Waterlogging real-time monitoring method and system based on urban electronic eye system

Abstract

The invention provides a waterlogging real-time monitoring method and system based on an urban electronic eye system, and relates to the field of fluid measurement, and the method comprises the steps: setting a plurality of camera groups, carrying out calibration, and determining camera calibration parameters; feature matching is carried out on an upstream image collected by the upstream camera at the t moment and a downstream image collected by the downstream camera at the t moment, feature matching is carried out on an upstream image collected by the upstream camera at the t + 1 moment and a downstream image collected by the downstream camera at the t + 1 moment, and waterlogging water depth distribution is determined; floating substance center identification and floating substance center matching are carried out, and three-dimensional coordinates of the floating object at the t moment and the t + 1 moment are determined according to camera calibration parameters; and according to the three-dimensional coordinate of the floater at the t moment, the three-dimensional coordinate of the floater at the t + 1 moment and the time difference between the t moment and the t + 1 moment, the waterlogging flow velocity distribution is determined, and the method has the advantage of improving the precision and efficiency of urban flood detection.

Description

Technical Field

[0001] The present invention relates to the field of fluid measurement, and particularly to a real-time monitoring method and system for urban waterlogging based on an urban electronic eye system. Background Art

[0002] Urban flood disasters mainly include two types: external floods and waterlogging. Flash floods or river floods caused by heavy rain, snowmelt, etc. are the main sources of external floods in cities; while urban waterlogging is more due to the obstruction of urban rainwater outflow, resulting in overly concentrated water accumulation on urban roads and turning streets into "rivers". The causes of both external floods and waterlogging cover multiple factors, mainly including two types of factors: natural factors and human activities.

[0003] Currently, urban flood monitoring technologies mainly include aspects such as water accumulation level, water accumulation range, and surface runoff velocity. The methods for monitoring urban flood water accumulation level mainly include: (1) direct measurement tools such as water gauges or sounding needles, which rely entirely on manual measurement of water level, with low accuracy and efficiency; (2) water level sensors such as ultrasonic water level gauges, radar water level gauges, and electronic water gauges, which are limited by the instrument layout and can only monitor the water level at a single point, and are easily affected by the external environment; (3) image processing technology and machine learning algorithms, which can currently only monitor the water level at a single point or locally, and the monitoring data highly depends on water gauges and other markers, and also requires a large amount of data to train the model, and the accuracy and efficiency are unstable in actual use. The methods for monitoring urban flood water accumulation range mainly include: (1) image processing technology, which can only identify the contour range of the water accumulation area and extract edge information; (2) deep learning models, which can only identify the proportion of the actual water accumulation area and cannot obtain contour information. The methods for monitoring urban flood surface runoff velocity mainly include: (1) sensors such as radio current meters and ADV (Acoustic Doppler Velocimetry), which can only obtain single-point information and are easily affected by the environment; (2) light refraction and light absorption, which are mainly applied to indoor monitoring, are sensitive to the external environment, and are rarely applied to outdoor flood environments; (3) image velocity measurement technologies such as PIV (Particle Image Velocimetry) and PTV (Particle Tracking Velocimetry) and the corresponding large-scale measurement technologies LSPIV and LSPTV, etc., whose accuracy is easily affected by conditions such as lighting.

[0004] Therefore, there is a need to provide a real-time monitoring method and system for urban waterlogging based on an urban electronic eye system to improve the accuracy and efficiency of urban flood detection. Summary of the Invention

[0005] The present invention provides a real-time monitoring method for urban waterlogging based on an urban electronic eye system, including: setting a plurality of camera groups, where each camera group includes an upstream camera and a downstream camera, and there is an overlapping shooting area between the shooting areas of the upstream camera and the downstream camera; for each camera group, calibrating the upstream camera and the downstream camera to determine camera calibration parameters; acquiring upstream images collected by the upstream camera at time t and time t + 1 and downstream images collected by the downstream camera at time t and time t + 1; performing feature matching on the upstream image collected by the upstream camera at time t and the downstream image collected by the downstream camera at time t, and performing feature matching on the upstream image collected by the upstream camera at time t + 1 and the downstream image collected by the downstream camera at time t + 1 to determine the waterlogging depth distribution; performing floating object centroid recognition and floating object centroid matching on the upstream image collected by the upstream camera at time t and the downstream image collected by the downstream camera at time t, and determining the three-dimensional coordinates of the floating object at time t according to the camera calibration parameters; performing floating object centroid recognition and floating object centroid matching on the upstream image collected by the upstream camera at time t + 1 and the downstream image collected by the downstream camera at time t + 1, and determining the three-dimensional coordinates of the floating object at time t + 1 according to the camera calibration parameters; determining the waterlogging flow velocity distribution according to the three-dimensional coordinates of the floating object at time t, the three-dimensional coordinates of the floating object at time t + 1, and the time difference between time t and time t + 1.

[0006] Further, calibrating the upstream camera and the downstream camera to determine camera calibration parameters includes: setting a calibration board in the overlapping shooting area of the upstream camera and the downstream camera, and acquiring images of the calibration board in different states through the upstream camera and the downstream camera; determining the internal parameter matrices of the upstream camera and the downstream camera according to the images of the calibration board in different states acquired by the upstream camera and the downstream camera; setting a plurality of ground control points in the overlapping shooting area of the upstream camera and the downstream camera, and acquiring images of the plurality of ground control points through the upstream camera and the downstream camera; determining the external parameter matrices of the upstream camera and the downstream camera according to the images of the plurality of ground control points acquired by the upstream camera and the downstream camera and the world coordinates of the plurality of ground control points, where the camera calibration parameters at least include the internal parameter matrices and the external parameter matrices of the upstream camera and the downstream camera.

[0007] Further, perform feature matching on the upstream image captured by the upstream camera at time t and the downstream image captured by the downstream camera at time t, and perform feature matching on the upstream image captured by the upstream camera at time t + 1 and the downstream image captured by the downstream camera at time t + 1 to determine the water depth distribution of the waterlogging, including: performing image preprocessing on the upstream image captured by the upstream camera at time t and the downstream image captured by the downstream camera at time t, and extracting the upstream target region image and the downstream target region image corresponding to time t; determining feature point extraction parameters according to the position parameters of the upstream camera and the downstream camera; extracting the upstream image feature points corresponding to time t from the upstream target region image corresponding to time t according to the feature point extraction parameters, extracting the downstream image feature points corresponding to time t from the downstream target region image corresponding to time t, extracting the upstream image feature points corresponding to time t + 1 from the upstream target region image corresponding to time t + 1, and extracting the downstream image feature points corresponding to time t + 1 from the downstream target region image corresponding to time t + 1; performing feature point matching and error elimination on the upstream image feature points and the downstream image feature points corresponding to time t to determine the feature points corresponding to time t; determining the three-dimensional coordinates of the feature points corresponding to time t according to the camera calibration parameters; performing feature point matching and error elimination on the upstream image feature points and the downstream image feature points corresponding to time t + 1 to determine the feature points corresponding to time t; determining the three-dimensional coordinates of the feature points corresponding to time t + 1 according to the camera calibration parameters; and determining the water depth distribution of the waterlogging according to the three-dimensional coordinates of the feature points corresponding to time t and the three-dimensional coordinates of the feature points corresponding to time t + 1.

