A method for setting the time step of monitoring points for water depth video monitoring
Dynamic categorization and time step setting for waterlogging monitoring points in video surveillance optimizes resource use and real-time performance, addressing the inefficiencies of fixed time steps in existing methods.
Patent Information
- Application Number
- CN202410746118.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-06-11
- Publication Date
- 2024-12-13
- Estimated Expiration
- 2044-06-11
AI Technical Summary
Existing waterlogging monitoring methods using fixed time steps in video surveillance data lead to excessive network and computing resource demands or insufficient real-time performance, failing to effectively detect dangerous situations.
Dynamically divide monitoring points into representative and non-representative categories, setting distinct time steps based on water depth and change rate for each category to optimize resource usage and real-time performance.
Enhances real-time monitoring capability while reducing data transmission and computing requirements, enabling effective detection of waterlogging changes.
Smart Images
Figure CN118747868B_ABST
Abstract
Description
Technical Field
[0001] The invention belongs to the technical field of urban waterlogging monitoring, and the technical application field involves urban waterlogging monitoring, drainage and waterlogging prevention, and emergency management, and specifically relates to a method for setting the time step of monitoring points for waterlogging depth video monitoring. Background Art
[0002] Monitoring the depth of urban waterlogging can provide a reference for early warning and forecasting, disaster assessment, flood control and drainage, emergency rescue, etc., and help alleviate and reduce the damage caused by urban floods.
[0003] Water level sensors are currently the main means of monitoring urban waterlogging. However, due to the high cost of building water level sensors and the limited number of monitoring sites in cities, it is difficult to obtain large-scale and detailed water depth in the city. Therefore, it is urgent to introduce new monitoring methods. Video surveillance cameras are spread all over the city, monitoring urban dynamics in real time in the form of video images. The data generated is typical video big data. The various ubiquitous reference objects in the video images are like water rulers that can be used to "measure" the depth of water. Video monitoring of water depth can use the existing video surveillance in the city as a "sensing device" for water accumulation, and use the reference objects in the video images as monitoring points to monitor the depth of water accumulation in real time. It has the characteristics of wide spatial coverage and high density, and is an important means to obtain large-scale and detailed water depth in the city.
[0004] Setting the time step of the monitoring point (i.e. the time interval between two monitoring moments before and after the monitoring point) is an essential step in monitoring the depth of waterlogging. In the prior art, a static time step is usually used, and a fixed time step is uniformly set for all the monitoring points of the depth of waterlogging.
[0005] However, water depth video monitoring requires real-time transmission of video data to the processing center. In addition, there are a large number of reference objects in the video image. If a fixed and short time step is used for all monitoring points (reference objects) in the city, it will generate a huge amount of monitoring video transmission and calculation, which will put great pressure on network transmission and computing resources; if a fixed and long time step is used, it will be difficult to effectively monitor the change process of water depth, and the real-time performance is insufficient, resulting in the failure to detect dangerous situations in time. Therefore, in this real-time demand scenario, setting a reasonable time step is very important for using video big data to monitor water depth. Summary of the invention
[0006] In view of the shortcomings of the prior art, the present invention proposes a method for setting the time step of monitoring points for video monitoring of water depth, which dynamically divides the monitoring points in the video image into representative monitoring points and non-representative monitoring points, and sets different time steps for different types of monitoring points, so as to solve the problems of insufficient real-time performance or excessive calculation caused by fixed time steps in the prior art.
[0007] A method for setting the time step of a monitoring point for water depth video monitoring specifically comprises the following steps:
[0008] Step 1: Reference object detection and reference object information setting
[0009] Select specific reference objects with fixed spatial positions and known heights (such as crash barrels, isolation piers, etc.) as water depth reference objects. Construct a water depth reference object detection model. Use the water depth reference object detection model to detect water depth reference objects (hereinafter referred to as reference objects) from existing video surveillance images. Select a corresponding video surveillance device for each reference object. For reference objects within the monitoring range of multiple devices, select the device with the highest image quality as the corresponding video surveillance device for the reference object. Use spatial positioning devices such as handheld GPS locators to determine the spatial position information of each reference object.
[0010] Step 2: Specify the maximum time step and minimum time step parameters
[0011] Take each reference object in step 1 as a water depth monitoring point p i , define the monitoring point set P = {p1,p2,...,p n}, where n is the number of all monitoring points, i = 1, 2, ..., n. The maximum time step and minimum time step of each monitoring point are uniformly specified as Δt max and Δt min , specify Δt min The corresponding average water depth d and average water depth change rate r.
