R extends; detect differences between each of the images; and identify the location of the damage X based on the detected differences. [Explanation of Symbols]

[0114] 1…Damage identification system 10…Damage identification device 11…Acquisition unit 12…Detection unit 13…Identification unit 14…Notification unit 15…Estimation unit 20…Camera 21…First camera 22…Second camera 23…Third camera 24…Fourth camera 25…Fifth camera 26…Sixth camera 27…Seventh camera 1100…Computer 1110…Processor 1120…Main memory 1130…Storage 1140…Interface B…Embankment Br…Bridge Dw…Driftwood I1…First section I2…Second section I3…Third section I4…Fourth section I5…Fifth section I6…Sixth section I7…Seventh section R…River Rv…River basin X…Damage

Claims

1. An acquisition unit that acquires images of each camera, including the river, from multiple cameras installed in the river basin where the river extends, A detection unit for detecting differences between each of the aforementioned images, Based on the detected differences, the identification unit identifies the location of the damage in the river basin, Equipped with, Each camera is installed in the river basin so as to capture one section of the river which is divided into multiple sections, and so as to capture different sections from each other. The detection unit detects the difference in water level of the river in the images as a difference between each of the images, Based on the detected differences, the identifying unit identifies the location of the overflow or levee breach that occurred in the river basin as the location of the damage. The specified unit is a damage identification device that identifies the land facing a particular section as the location of the damage where the overflow or levee breach occurred, when the water level of the river in that particular section is lower than the water level of the river in a section adjacent to that section from the upstream side.

2. The steps include acquiring images of each camera, including the river, from multiple cameras installed in the river basin where the river extends, A step of detecting the differences between each of the aforementioned images, Based on the detected differences, the steps include identifying the location of the damage in the river basin, Execute, Each camera is installed in the river basin so as to capture one section of the river which is divided into multiple sections, and so as to capture different sections from each other. In the step of detecting differences between each of the aforementioned images, the difference in the water level of the river in the images is detected as a difference between each of the aforementioned images. In the step of identifying the location of the damage, based on the detected difference, the location of the overflow or levee breach that occurred in the river basin is identified as the location of the damage. A damage identification method in which, in the step of identifying the location of the damage, if the water level of the river in one section is lower than the water level of the river in a section adjacent to the one section from the upstream side, the land facing the one section is identified as the location of the damage where the overflow or levee breach occurred.

3. The steps include acquiring images of each camera, including the river, from multiple cameras installed in the river basin where the river extends, A step of detecting the differences between each of the aforementioned images, Based on the detected differences, the steps include identifying the location of the damage in the river basin, Have the computer run it, Each camera is installed in the river basin so as to capture one section of the river which is divided into multiple sections, and so as to capture different sections from each other. In the step of detecting differences between each of the aforementioned images, the difference in the water level of the river in the images is detected as a difference between each of the aforementioned images. In the step of identifying the location of the damage, based on the detected difference, the location of the overflow or levee breach that occurred in the river basin is identified as the location of the damage. A program that, in the step of identifying the location of the damage, identifies the land facing the section as the location of the damage where the overflow or levee breach occurred, if the water level of the river in the section is lower than the water level of the river in the section adjacent to the section upstream of the section.

4. An acquisition unit that acquires images of each camera, including the river, from multiple cameras installed in the river basin where the river extends, A detection unit for detecting differences between each of the aforementioned images, Based on the detected differences, the identification unit identifies the location of the damage in the river basin, Equipped with, Each camera is installed in the river basin so as to capture one section of the river which is divided into multiple sections, and so as to capture different sections from each other. The detection unit detects the difference between the amount of debris flowing in one section of the river and the amount of debris flowing in other sections upstream or downstream of that section as a difference between the respective images. The detection unit is Using a pre-trained model that has been repeatedly trained on a dataset in which images of each of the aforementioned sections acquired in advance and the features at the time of acquisition of those images are input data, and a classification indicating the amount of the flowing material in the images is output data, The difference between the amount of debris flowing through one section and the amount of debris flowing through the other sections is detected as a difference between each of the images. The features obtained when acquiring the aforementioned image include the time data at the time of image capture, the seasonal data at the time of image capture, the water level data at the time of image capture, and the water temperature data at the time of image capture. The detection unit inputs the image and feature quantities received from the acquisition unit into a trained model, thereby obtaining the amount of driftwood as the output of the trained model and detecting the differences between each of the images, thus providing a damage identification device.

5. The steps include acquiring images of each camera, including the river, from multiple cameras installed in the river basin where the river extends, A step of detecting the differences between each of the aforementioned images, Based on the detected differences, the steps include identifying the location of the damage in the river basin, Execute, Each camera is installed in the river basin so as to capture one section of the river which is divided into multiple sections, and so as to capture different sections from each other. In the step of detecting differences between each of the aforementioned images, a pre-trained model, which has been repeatedly trained on a dataset in which previously acquired images of each section and the features at the time of acquisition of those images are input data, and classifications indicating the amount of debris flowing in the images are output data, is used to detect the difference between the amount of debris flowing in one section and the amount of debris flowing in the other sections as differences between each of the aforementioned images. The features obtained when acquiring the aforementioned image include the time data at the time of image capture, the seasonal data at the time of image capture, the water level data at the time of image capture, and the water temperature data at the time of image capture. In the step of detecting differences between each of the aforementioned images, the image and the feature quantities received in the acquisition step are input to a trained model, and the amount of driftwood is obtained as the output of the trained model, thereby detecting the differences between each of the aforementioned images, in a method for identifying damage.

6. The steps include acquiring images of each camera, including the river, from multiple cameras installed in the river basin where the river extends, A step of detecting the differences between each of the aforementioned images, Based on the detected differences, the steps include identifying the location of the damage in the river basin, Have the computer run it, Each camera is installed in the river basin so as to capture one section of the river which is divided into multiple sections, and so as to capture different sections from each other. In the step of detecting differences between each of the aforementioned images, a pre-trained model, which has been repeatedly trained on a dataset in which previously acquired images of each section and the features at the time of acquisition of those images are input data, and classifications indicating the amount of debris flowing in the images are output data, is used to detect the difference between the amount of debris flowing in one section and the amount of debris flowing in the other sections as differences between each of the aforementioned images. The features obtained when acquiring the aforementioned image include the time data at the time of image capture, the seasonal data at the time of image capture, the water level data at the time of image capture, and the water temperature data at the time of image capture. In the step of detecting differences between each of the aforementioned images, the program inputs the images and features received in the acquisition step into a trained model, thereby obtaining the amount of driftwood as the output of the trained model and detecting differences between each of the aforementioned images.

Images