[0008] Further, the position parameters at least include the installation height and the baseline distance; the feature point extraction parameters at least include the measurement threshold of the feature points, the number of octaves, and the number of scale levels used in each octave; extracting the upstream image feature points corresponding to time t from the upstream target region image corresponding to time t according to the feature point extraction parameters, and extracting the downstream image feature points corresponding to time t from the downstream target region image corresponding to time t includes: extracting the upstream image feature points corresponding to time t from the upstream target region image corresponding to time t through the SURF algorithm according to the feature point extraction parameters, and extracting the downstream image feature points corresponding to time t from the downstream target region image corresponding to time t.

[0009] Further, perform feature point matching and error elimination on the upstream image feature points and downstream image feature points corresponding to the t-th moment to determine the feature points corresponding to the t-th moment, including: calculating the distance between any upstream image feature point corresponding to the t-th moment and any downstream image feature point corresponding to the t-th moment; determining the matching feature point group corresponding to the t-th moment according to the distance threshold and the distance between any upstream image feature point corresponding to the t-th moment and any downstream image feature point corresponding to the t-th moment, where the matching feature point group includes an upstream image feature point and a downstream image feature point; determining the error elimination parameter according to the position parameters of the upstream camera and the downstream camera, where the error elimination parameter at least includes the maximum allowable number of trials, the confidence level, and the allowable maximum distance; performing error elimination on the matching feature point group corresponding to the t-th moment through MSAC according to the error elimination parameter to determine the feature points corresponding to the t-th moment.

[0010] Further, determine the water depth distribution of waterlogging according to the three-dimensional coordinates of the feature points corresponding to the t-th moment and the three-dimensional coordinates of the feature points corresponding to the (t + 1)-th moment, including: obtaining the water depth point set according to the three-dimensional coordinates of the feature points corresponding to the t-th moment and the three-dimensional coordinates of the feature points corresponding to the (t + 1)-th moment; performing scatter interpolation on the water depth point set to obtain surface data; performing median filtering and PCHIP interpolation on the surface data to determine the water depth distribution of waterlogging.

[0011] Further, perform floating object centroid recognition and floating object centroid matching on the upstream image collected by the upstream camera at the t-th moment and the downstream image collected by the downstream camera at the t-th moment, and determine the three-dimensional coordinates of the floating object at the t-th moment according to the camera calibration parameters, including: calculating the gray threshold of the upstream target area image corresponding to the t-th moment using the OTSU algorithm, binarizing the upstream target area image corresponding to the t-th moment to generate the upstream target area binarized image corresponding to the t-th moment, and determining the centroid coordinates of the connected area formed by the floating object in the upstream target area binarized image corresponding to the t-th moment; converting the centroid coordinates of the connected area formed by the floating object in the upstream target area binarized image corresponding to the t-th moment into the centroid coordinates in the downstream camera coordinate system according to the homography matrix obtained by feature matching; calculating the gray threshold of the downstream target area image corresponding to the t-th moment using the OTSU algorithm, binarizing the downstream target area image corresponding to the t-th moment to generate the downstream target area binarized image corresponding to the t-th moment, and determining the centroid coordinates of the connected area formed by the floating object in the downstream target area binarized image corresponding to the t-th moment; performing floating object centroid matching according to the centroid coordinates in the downstream camera coordinate system corresponding to the t-th moment and the centroid coordinates of the connected area formed by the floating object in the downstream target area binarized image corresponding to the t-th moment to determine the matching floating object centroid corresponding to the t-th moment; determining the three-dimensional coordinates of the matching floating object centroid at the t-th moment according to the camera calibration parameters.

[0012] Furthermore, the waterlogging flow velocity is calculated based on the following formula: , where V is the waterlogging flow velocity corresponding to the centroid of the floating object, is the velocity component of the waterlogging flow velocity corresponding to the centroid of the floating object in the x - direction, is the x - axis coordinate of the centroid of the floating object at time t + 1, is the x - axis coordinate of the centroid of the floating object at time t, is the velocity component of the waterlogging flow velocity corresponding to the centroid of the floating object in the y - direction, is the y - axis coordinate of the centroid of the floating object at time t + 1, is the y - axis coordinate of the centroid of the floating object at time t, is the velocity component of the waterlogging flow velocity corresponding to the centroid of the floating object in the z - direction, is the z - axis coordinate of the centroid of the floating object at time t + 1, is the z - axis coordinate of the centroid of the floating object at time t, is the time difference between time points t and t + 1.

[0013] Furthermore, the method further includes: superimposing the waterlogging water depth distribution and the waterlogging flow velocity distribution to form an integrated monitoring result of water depth and flow velocity.

[0014] The present invention provides a real-time monitoring system for urban waterlogging based on an urban electronic eye system, and applies the above-mentioned real-time monitoring method for urban waterlogging based on an urban electronic eye system, including: an image acquisition module, including a plurality of camera groups, wherein each camera group includes an upstream camera and a downstream camera, and there is an overlap in the shooting areas of the upstream camera and the downstream camera; a camera calibration module, configured to calibrate the upstream camera and the downstream camera for each camera group to determine camera calibration parameters; an image acquisition module, configured to acquire an upstream image acquired by the upstream camera at time t and time t+1 and a downstream image acquired by the downstream camera at time t and time t+1; a waterlogging analysis module, configured to perform feature matching on the upstream image acquired by the upstream camera at time t and the downstream image acquired by the downstream camera at time t, and perform feature matching on the upstream image acquired by the upstream camera at time t+1 and the downstream image acquired by the downstream camera at time t+1 to determine the waterlogging water depth distribution; perform floating object centroid recognition and floating object centroid matching on the upstream image acquired by the upstream camera at time t and the downstream image acquired by the downstream camera at time t, and determine the three-dimensional coordinates of the floating object at time t according to the camera calibration parameters; perform floating object centroid recognition and floating object centroid matching on the upstream image acquired by the upstream camera at time t+1 and the downstream image acquired by the downstream camera at time t+1, and determine the three-dimensional coordinates of the floating object at time t+1 according to the camera calibration parameters; and determine the waterlogging flow velocity distribution according to the three-dimensional coordinates of the floating object at time t, the three-dimensional coordinates of the floating object at time t+1, and the time difference between time t and time t+1.