[0012] Step 3: Initialize the water depth time series
[0013] For each monitoring point p i Define a time series Denotes monitoring point p i At monitoring time t j The water depth monitoring value of j = 1, 2, ..., w, w is the time series D i The number of monitoring moments before the start of water depth monitoring, D i The data in is empty. Assume that the rainfall starts at t0. After the rainfall starts, i At t0+Δtmax and t0+2Δt max The water depth at the moment is monitored (i.e., the water depth at the monitoring point is estimated from the corresponding video surveillance image), and the water depth monitoring value is added to D i ,but
[0014] Step 4: Urban surface gridding and grid classification
[0015] The urban surface is discretized into a series of grids. The monitoring points falling into each grid are determined according to the spatial location information of the monitoring points. According to the number of monitoring points in the grid, the grids are divided into four categories: grids with no monitoring points, grids containing 1 monitoring point, grids containing 2 monitoring points, and grids containing 3 or more monitoring points.
[0016] Step 5: Dynamic division of representative and non-representative monitoring points
[0017] For grids without monitoring points, there is no need to determine representative monitoring points and non-representative monitoring points.
[0018] For a grid containing one monitoring point, this monitoring point is directly used as the representative monitoring point of the grid.
[0019] For a grid containing two monitoring points, the monitoring point with the larger water depth monitoring value at the latest monitoring time (i.e. the monitoring time closest to the current time) is taken as the representative monitoring point, and the other monitoring point is taken as the non-representative monitoring point.
[0020] For grids containing three or more monitoring points, the latest monitoring time t cur and time t cur -△t max The water accumulation depth monitoring value is used as the feature vector of the monitoring point, and the monitoring points in the grid are clustered. The monitoring point corresponding to the cluster center and the monitoring point with the largest water accumulation depth monitoring value at the latest monitoring moment are taken as representative monitoring points of the grid, and the remaining monitoring points are taken as non-representative monitoring points.
[0021] Step 6: Set the time step of representative monitoring points
[0022] According to the average water depth and average water depth change rate of the representative monitoring points, the monitoring time step is set as follows:
[0023] s6.1. Traverse all grids and obtain the representative monitoring point p in each grid z The corresponding time series D z and the latest monitoring time t cur , z=1,2,…m, m is the number of representative monitoring points in the current grid.
[0024] s6.2. Calculate the representative monitoring point p z Average water depth and the average water depth change rate
[0025]
[0026] Where c is the time series D z In the time period [t cur -△t max ,t cur ], t s 'For D z In the time period [t cur -△t max ,t cur ] is the sth monitoring moment, t1' = t cur -△t max , t' c =t cur , For p z At monitoring time t s 'The water depth value.
[0027] s6.3、Set up representative monitoring pointsp z The time step △t z :
[0028]
[0029] in, For p z The initial value of the time step.
[0030] Step 7: Set the time step of non-representative monitoring points
[0031] Traverse all grids and obtain the non-representative monitoring points p in each grid f , f=1,2,…a, a is the number of non-representative monitoring points in the current grid, and p f The time step is set to △t max .
[0032] Step 8: Monitor the water depth at representative monitoring points according to the time step
[0033] The water depth is monitored at representative monitoring points according to the time step. The specific method is as follows:
[0034] s8.1. Traverse all grids and obtain the representative monitoring point p in each grid z The corresponding time series D z , latest monitoring time tcur and the dynamic time step △t set in step 6 z ;
[0035] s8.2, in the period [t cur ,t cur +△t max ], according to the time step △t z P z Monitor the depth of water accumulation and add the acquired water accumulation depth monitoring data to the time series D z Inside;
[0036] s8.3, judgement D z Does it contain p z At time t cur +△t max If the water depth monitoring data is not included, then the p z At time t cur +△t max The depth of water accumulation is monitored and the p z At time t cur +△t max The water depth monitoring data of D z ;
[0037] Step 9: Monitor the water depth at non-representative monitoring points according to the time step
[0038] Traverse all grids and obtain the non-representative monitoring points p in each grid f The corresponding time series D f and the latest monitoring time t cur . f At time t cur +△t max The water depth is monitored and the time t cur +△t max The water depth monitoring data of D f middle.