[0015] Compared with the prior art, the real-time monitoring method and system for urban waterlogging provided by the present invention at least have the following beneficial effects: It can carry out real-time monitoring of the water depth, flow velocity and three-dimensional water surface in urban flood areas. It can not only solve the limitations of single-point monitoring of water level or flow velocity in the past, but also realize the simultaneous monitoring of water level and flow velocity. At the same time, it can realize three-dimensional monitoring of the water surface flow velocity. For the monitoring of water flow movement in urban flood areas, it improves the efficiency of measuring relevant hydraulic parameters, enriches the monitoring information of hydraulic parameters, and greatly reduces the monitoring risk and requirements. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] This specification will be further described in the form of exemplary embodiments, and these exemplary embodiments will be described in detail through the drawings. These embodiments are not restrictive. In these embodiments, the same numbers represent the same structures, wherein: Figure 1 is a schematic flowchart of a real-time monitoring method for urban waterlogging based on an urban electronic eye system shown in some embodiments of this specification; Figure 2It is a schematic diagram of the relative position relationship between the upstream camera and the downstream camera shown in some embodiments of this specification; Figure 3 It is a schematic diagram of the result after extracting feature points and removing incorrect matching points shown in some embodiments of this specification; Figure 4 It is a schematic diagram of the result after matching the centroid of floating substances shown in some embodiments of this specification; Figure 5 It is a schematic diagram of the water depth distribution shown in some embodiments of this specification; Figure 6 It is a schematic diagram of the result of the water depth-coupled three-dimensional flow velocity distribution shown in some embodiments of this specification; Figure 7 It is a schematic diagram of the modules of the urban waterlogging real-time monitoring system based on the urban electronic eye system shown in some embodiments of this specification. Detailed implementation manners

[0017] To more clearly illustrate the technical solutions of the embodiments of this specification, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following-described drawings are only some examples or embodiments of this specification. For those of ordinary skill in the art, without creative efforts, this specification can also be applied to other similar scenarios based on these drawings. Unless obvious from the language context or otherwise stated, the same reference numerals in the drawings represent the same structure or operation.

[0018] Figure 1 It is a schematic flowchart of the urban waterlogging real-time monitoring method based on the urban electronic eye system shown in some embodiments of this specification. As Figure 1 shown, the urban waterlogging real-time monitoring method based on the urban electronic eye system may include the following steps.

[0019] Step 110, set multiple camera groups, where each camera group includes an upstream camera and a downstream camera, and there is an overlapping shooting area between the upstream camera and the downstream camera.

[0020] Figure 2 It is a schematic diagram of the relative position relationship between the upstream camera and the downstream camera shown in some embodiments of this specification. As Figure 2 shown, during urban waterlogging, the accumulated water flows from the upstream camera to the downstream camera.

[0021] Step 120, for each camera group, calibrate the upstream camera and the downstream camera to determine the camera calibration parameters.

[0022] In some embodiments, step 120 specifically includes: Set a calibration board in the overlapping shooting area of the upstream camera and the downstream camera, and obtain images of the calibration board in different states through the upstream camera and the downstream camera; Determine the internal parameter matrices of the upstream camera and the downstream camera according to the images of the calibration board in different states obtained by the upstream camera and the downstream camera; Set multiple ground control points in the overlapping shooting area of the upstream camera and the downstream camera, and obtain images of the multiple ground control points through the upstream camera and the downstream camera; Determine the external parameter matrices of the upstream camera and the downstream camera according to the images of the multiple ground control points obtained by the upstream camera and the downstream camera and the world coordinates of the multiple ground control points. Among them, the camera calibration parameters at least include the internal parameter matrices and the external parameter matrices of the upstream camera and the downstream camera.

[0023] Specifically, the calibration board can be a checkerboard calibration board, and the size of the calibration board can be determined according to actual needs. For example, the size of the calibration board can be 500×600mm. The positions of the calibration boards in different states are different. The number of states can be determined according to actual needs. For example, the number of states can be 10 to 20.

[0024] The images of the calibration board in different states obtained by the upstream camera and the downstream camera can be imported into the calibration toolbox in MATLAB for calibration to determine the internal parameter matrices of the upstream camera and the downstream camera.

[0025] A solid circle shape with prominent color can be used as the ground control point. The number of ground control points can be determined according to actual needs. For example, the number of ground control points can be 10 to 15. The center pixel coordinates are extracted from the corrected picture by the circular Hough transform method, and the internal parameter matrices and the external parameter matrices of the upstream camera and the downstream camera are solved by using the direct linear transformation and the least squares method.

[0026] Step 130, obtain the upstream image collected by the upstream camera at time t and t+1 and the downstream image collected by the downstream camera at time t and t+1.

[0027] Step 140, perform feature matching on the upstream image collected by the upstream camera at time t and the downstream image collected by the downstream camera at time t, and perform feature matching on the upstream image collected by the upstream camera at time t+1 and the downstream image collected by the downstream camera at time t+1 to determine the distribution of the water depth of waterlogging.

[0028] In some embodiments, step 140 specifically includes: Perform image preprocessing on the upstream image collected by the upstream camera at time t and the downstream image collected by the downstream camera at time t, and extract the upstream target area image and the downstream target area image corresponding to time t; Determine the feature point extraction parameters according to the position parameters of the upstream camera and the downstream camera; According to the feature point extraction parameters, extract the upstream image feature points corresponding to time t from the upstream target area image corresponding to time t, extract the downstream image feature points corresponding to time t from the downstream target area image corresponding to time t, extract the upstream image feature points corresponding to time t + 1 from the upstream target area image corresponding to time t + 1, and extract the downstream image feature points corresponding to time t + 1 from the downstream target area image corresponding to time t + 1; Perform feature point matching and error elimination on the upstream image feature points and downstream image feature points corresponding to time t to determine the feature points corresponding to time t; Determine the three-dimensional coordinates of the feature points corresponding to time t according to the camera calibration parameters; Perform feature point matching and error elimination on the upstream image feature points and downstream image feature points corresponding to time t + 1 to determine the feature points corresponding to time t + 1; Determine the three-dimensional coordinates of the feature points corresponding to time t + 1 according to the camera calibration parameters; Determine the waterlogging water depth distribution according to the three-dimensional coordinates of the feature points corresponding to time t and the three-dimensional coordinates of the feature points corresponding to time t + 1.

[0029] Specifically, in the images captured by the upstream camera and the downstream camera, it is inevitable that the water flow (target area) and the two banks (background area) exist at the same time. It is necessary to only obtain the feature points of the water flow (target area) and not obtain the feature points of the two banks (background area) additionally to increase the calculation load and disrupt information extraction. Therefore, the poly2mask function is used in the image processing stage to create a binary mask to separate the target area from the background area. The specific method is as follows: S11. Convert the original RGB image into a grayscale image, and the grayscale value G can be calculated by the following formula: , where R, G, and B respectively correspond to the values of the red, green, and blue channels.

[0030] S12. Obtain the image size, generally 1080*1920, and customize the polygon vertex coordinates of a specific area of the image, noting that they should be within the image size.

[0031] S13. Use the poly2mask function to create a binary mask. The pixel value M in the mask is 1 for the inside of the polygon and 0 for the outside. The pixel value M in the mask can be defined by the following conditions: , S14. Multiply the grayscale image with the mask using element-wise multiplication to retain the region inside the polygon and remove the external region. The image Imasked after mask processing can be expressed as: , where I is the original grayscale image and M is the binary mask.

[0032] In some embodiments, according to the feature point extraction parameters, extract the upstream image feature points corresponding to time t from the upstream target region image corresponding to time t, and extract the downstream image feature points corresponding to time t from the downstream target region image corresponding to time t, including: According to the feature point extraction parameters, extract the upstream image feature points corresponding to time t from the upstream target region image corresponding to time t and the downstream image feature points corresponding to time t from the downstream target region image corresponding to time t through the SURF algorithm. Specifically, extract the upstream image feature point descriptor corresponding to time t through the SURF (Speeded Up Robust Features) algorithm and the downstream image feature point descriptor corresponding to time t .

[0033] In some embodiments, the position parameters at least include the installation height and the baseline distance. Among them, the installation height can be the installation heights of the upstream camera and the downstream camera, and the baseline distance can be the distance between the optical centers of the upstream camera and the downstream camera. The feature point extraction parameters at least include the measurement threshold of the feature points, the number of octaves, and the number of scale levels used at each octave.

[0034] Specifically, the measurement threshold of the feature points determines the significance of the feature points to be detected. When the installation heights of the upstream camera and the downstream camera increase or the baseline distance between the upstream camera and the downstream camera increases, the details in the captured images will decrease. Therefore, it is necessary to consider reducing this value to ensure that enough significant feature points can be detected. Only as an example, when the baseline distance increases from 2.2 m to 4.4 m, the measurement threshold of the feature points decreases from 40 to 5.

[0035] The number of octaves refers to the number of pyramid levels used when constructing the multi-scale representation of an image. In feature point extraction algorithms, a Gaussian pyramid or a Difference of Gaussian (DoG) pyramid is usually constructed to represent the features of an image at different scales. Each octave represents a scale range of the image. By detecting feature points at different octaves, stable feature points at different scales can be obtained. The choice of the number of octaves depends on the resolution of the image and the requirements of the feature point detection algorithm. Generally speaking, the higher the resolution of the image, the more octaves are required to capture more feature points at different scales. When the mounting height of the camera increases or the baseline distance between two cameras increases, the features in the image may exhibit different scale changes, and the number of octaves needs to be adjusted.

[0036] The number of scale levels used in each octave refers to the number of scale levels within each octave for feature point detection. Scale levels are obtained by changing the standard deviation of the Gaussian filter. Different scale levels correspond to different degrees of image smoothing and detail features. In each octave, the algorithm performs feature point detection at multiple scale levels to find points with significant characteristics at different scales. The choice of the number of scale levels affects the stability and robustness of the feature points. If the number of scale levels is too small, it may lead to insufficiently significant changes in feature points at different scales, affecting the accuracy of the algorithm; if the number of scale levels is too large, it will increase the computational complexity and reduce the efficiency of the algorithm. When the mounting height of the camera increases or the baseline distance between two cameras increases, the features in the image may exhibit different scale changes. In the case where the mounting height of the upstream camera and the downstream camera increases or the baseline distance between the upstream camera and the downstream camera increases, adjust the number of scale levels used in each octave as the image resolution changes. For example, if the image clarity decreases, reduce this value to avoid excessive dispersion and redundancy of feature points.

[0037] By way of example only, the feature extraction parameter determination model can determine the feature point extraction parameters according to the position parameters of the upstream camera and the downstream camera. Among them, the feature extraction parameter determination model can be a convolutional neural network model.

[0038] The method of extracting the upstream image feature points corresponding to time t from the upstream target region image corresponding to time t + 1 and the downstream image feature points corresponding to time t + 1 from the downstream target region image corresponding to time t + 1 is the same as the method of extracting the upstream image feature points corresponding to time t from the upstream target region image corresponding to time t and the downstream image feature points corresponding to time t from the downstream target region image corresponding to time t, and will not be elaborated here.

[0039] In some embodiments, feature point matching and error elimination are performed on the upstream image feature points and the downstream image feature points corresponding to time t to determine the feature points corresponding to time t, including: Calculate the distance between the upstream image feature points corresponding to any moment t and the downstream image feature points corresponding to any moment t. According to the distance threshold and the distance between the upstream image feature points corresponding to any moment t and the downstream image feature points corresponding to any moment t, determine the matching feature point group corresponding to moment t, where the matching feature point group includes an upstream image feature point and a downstream image feature point. According to the position parameters of the upstream camera and the downstream camera, determine the error rejection parameters, where the error rejection parameters at least include the maximum allowable number of trials, the confidence level, and the allowable maximum distance. According to the error rejection parameters, perform error rejection on the matching feature point group corresponding to moment t through MSAC (M-Estimator Sample Consensus) to determine the feature points corresponding to moment t.

[0040] Specifically, use the distance metric of local features (Euclidean distance) to determine the matching, which can be expressed as: , Set a fixed threshold to judge the maximum distance of the matching. When the distance is less than this value, it is judged as a valid matching feature point, where is the distance between the upstream image feature point corresponding to moment t and the downstream image feature point corresponding to moment t, is the descriptor of the upstream image feature point corresponding to moment t, is the descriptor of the downstream image feature point corresponding to moment t.

[0041] Let the upstream image feature point corresponding to moment t and the downstream image feature point corresponding to moment t be , where and are the upstream image feature point and the downstream image feature point corresponding to moment t respectively, and perform a geometric transformation, which is expressed as: , The reprojection error is: , where is the reprojection error, and H is a 3×3 homography matrix containing transformation information.

[0042] Set the distance threshold , and screen out the feature points that meet the conditions: if then keep.

[0043] By determining the error rejection parameters according to the position parameters of the upstream camera and the downstream camera, it is hoped to obtain a sufficient number of accurate feature point matching results under appropriate computing power.