[0039] Step 10: Repeat steps 5 to 9 until the specified end time is reached.
[0040] The present invention has the following beneficial effects:
[0041] The monitoring points are dynamically divided into representative monitoring points and non-representative monitoring points. The time step of the representative monitoring points is dynamically set according to the water accumulation depth and the water accumulation depth change rate of the representative monitoring points, so that the water accumulation depth monitoring is more targeted and reasonable; at the same time, it can greatly reduce the requirements for data transmission and computing power, which is conducive to the promotion and application of water accumulation depth video monitoring. BRIEF DESCRIPTION OF THE DRAWINGS
[0042] Figure 1 Flowchart of method for setting time steps for monitoring points;
[0043] Figure 2 It is a schematic diagram of the dynamic changes of representative monitoring points in the grid;
[0044] Figure 3 Schematic diagram of the dynamic changes in the time step of representative monitoring points. DETAILED DESCRIPTION
[0045] The present invention will be further explained below with reference to the accompanying drawings;
[0046] A method for setting the time step of monitoring points for water depth video monitoring, such as Figure 1 As shown, the specific steps include:
[0047] Step 1: Reference object detection and reference object information setting
[0048] s1.1. Select a specific type of reference object with a fixed spatial position and known height (such as a crash bucket, isolation pier, etc.) as a reference object for the depth of accumulated water.
[0049] s1.2. Construct a water depth reference object detection model. The specific method is as follows:
[0050] s1.2.1. Collect video images of urban flooding that include water depth reference objects to form a water depth reference object image set.
[0051] s1.2.2. Use the Labelme tool to label the water depth reference objects in the water depth reference object image set to form a water depth reference object labeling dataset, and convert the labeled dataset into the data format required by the Single-shot MultiboxDetector (SSD) model.
[0052] s1.2.3. Use the data after the data format conversion to train the SSD model, so that the trained SSD model can automatically identify the water depth reference object from the video image, and use the trained SSD model as the water depth reference object detection model.
[0053] s1.3. Detect the water depth reference object (hereinafter referred to as reference object) from the existing video surveillance image using the water depth reference object detection model.
[0054] s1.4. Select a corresponding video surveillance device for each reference object: If a reference object appears in multiple video surveillance images, select the device with the highest image quality, such as resolution, as the corresponding video surveillance device for the reference object.
[0055] s1.5. Use spatial positioning devices such as handheld GPS locators to determine the spatial location information of each reference object.
[0056] Step 2: Specify the maximum time step and minimum time step parameters
[0057] Take each reference object in step 1 as a water depth monitoring point p i , define the monitoring point set P = {p1,p2,...,p n}, where n is the number of all monitoring points, i = 1, 2, ..., n. The maximum time step and minimum time step of each monitoring point are uniformly specified as Δt max and Δt min , specify Δt min The corresponding average water depth d and average water depth change rate r.
[0058] Step 3: Initialize the water depth time series
[0059] For each monitoring point p i Define a time series Denotes monitoring point p i At monitoring time t j The water depth monitoring value of j = 1, 2, ..., w, w is the time series D i The number of monitoring moments before the start of water depth monitoring, D i The data in is empty. Assuming that the precipitation starts at t0, after the precipitation starts, first i At t0+Δt max and t0+2Δt max The water depth at the moment t0+Δt max and t0+2Δt max The water depth monitoring value at the moment is added to D i ,but The present invention adopts the waterlogging depth estimation method in the document "Automatic estimation of urban waterlogging depths from video images based on ubiquitous reference object" to monitor the waterlogging depth at the monitoring point. Firstly, the reference object detection model is used to detect the "same reference object" in the waterlogging state and the non-waterlogging state from the corresponding video surveillance image of the reference object. Then, the waterlogging depth is estimated by using the height difference of the "same reference object" in the two states (in the waterlogging state, part of the reference object will be below the water surface, resulting in that only the part of the reference object above the water surface will be detected).
[0060] Step 4: Urban surface gridding and grid classification
[0061] s4.1. Specify the spatial extent and grid size of the city, and use the create fishnet tool in ArcMap to generate the initial grid set data. Then use the clipping function in ArcMap to clip the initial grid set data according to the city administrative district boundary data, so as to discretize the urban area into a series of grids.
[0062] s4.2. Determine the monitoring points that fall into each grid based on the spatial location information of the monitoring points. According to the number of monitoring points in the grid, divide the grid into four categories: grid with no monitoring points, grid with 1 monitoring point, grid with 2 monitoring points, and grid with 3 or more monitoring points.
[0063] Step 5: Dynamic division of representative and non-representative monitoring points
[0064] For grids without monitoring points, it is not necessary to determine representative monitoring points and non-representative monitoring points.
[0065] For a grid containing one monitoring point, this monitoring point is directly used as the representative monitoring point of the grid.
[0066] For a grid containing two monitoring points, the water depth monitoring values corresponding to the latest monitoring moments are read from the time series corresponding to the two monitoring points respectively. The monitoring point with the larger water depth monitoring value at the latest monitoring moment is taken as the representative monitoring point, and the other monitoring point is taken as the non-representative monitoring point.
[0067] For grids containing three or more monitoring points, the monitoring points in the grid are clustered, and the monitoring point corresponding to the cluster center and the monitoring point with the largest water accumulation depth value are selected as representative monitoring points of the grid, and the remaining monitoring points are regarded as non-representative monitoring points.
[0068] In one embodiment, for a grid containing three or more monitoring points, representative monitoring points are selected by k-medoids, and the specific method is as follows:
[0069] s5.1、Real-time reading of each monitoring point p in the grid v The corresponding time series D v , where v = 1, 2, ... M, M is the number of all monitoring points in the grid; from D v Get the latest monitoring time t cur Corresponding water depth monitoring value and time t cur -△t max Corresponding water depth monitoring value These two water depth values are used as monitoring points p vThe feature vector of .
[0070] s5.2. Randomly select k monitoring points from the grid monitoring points as the initial cluster centers of k clusters, where k is the number of clusters, determined by the silhouette coefficient method, and k<M.
[0071] s5.3. Calculate the Euclidean distance between the feature vector of each monitoring point in the grid and the feature vectors of the k cluster centers respectively, and assign each monitoring point in the grid to the cluster where the cluster center with the closest Euclidean distance to its feature vector is located, forming k clusters.
[0072] s5.4. For each cluster, calculate the sum of the Euclidean distances between the feature vector of each monitoring point in the cluster and the feature vectors of all other monitoring points in the cluster, and take the monitoring point with the smallest sum of the Euclidean distances to the feature vectors of other monitoring points as the cluster center.
[0073] s5.5. Repeat s5.3 and s5.4 until the cluster center no longer changes or the preset number of iterations is reached.
[0074] s5.6. The monitoring points corresponding to the k cluster centers obtained by clustering and the monitoring point with the largest water depth monitoring value at the latest monitoring moment are taken as representative monitoring points of the grid, and the remaining monitoring points are taken as non-representative monitoring points.
[0075] like Figure 2 As shown, at different times, the representative monitoring points and non-representative monitoring points in the same grid may change.
[0076] Step 6: Set the time step of representative monitoring points
[0077] According to the average water depth and average water depth change rate of the representative monitoring points, the monitoring time step is set as follows:
[0078] s6.1. Traverse all grids and obtain the representative monitoring point p in each grid z The corresponding time series D z and the latest monitoring time t cur , z=1,2,…m, m is the number of representative monitoring points in the current grid.
[0079] s6.2. Calculate the representative monitoring point p z Average water depth and the average water depth change rate r z :
[0080]
[0081] Where c is the time series D z In the time period [t cur -△tmax ,t cur ], t s 'For D z In the time period [t cur -△t max ,t cur ] is the sth monitoring moment, t1' = t cur -△t max , t' c =t cur , For p z At monitoring time t s 'Water depth monitoring value.
[0082] s6.3、Set up representative monitoring pointsp z The time step △t z :
[0083]
[0084] in, For p z The initial value of the time step.
[0085] like Figure 3 As shown, since the time step is set according to the average water depth and the rate of change of the average water depth at the representative monitoring points, the time step of the representative monitoring points may change.
[0086] Step 7: Set the time step of non-representative monitoring points
[0087] Traverse all grids and obtain the non-representative monitoring points p in each grid f , f=1,2,…a, a is the number of non-representative monitoring points in the current grid, and p f The time step is set to △t max .