[0044] In each trial, the algorithm randomly selects a part of the matching points to estimate the transformation. The more trials are conducted, the greater the likelihood of finding the optimal transformation model, but it will also lead to an increase in computing time. When the installation height of the upstream camera and the downstream camera increases or the baseline distance between the upstream camera and the downstream camera increases, the change in image details may make the matching of feature points less reliable. Therefore, the maximum allowable number of trials is increased to handle some mismatched points and ensure effective transformation estimation.

[0045] When the installation height of the upstream camera and the downstream camera increases or the baseline distance between the upstream camera and the downstream camera increases, the geometric relationship between the feature points may become more complex. Therefore, the confidence level can be appropriately increased to 95 or 99 to ensure that the influence of outliers is minimized. In the case of reduced resolution, if the number of feature points decreases, the confidence level can also be considered to be reduced (such as 80) to ensure that enough inliers can be found for effective transformation estimation.

[0046] When the installation height of the upstream camera and the downstream camera increases or the baseline distance between the upstream camera and the downstream camera increases, the distance difference between the points in the image after projection increases. Therefore, the maximum allowable distance needs to be increased.

[0047] Only by way of example, the outlier rejection parameter determination model can determine the outlier rejection parameter according to the position parameters of the upstream camera and the downstream camera. Among them, the outlier rejection parameter determination model can be a convolutional neural network model.

[0048] Feature point matching and outlier rejection are performed on the upstream image feature points and downstream image feature points corresponding to the t+1 moment to determine the feature points corresponding to the t+1 moment in the same way as the feature points corresponding to the t moment are determined by performing feature point matching and outlier rejection on the upstream image feature points and downstream image feature points corresponding to the t moment, which will not be elaborated here.

[0049] In some embodiments, according to the three-dimensional coordinates of the feature points corresponding to the t moment and the three-dimensional coordinates of the feature points corresponding to the t+1 moment, the distribution of the water depth of waterlogging is determined, including: Based on the three-dimensional coordinates of the feature points corresponding to the t moment and the three-dimensional coordinates of the feature points corresponding to the t+1 moment, a water depth point set is obtained; Scattered point interpolation is performed on the water depth point set to obtain surface data; Median filtering and PCHIP (Piecewise Cubic Hermite Interpolating Polynomial) interpolation are performed on the surface data to determine the distribution of the water depth of waterlogging.

[0050] Specifically, scatter interpolation is performed on the water depth point set, that is, a continuous surface is generated between discrete water depth data points, and the result can be represented by an interpolation function representation.

[0051] Median filtering is performed on the interpolated surface data, that is, the median of the neighboring pixels is used to replace the central pixel. The formula is: , where G(x, y) is the gray value of the original image, is after filtering, median(·) represents selecting the median of all pixels within the window, , N is the size of the window, and the window range is [-k, k].

[0052] PCHIP interpolation, a piecewise cubic interpolation method, first fills the columns and then fills the rows, ensuring the monotonicity of the interpolation result and making the generated water depth distribution have good smoothness. The specific steps are as follows: For the discrete water depth point set , calculate the slope of each segment: , Determine the tangent slope of each point according to the slope of each segment . Use boundary conditions to ensure that the tangent slope at the endpoints does not exceed the slope limit between data points.

[0053] Calculate the interpolation curve y(x), which is defined according to the slopes in different intervals: For : , where, , .

[0054] Use the filled data to generate a grid again and perform visualization to obtain the water depth distribution of the entire monitoring area.

[0055] Step 150, perform floating object centroid recognition and floating object centroid matching on the upstream image collected by the upstream camera at time t and the downstream image collected by the downstream camera at time t, and determine the three-dimensional coordinates of the floating object at time t according to the camera calibration parameters.

[0056] In some embodiments, step 150 specifically includes: Use the OTSU algorithm to calculate the gray threshold of the upstream target area image corresponding to time t, binarize the upstream target area image corresponding to time t, generate the binarized image of the upstream target area corresponding to time t, and determine the centroid coordinates of the connected region formed by the floating object in the binarized image of the upstream target area corresponding to time t; According to the homography matrix obtained by feature matching, convert the centroid coordinates of the connected regions formed by floating objects in the binary image of the upstream target region corresponding to time t into the centroid coordinates in the downstream camera coordinate system; Use the OTSU algorithm to calculate the gray threshold of the downstream target region image corresponding to time t, binarize the downstream target region image corresponding to time t, generate the binary image of the downstream target region corresponding to time t, and determine the centroid coordinates of the connected regions formed by floating objects in the binary image of the downstream target region corresponding to time t; According to the centroid coordinates in the downstream camera coordinate system corresponding to time t and the centroid coordinates of the connected regions formed by floating objects in the binary image of the downstream target region corresponding to time t, perform centroid matching of floating objects to determine the centroid of the matched floating objects corresponding to time t; According to the camera calibration parameters, determine the three-dimensional coordinates of the centroid of the matched floating objects at time t.

[0057] Specifically, convert the upstream target region image and the downstream target region image corresponding to time t from the original RGB image to a grayscale image. The grayscale value G can be calculated by the following formula: , where R, G, and B correspond to the values of the red, green, and blue channels respectively.

[0058] Use the OTSU inter-class variance method to calculate the appropriate gray threshold to separate the floating objects and the water body into the foreground and the background, and generate a binary image. The steps can be summarized as follows: Calculate the gray histogram of the image, that is, calculate the number of pixels corresponding to each gray level and visualize it in the form of a bar chart; Normalize the histogram, where, is the number of pixels of gray level i, N is the total number of all pixels in the image, is the probability of the occurrence of gray level i; For each possible threshold , calculate the inter-class variance of the foreground and the background: , where, is the global gray mean of the image, is the probability of the foreground (pixels with gray level less than or equal to ), is the probability of the background (pixels with gray level greater than ), is the gray mean of the foreground, is the gray mean of the background, L is the total number of possible gray levels in the image.

[0059] The between-class variance is expressed as: , Select the threshold that maximizes the between-class variance : , Based on the obtained grayscale threshold Binarize the image: , where is the binary image, the foreground (floating object) value is 1, and the background (water body) value is 0.

[0060] In the binarized image, the connected regions formed by floating objects are displayed as white foreground, and the water surface is displayed as black background. Assign a unique label to each region and calculate its properties to obtain the centroid. The centroid calculation formula is: , In the formula: is the pixel value in the binary image; C x ，C y are the centroid coordinates of the connected region respectively; x, y are the coordinates of the pixel point.