[0088] Step 8: Monitor the water depth at representative monitoring points according to the time step
[0089] The water depth is monitored at representative monitoring points according to the time step. The specific method is as follows:
[0090] s8.1. Traverse all grids and obtain the representative monitoring point p in each grid z The corresponding time series D z , latest monitoring time t cur and the dynamic time step △t set in step 6 z ;
[0091] s8.2, in the period [t cur ,tcur +△t max ], according to the time step △t z P z Monitor the depth of water accumulation (i.e. from t cur +△t z Start by checking p z At time t cur +u△t z The depth of water accumulation is monitored, u=1,2,...l, l is p z In the period [t cur ,t cur +△t max ] within △t z The maximum number of times waterlogging is monitored in a time step, t cur +l△t z ≤t cur +△t max ), and add the water depth monitoring data to the time series D z Inside;
[0092] s8.3, judgement D z Does it contain p z At time t cur +△t max If the water depth monitoring data is not included, then the p z At time t cur +△t max The water depth is monitored and the time t cur +△t max The water depth monitoring data of D z ;
[0093] Step 9: Monitor the water depth at non-representative monitoring points according to the time step
[0094] Traverse all grids and obtain the non-representative monitoring points p in each grid f The corresponding time series D f and the latest monitoring time t cur . f At time t cur +△t max The water depth is monitored and the time t cur +△t max The water depth monitoring data of D f middle.
[0095] Step 10: Repeat steps 5 to 9 until the specified end time is reached.
Claims
1. A method for setting the time step of monitoring points for water depth video monitoring, characterized in that: The specific steps include: Step 1: Reference object detection and reference object information setting Detecting water depth reference objects from the video surveillance image, and determining the spatial position information of each reference object using a spatial positioning device; Step 2: Specify the maximum and minimum time steps Take each reference object in step 1 as a water depth monitoring point p i , i=1,2,...,n, n is the number of monitoring points; the maximum time step and the minimum time step of each monitoring point are uniformly designated as Δt max and Δt min , specify Δt min The corresponding average water depth d and average water depth change rate r; Step 3: Initialize the water depth time series For each monitoring point p i Define a time series D i , used to store the water depth monitoring values at different monitoring times; after the rainfall begins, the monitoring point p i At t0+Δt max and t0+2Δt max The water depth at each moment is monitored and the water depth monitoring value is recorded. Add to the time series, t0 is the time when rainfall starts; Step 4: Urban surface gridding and grid classification Discretize the urban surface into a series of grids and classify the grids according to the number of monitoring points in the grids; Step 5: Dynamic division of representative and non-representative monitoring points For grids without monitoring points, it is not necessary to determine representative and non-representative monitoring points; For a grid containing one monitoring point, this monitoring point is directly used as the representative monitoring point of the grid; For a grid containing two monitoring points, the latest monitoring time t cur The monitoring point with the larger water depth monitoring value is taken as the representative monitoring point, and the other monitoring point is taken as the non-representative monitoring point; For grids containing three or more monitoring points, the latest monitoring time t cur and time t cur -△t max The water depth monitoring value is used as the feature vector of the monitoring point, and the monitoring points in the grid are clustered. The monitoring point corresponding to the cluster center and the monitoring point with the largest water depth monitoring value at the latest monitoring time are taken as the representative monitoring points of the grid, and the remaining monitoring points are taken as non-representative monitoring points. Step 6: Set the time step of representative monitoring points According to the average water depth and average water depth change rate of the representative monitoring points, the monitoring time step is set as follows: s6.
1. Traverse all grids and obtain the representative monitoring point p in each grid z The corresponding time series D z and the latest monitoring time t cur , z=1,2,…m, m is the number of representative monitoring points in the current grid; s6.
2. Calculate the representative monitoring point p z In the time period [t cur -△t max ,t cur The average depth of water in and the average water depth change rate s6.3、Set up representative monitoring pointsp z The time step △t z : in, For p z The initial value of the time step; Step 7: Set the time step of non-representative monitoring points The non-representative monitoring points p in each grid f The time step is set to △t max , f=1,2,…a, a is the number of non-representative monitoring points in the current grid; Step 8: Monitor the water depth at representative monitoring points according to the time step In the period [t cur ,t cur +△t max ], according to the time step △t z For representative monitoring point p z Monitor the depth of water accumulation and add the acquired water accumulation depth monitoring data to the time series D z If D z does not contain p z At time t cur +△t max The water depth monitoring data of p z At time t cur +△t max Monitor the depth of waterlogging and add the waterlogging depth monitoring data to D z ; Step 9: Monitor the water depth at non-representative monitoring points according to the time step For non-representative monitoring points p f At time t cur +△t max Monitor the depth of waterlogging and add the waterlogging depth monitoring data to D f middle; Step 10: Repeat steps 5 to 9 until the specified end time is reached.
2. A method for setting the time step of monitoring points for water depth video monitoring according to claim 1, characterized in that: A specific reference object with a fixed spatial position and a known height is selected as a water accumulation depth reference object; a water accumulation depth reference object detection model is constructed; and the reference object is detected from the video surveillance image using the water accumulation depth reference object detection model; For reference objects that appear within the monitoring range of multiple devices at the same time, the device with the highest image quality is selected as the corresponding video monitoring device for the reference object; a handheld GPS locator is used to determine the spatial location information of each reference object.
3. A method for setting the time step of monitoring points for water depth video monitoring as claimed in claim 2, characterized in that: The method for constructing a waterlogging depth reference object detection model is as follows: collect waterlogging video images containing waterlogging depth reference objects to form a waterlogging depth reference object image set; use the Labelme tool to annotate the waterlogging depth reference objects in the waterlogging depth reference object image set and perform format conversion to form a waterlogging depth reference object annotated dataset; The SSD model is trained using the labeled data set, so that the trained SSD model can automatically identify water depth reference objects from video images. The trained SSD model is the water depth reference object detection model.
4. A method for setting the time step of monitoring points for water depth video monitoring according to claim 1, characterized in that: The spatial scope and grid size of the city were specified, and the initial grid set data was generated using the create fishnet tool in ArcMap. The initial grid set data was then clipped using the clipping function in ArcMap according to the city administrative district boundary data, thereby discretizing the urban area into a series of grids. The monitoring points that fell into each grid were determined based on the spatial location information of the monitoring points, and then the grids were divided into grids with no monitoring points, grids with 1 monitoring point, grids with 2 monitoring points, and grids with 3 or more monitoring points.
5. The method for setting the time step of monitoring points for water depth video monitoring according to claim 1, characterized in that: For grids containing three or more monitoring points, the clustering algorithm used is the k-medoids algorithm.
6. A method for setting the time step of monitoring points for water depth video monitoring as claimed in claim 5, characterized in that: The method for selecting representative monitoring points for a grid containing 3 or more monitoring points is as follows: s5.1、Real-time reading of each monitoring point p in the grid v The corresponding time series D v , where v = 1, 2, ... M, M is the number of all monitoring points in the grid; from D v Get the latest monitoring time t cur Corresponding water depth monitoring value and time t cur -△t max Corresponding water depth monitoring value These two water depth values are used as monitoring points p v The eigenvector of s5.2, randomly select k monitoring points from the grid monitoring points as the initial cluster centers, k<M; s5.
3. Calculate the Euclidean distance between the feature vector of each monitoring point in the grid and the feature vectors of the k cluster centers respectively, and assign each monitoring point in the grid to the cluster where the cluster center with the closest Euclidean distance to its feature vector is located, forming k clusters; s5.
4. For each cluster, calculate the sum of the Euclidean distances between the feature vectors of each monitoring point in the cluster and the feature vectors of all other monitoring points in the cluster, and take the monitoring point with the smallest sum of the Euclidean distances to the feature vectors of other monitoring points as the cluster center; s5.5, repeat s5.3 and s5.4 until the cluster center no longer changes or the preset number of iterations is reached; s5.
6. The monitoring points corresponding to the k cluster centers obtained by clustering and the monitoring point with the largest water depth monitoring value at the latest monitoring moment are taken as representative monitoring points of the grid, and the remaining monitoring points are taken as non-representative monitoring points.
7. A method for setting the time step of monitoring points for water depth video monitoring according to claim 6, characterized in that: The number k of initial cluster centers is determined by the silhouette coefficient method.
8. The method for setting the time step of monitoring points for water depth video monitoring according to claim 1, characterized in that: Calculate the representative monitoring point p z Average water depth and the average water depth change rate Where c is the time series D z In the time period [t cur -△t max ,t cur ], t s 'For D z In the time period [t cur -△t max ,t cur ] is the sth monitoring moment, t1' = t cur -△t max , t' c =t cur , ' For p z At monitoring time t s Monitoring value of water depth.
9. A computer-readable storage medium having a computer program stored thereon, which, when executed in a computer, causes the computer to execute the method according to any one of claims 1 to 8.