[0061] In the feature point matching stage, there is a 3×3 homography matrix H containing transformation information between the feature points and of the upstream camera and the downstream camera. Use this relationship to adjust the centroid coordinates of the identified floating objects, that is, adjust the centroid coordinates of the floating objects identified in the image captured by the upstream camera to the downstream camera coordinate system so that they are in the same coordinate system, and obtain .

[0062] Use the matching probability method to match the centroid coordinates of the floating objects according to the centroid coordinates in the downstream camera coordinate system corresponding to time t and the centroid coordinates of the connected regions formed by the floating objects in the binary image of the downstream target region corresponding to time t. Retain the centroid coordinates of the floating objects with successful matches in the centroid point set, delete the centroid coordinates of the floating objects that are not matched, and then add the M-estimation sample consensus algorithm MSAC to ensure that all the remaining centroid coordinates are accurate. The detailed description of the matching probability method is as follows: Initialize the matching probability: For each pair of centroids to be matched, initialize the matching probability between them. Set the initial probability value based on the distance and the position of neighboring centroids.

[0063] Calculate the distance: Calculate the distances to all possible matching centroids. Similarly, the Euclidean distance is used for measurement and can be expressed as: , Set the threshold: Filter the matches through a pre-set threshold. Only the centroids with distances within the threshold range will be considered as possible matches.

[0064] Influence of neighboring particles: Consider the neighborhood information of the centroids to adjust the matching probability.

[0065] Iterative update: Perform iterative calculations based on the final matching probability, and gradually update the matching probability of each centroid until convergence or reaching the preset number of iterations.

[0066] Normalize the probability: Ensure that all matching probabilities and non-matching probabilities are normalized so that they can be conveniently compared and selected.

[0067] Select the matching particles: Set a probability threshold. Once the matching probability of a certain centroid exceeds this threshold, consider this centroid as a successful match.

[0068] According to the pixel coordinates of the floating matter centroid correctly matched by the upstream camera and the downstream camera , using the projection matrix Q obtained by calibration, where the projection matrix Q is the product of the internal parameter matrix and the external parameter matrix of the camera, calculate the corresponding three-dimensional world coordinates , and obtain the floating matter centroid point sets at time t and time t+1.

[0069] Specifically, the point in space and the feature point and are related as follows: , Combining the above two equations, the system of equations for solving the three-dimensional coordinates of point P is: , where, S 1 and S 2 are the scale factors of the upstream camera and the downstream camera respectively, M 1 and M 2 are the internal parameter matrices of the upstream camera and the downstream camera respectively, N 1 and N 2 are the external parameter matrices of the upstream camera and the downstream camera respectively, Q 1 and Q2 They are the projection matrices of the upstream camera and the downstream camera respectively, that is, the product of the camera internal parameter matrix and the external parameter matrix.

[0070] Step 160: Identify the centroid of floating objects and match the centroids of floating objects in the upstream image captured by the upstream camera at time t + 1 and the downstream image captured by the downstream camera at time t + 1. According to the camera calibration parameters, determine the three-dimensional coordinates of the floating object at time t + 1.

[0071] Specifically, the method for determining the three-dimensional coordinates of the matched floating object centroid at time t + 1 is the same as that for determining the three-dimensional coordinates of the matched floating object centroid at time t, which will not be elaborated here.

[0072] Step 170: Determine the distribution of the urban waterlogging flow velocity according to the three-dimensional coordinates of the floating object at time t, the three-dimensional coordinates of the floating object at time t + 1, and the time difference between time t and time t + 1.

[0073] In some embodiments, the urban waterlogging flow velocity is calculated based on the following formula: , where, V is the urban waterlogging flow velocity corresponding to the centroid of the floating object, is the velocity component of the urban waterlogging flow velocity corresponding to the centroid of the floating object in the x direction, is the x-axis coordinate of the centroid of the floating object at time t + 1, is the x-axis coordinate of the centroid of the floating object at time t, is the velocity component of the urban waterlogging flow velocity corresponding to the centroid of the floating object in the y direction, is the y-axis coordinate of the centroid of the floating object at time t + 1, is the y-axis coordinate of the centroid of the floating object at time t, is the velocity component of the urban waterlogging flow velocity corresponding to the centroid of the floating object in the z direction, is the z-axis coordinate of the centroid of the floating object at time t + 1, is the z-axis coordinate of the centroid of the floating object at time t, is the time difference between time t and time t + 1.

[0074] Interpolate the calculated discrete three-dimensional flow velocities into a uniform grid and fill them in parallel to generate a dense three-dimensional surface flow velocity.

[0075] In some embodiments, the real-time urban waterlogging monitoring method based on the urban electronic eye system may further include step 180: superimpose the urban waterlogging water depth distribution and the urban waterlogging flow velocity distribution to form an integrated monitoring result of water depth and flow velocity.

[0076] Specifically, according to the original water surface shape , in the vector map after three-dimensional interpolation, find the flow velocity layer closest to the water surface height, which can be achieved by minimizing the height difference and can be expressed as: , where, is a two-dimensional grid representing the water level at each location; is a variable used to index and locate the flow velocity data layer in three-dimensional space to ensure that the output velocity components match the current water surface height, is the height function in three-dimensional space, where i and j represent positions on the horizontal grid, and k represents the layer index in the vertical direction.

[0077] Next, in combination with experiments, the beneficial effects of the real-time monitoring method for urban waterlogging based on the urban electronic eye system will be described.

[0078] An experimental flume is used to simulate urban waterlogging. The experimental flume is 22.8 m long, 1 m wide, and 1.09 m high, with a bottom slope of 0.001. It consists of a 4.8-m-long reservoir area, an 18-m-long flood area, and a gate, and is a compound river channel. The main channel is 0.334 m wide, the floodplain is 0.666 m wide, and the floodplain height is 0.15 m. The water surface floating objects used in the experiment are naturally fallen leaves. The camera used in the experiment is the Hikvision DS-2CD7A427FWD, with a resolution of 1920×1080 and a frame rate of 25 frames / s. The calibration board is a checkerboard with a size of 500×600 mm. Stationary water bodies with depths of 0.4 m and 0.23 m are stored in the reservoir area and the flood area of the flume respectively. Floating objects are scattered on the water surface in advance to make them as evenly distributed as possible. After the above preparations are completed, two cameras are turned on to take pictures in advance, and then the gate is opened to simulate the flood propagation process in the urban flood area. After the entire flood propagation process is completely recorded, the cameras are turned off. Here, the flood at t = 7.6 s is selected for analysis. The results after extracting the feature points of the pictures by the Speeded Up Robust Features (SURF) algorithm and removing the wrong matching points are shown in the appendix Figure 3 as shown, and the results of matching the centroids of the floating objects in the pictures taken by the upstream camera and the downstream camera by the coordinate fitting method and the matching probability method are shown in Figure 4 as shown. The results of the water depth distribution and the water depth-coupled three-dimensional flow velocity distribution obtained by calculation are shown in Figure 5 and Figure 6 respectively.

[0079] Figure 7 is a schematic diagram of the modules of the real-time monitoring system for urban waterlogging based on the urban electronic eye system according to some embodiments of this specification. As shown in Figure 7 , the real-time monitoring system for urban waterlogging based on the urban electronic eye system may include an image acquisition module, a camera calibration module, an image acquisition module, and an urban waterlogging analysis module.

[0080] The image acquisition module may include multiple camera groups. Among them, each camera group includes an upstream camera and a downstream camera, and there is an overlapping shooting area between the upstream camera and the downstream camera.

[0081] The camera calibration module can be used to calibrate the upstream camera and the downstream camera for each camera group to determine the camera calibration parameters.

[0082] The image acquisition module can be used to acquire the upstream images acquired by the upstream camera at time t and t + 1 and the downstream images acquired by the downstream camera at time t and t + 1.

[0083] The waterlogging analysis module can be used to perform feature matching on the upstream images acquired by the upstream camera at time t and the downstream images acquired by the downstream camera at time t, and perform feature matching on the upstream images acquired by the upstream camera at time t + 1 and the downstream images acquired by the downstream camera at time t + 1 to determine the waterlogging water depth distribution; perform floating object centroid recognition and floating object centroid matching on the upstream images acquired by the upstream camera at time t and the downstream images acquired by the downstream camera at time t, and determine the three-dimensional coordinates of the floating object at time t according to the camera calibration parameters; perform floating object centroid recognition and floating object centroid matching on the upstream images acquired by the upstream camera at time t + 1 and the downstream images acquired by the downstream camera at time t + 1, and determine the three-dimensional coordinates of the floating object at time t + 1 according to the camera calibration parameters; determine the waterlogging flow velocity distribution according to the three-dimensional coordinates of the floating object at time t, the three-dimensional coordinates of the floating object at time t + 1, and the time difference between time t and time t + 1.

[0084] The real-time waterlogging monitoring system based on the urban electronic eye system can be used to execute the real-time waterlogging monitoring method based on the urban electronic eye system. For more descriptions of the real-time waterlogging monitoring system based on the urban electronic eye system, reference can be made to the relevant descriptions of the real-time waterlogging monitoring method based on the urban electronic eye system, which will not be described here.

[0085] Finally, it should be understood that the embodiments described in this specification are only used to illustrate the principles of the embodiments of this specification. Other deformations may also fall within the scope of this specification. Therefore, by way of example rather than limitation, alternative configurations of the embodiments of this specification can be regarded as consistent with the teachings of this specification. Accordingly, the embodiments of this specification are not limited to the embodiments explicitly introduced and described in this specification.

Claims

1. A real-time monitoring method for urban flooding based on an urban electronic eye system, characterized in that: include: Setting a plurality of camera groups, wherein the camera group includes an upstream camera and a downstream camera, and the shooting areas of the upstream camera and the downstream camera overlap; For each camera group, calibrate the upstream camera and the downstream camera to determine camera calibration parameters; Acquire the upstream images captured by the upstream camera at time t and time t+1 and the downstream images captured by the downstream camera at time t and time t+1; Perform feature matching on the upstream image captured by the upstream camera at time t and the downstream image captured by the downstream camera at time t, and perform feature matching on the upstream image captured by the upstream camera at time t+1 and the downstream image captured by the downstream camera at time t+1, to determine the water depth distribution of the waterlogging; Perform floating material centroid identification and floating material centroid matching on the upstream image collected by the upstream camera at time t and the downstream image collected by the downstream camera at time t, and determine the three-dimensional coordinates of the floating object at time t according to the camera calibration parameters; Perform floating material centroid identification and floating material centroid matching on the upstream image collected by the upstream camera at time t+1 and the downstream image collected by the downstream camera at time t+1, and determine the three-dimensional coordinates of the floating object at time t+1 according to the camera calibration parameters; The waterlogging flow velocity distribution is determined based on the three-dimensional coordinates of the floating object at time t, the three-dimensional coordinates of the floating object at time t+1, and the time difference between time t and time t+1.

2. The real-time monitoring method for urban flooding based on the urban electronic eye system according to claim 1 is characterized in that: Calibrate the upstream camera and the downstream camera to determine camera calibration parameters, including: A calibration plate is set in the overlapping shooting area of ​​the upstream camera and the downstream camera, and images of the calibration plate in different states are acquired through the upstream camera and the downstream camera; According to the images of the calibration plate in different states acquired by the upstream camera and the downstream camera, the intrinsic parameter matrices of the upstream camera and the downstream camera are determined; Setting a plurality of ground control points in the overlapping shooting area of ​​the upstream camera and the downstream camera, and acquiring images of the plurality of ground control points through the upstream camera and the downstream camera; According to the images of multiple ground control points and the world coordinates of multiple ground control points obtained by the upstream camera and the downstream camera, the external parameter matrices of the upstream camera and the downstream camera are determined, wherein the camera calibration parameters at least include the internal parameter matrix and the external parameter matrix of the upstream camera and the downstream camera.

3. The real-time monitoring method for urban waterlogging based on the urban electronic eye system according to claim 2 is characterized in that: Performing feature matching on the upstream image captured by the upstream camera at time t and the downstream image captured by the downstream camera at time t, and performing feature matching on the upstream image captured by the upstream camera at time t+1 and the downstream image captured by the downstream camera at time t+1, to determine the water depth distribution of the waterlogging, including: Performing image preprocessing on the upstream image captured by the upstream camera at time t and the downstream image captured by the downstream camera at time t, and extracting the upstream target area image and the downstream target area image corresponding to time t; Determining feature point extraction parameters according to position parameters of the upstream camera and the downstream camera; According to the feature point extraction parameters, extract the upstream image feature points corresponding to time t from the upstream target area image corresponding to time t, extract the downstream image feature points corresponding to time t from the downstream target area image corresponding to time t, extract the upstream image feature points corresponding to time t+1 from the upstream target area image corresponding to time t+1, and extract the downstream image feature points corresponding to time t+1 from the downstream target area image corresponding to time t+1; Perform feature point matching and error elimination on the feature points of the upstream image and the downstream image corresponding to time t to determine the feature points corresponding to time t; According to the camera calibration parameters, determine the three-dimensional coordinates of the feature point corresponding to time t; Perform feature point matching and error elimination on the upstream image feature points and the downstream image feature points corresponding to time t+1 to determine the feature points corresponding to time t+1; According to the camera calibration parameters, determine the three-dimensional coordinates of the feature point corresponding to time t+1; The water depth distribution of the waterlogging is determined based on the three-dimensional coordinates of the feature points corresponding to time t and the three-dimensional coordinates of the feature points corresponding to time t+1.

4. The real-time monitoring method for urban waterlogging based on the urban electronic eye system according to claim 3 is characterized in that: The location parameters include at least the erection height and the baseline distance; The feature point extraction parameters include at least a metric threshold of the feature point, an octave number, and a number of scale levels used in each octave; According to the feature point extraction parameters, extracting the upstream image feature points corresponding to time t from the upstream target area image corresponding to time t, and extracting the downstream image feature points corresponding to time t from the downstream target area image corresponding to time t, including: According to the feature point extraction parameters, the upstream image feature points corresponding to time t are extracted from the upstream target area image corresponding to time t by using the SURF algorithm, and the downstream image feature points corresponding to time t are extracted from the downstream target area image corresponding to time t.

5. The real-time monitoring method for urban waterlogging based on the urban electronic eye system according to claim 3 is characterized in that: Perform feature point matching and error elimination on the upstream image feature points and the downstream image feature points corresponding to time t to determine the feature points corresponding to time t, including: Calculate the distance between any upstream image feature point corresponding to time t and any downstream image feature point corresponding to time t; Determine a matching feature point group corresponding to time t according to a distance threshold and a distance between any upstream image feature point corresponding to time t and any downstream image feature point corresponding to time t, wherein the matching feature point group includes an upstream image feature point and a downstream image feature point; Determine error-eliminating parameters according to the position parameters of the upstream camera and the downstream camera, wherein the error-eliminating parameters at least include a maximum allowable number of trials, a confidence level, and a maximum allowable distance; According to the error-removing parameters, the matched feature point group corresponding to time t is removed by MSAC to determine the feature point corresponding to time t.

6. The real-time monitoring method for urban waterlogging based on the urban electronic eye system according to claim 3 is characterized in that: According to the three-dimensional coordinates of the feature point corresponding to time t and the three-dimensional coordinates of the feature point corresponding to time t+1, the water depth distribution of the waterlogging is determined, including: According to the three-dimensional coordinates of the feature points corresponding to time t and the three-dimensional coordinates of the feature points corresponding to time t+1, a water depth point set is obtained; Performing scattered point interpolation on the water depth point set to obtain surface data; The surface data is subjected to median filtering and PCHIP interpolation to determine the water depth distribution of the waterlogging.

7. The real-time monitoring method for urban flooding based on the urban electronic eye system according to claim 5 is characterized in that: Performing floating material centroid identification and floating material centroid matching on the upstream image collected by the upstream camera at time t and the downstream image collected by the downstream camera at time t, and determining the three-dimensional coordinates of the floating object at time t according to the camera calibration parameters, including: Calculate the grayscale threshold of the upstream target area image corresponding to time t using the OTSU algorithm, binarize the upstream target area image corresponding to time t, generate a binary image of the upstream target area corresponding to time t, and determine the centroid coordinates of the connected area formed by the floating objects in the binary image of the upstream target area corresponding to time t; According to the homography matrix obtained by feature matching, the centroid coordinates of the connected area formed by the floating objects in the binary image of the upstream target area corresponding to time t are converted into the centroid coordinates in the downstream camera coordinate system; Calculate the gray threshold of the downstream target area image corresponding to time t using the OTSU algorithm, binarize the downstream target area image corresponding to time t, generate a binary image of the downstream target area corresponding to time t, and determine the centroid coordinates of the connected area formed by the floating objects in the binary image of the downstream target area corresponding to time t; According to the centroid coordinates of the downstream camera coordinate system corresponding to time t and the centroid coordinates of the connected area formed by the floating objects in the binary image of the downstream target area corresponding to time t, the floating material centroid is matched to determine the matching floating material centroid corresponding to time t; According to the camera calibration parameters, the three-dimensional coordinates of the center of the matching floating material at time t are determined.

8. The real-time monitoring method for urban flooding based on the urban electronic eye system according to claim 7 is characterized in that: The waterlogging flow velocity is calculated based on the following formula: ， in, V is the waterlogging velocity corresponding to the floating mass center, is the velocity component of the waterlogging velocity in the x direction corresponding to the floating mass center, is the x-axis coordinate of the floating material center at time t+1, is the x-axis coordinate of the floating mass center at time t, is the velocity component of the waterlogging velocity in the y direction corresponding to the floating mass center, is the y-axis coordinate of the floating material center at time t+1, is the y-axis coordinate of the floating mass center at time t, is the velocity component of the waterlogging velocity in the z direction corresponding to the floating mass center, is the z-axis coordinate of the floating material center at time t+1, is the z-axis coordinate of the floating mass center at time t, is the time difference between time points t and t+1.

9. The method for real-time monitoring of urban waterlogging based on the urban electronic eye system according to any one of claims 1 to 8, characterized in that: Also includes: The water depth distribution of urban flooding and the flow velocity distribution of urban flooding are superimposed to form an integrated monitoring result of water depth and flow velocity.

10. The real-time monitoring system for urban flooding based on the urban electronic eye system is characterized by: The method for real-time monitoring of urban flooding based on the urban electronic eye system according to any one of claims 1 to 9 comprises: An image acquisition module, comprising a plurality of camera groups, wherein the camera group comprises an upstream camera and a downstream camera, and the shooting areas of the upstream camera and the downstream camera overlap; A camera calibration module, used for calibrating the upstream camera and the downstream camera for each camera group, and determining camera calibration parameters; An image acquisition module, used to acquire the upstream images captured by the upstream camera at time t and time t+1 and the downstream images captured by the downstream camera at time t and time t+1; The waterlogging analysis module is used to perform feature matching on the upstream image captured by the upstream camera at time t and the downstream image captured by the downstream camera at time t, and to perform feature matching on the upstream image captured by the upstream camera at time t+1 and the downstream image captured by the downstream camera at time t+1, so as to determine the water depth distribution of the waterlogging; to perform floating material centroid identification and floating material centroid matching on the upstream image captured by the upstream camera at time t and the downstream image captured by the downstream camera at time t, and to determine the three-dimensional coordinates of the floating object at time t according to the camera calibration parameters; to perform floating material centroid identification and floating material centroid matching on the upstream image captured by the upstream camera at time t+1 and the downstream image captured by the downstream camera at time t+1, and to determine the three-dimensional coordinates of the floating object at time t+1 according to the camera calibration parameters; and to determine the waterlogging flow velocity distribution according to the three-dimensional coordinates of the floating object at time t, the three-dimensional coordinates of the floating object at time t+1, and the time difference between the time points at time t and time t+1.

Images