A method and system for real-time identification of water body information of a flowing water body

Abstract

The present invention relates to a method and system for real-time identification of water information of a flowing water body, comprising the following steps: using a depth camera to continuously shoot an identification object containing a flowing water body to obtain depth data; reorganizing the depth data in a fixed format to reconstruct a three-dimensional point cloud of the identification object and dynamically update it to obtain a digital twin model of the flowing water body; using a plane fitting algorithm to identify the water surface based on the reconstructed three-dimensional point cloud, and segmenting the water surface; combining the segmentation result of the water surface, and identifying the water body information based on the reconstructed three-dimensional point cloud. Compared with the prior art, the present invention can realize digital modeling of flowing water bodies in various scenarios, and intelligently identify water body information including depth and volume, which not only breaks through the deficiency of traditional identification methods being limited to the two-dimensional level, but also realizes non-contact and precise measurement of water body information without a water gauge.

Description

Technical Field

[0001] The present invention relates to the technical field of water body monitoring, and in particular to a method and system for real-time identification of water body information of a flowing water body. Background Art

[0002] Water body information, such as depth and volume, is a key parameter in water resources management, and its accurate acquisition is of great significance for flood control, irrigation system design, water quality monitoring and other fields.

[0003] On the one hand, traditional water monitoring often relies on physical sampling or installation of fixed monitoring equipment, such as water level measurement stations, sensors, flow meters, etc. These methods require high costs in terms of equipment purchase, installation, maintenance, and personnel operation. On the other hand, traditional monitoring methods often rely on regular on-site sampling and data recording, which means that the frequency of data updates is limited by the sampling frequency of personnel. In some cases, such as bad weather or remote geographical locations, the interval between data collection may be long, making it impossible to reflect changes in water conditions in real time. Therefore, traditional water monitoring methods face challenges such as high cost, low frequency of data updates, and difficulty in achieving large-scale real-time monitoring.

[0004] The current image recognition field focuses on water level recognition for the extraction of water body information, but these methods have the following two disadvantages: ① Most water level recognition methods require the installation of a water gauge in the recognition area, and the water level information is obtained by image processing of the water gauge image. However, in practical applications, there are many scenarios where the installation of a water gauge cannot be guaranteed, such as turbulent rivers, sewers, etc. ② The recognition methods for non-installed water gauges are mostly machine learning methods. In order to ensure the accuracy of recognition, the model must be trained with a large amount of data before recognition. Therefore, this type of method is highly dependent on the amount and accuracy of data. At the same time, the current image recognition methods are still limited to the two-dimensional level and do not recognize the volume of the water body. Summary of the invention

[0005] The purpose of the present invention is to provide a method and system for real-time identification of water information of flowing water bodies in order to overcome the defects of the above-mentioned prior art that water body monitoring requires the installation of a water gauge, which is inconvenient to install; or the use of machine learning methods, but is highly dependent on the amount and accuracy of data.

[0006] The purpose of the present invention can be achieved by the following technical solutions:

[0007] A method for real-time identification of water body information of a flowing water body comprises the following steps:

[0008] Use a depth camera to continuously shoot the identified object containing flowing water to obtain depth data;

[0009] The depth data is reorganized in a fixed format to reconstruct and dynamically update the three-dimensional point cloud of the identified object to obtain a digital twin model of the flowing water body;

[0010] According to the reconstructed 3D point cloud, the water surface is identified and segmented using a plane fitting algorithm;

[0011] Combined with the segmentation results of the water surface, the water body information is identified according to the reconstructed 3D point cloud.

[0012] Furthermore, the process of acquiring the depth data specifically includes:

[0013] Using a depth camera, take photos when the identified object does not contain water as reference data, and take photos when the identified object contains water to obtain water body change data.

[0014] Furthermore, the method further includes: after obtaining the depth data, performing data preprocessing on the depth data, and the data preprocessing includes: filtering outliers.

[0015] Furthermore, the process of filtering outliers includes:

[0016] Calculate the average value of the depth data obtained by the current frame image and standard deviation ;

[0017] Determine the current depth data Is it satisfied? If it is not satisfied, it is an outlier and the depth data is removed.

[0018] Furthermore, the reorganization of the depth data according to the fixed format is specifically as follows:

[0019] The depth data is arranged in a ply format and reorganized in a predetermined order of the pixel's x coordinate value, y coordinate value, z depth value, pixel color red R attribute value, green G attribute value, and blue B attribute value.

[0020] Furthermore, the plane fitting algorithm is a random sampling consistency algorithm, which specifically includes the following steps:

[0021] S31: Set the number of iterations, the minimum number of samples required for fitting, and the fitting error threshold parameters;

[0022] S32: randomly selecting a point cloud with a minimum number of samples from the reconstructed 3D point cloud for initial plane estimation;

[0023] S33: performing plane fitting on the selected point cloud to obtain a fitting plane;

[0024] S34: Calculate the distances from all points in the point cloud to the fitting plane, mark the points whose distances are less than or equal to the fitting error threshold as inliers; mark the points whose distances are greater than the fitting error threshold as outliers, and count the number of inliers;

[0025] S35: Return to step S32 and perform loop iteration until the number of iterations is reached;

[0026] S36: Select a fitting plane with a maximum number of inner points as a water surface recognition result.

[0027] Furthermore, the identified water body information includes water body depth, and the identification process of the water body depth is specifically as follows:

[0028] According to the reference data of the identified object without water body, the distance from each point in the reference data to the water surface is calculated to obtain the depth value of each point.

[0029] Furthermore, the calculation expression of the depth value is:

[0030]

[0031] In the formula, is the depth value, Indicates the coordinate value of the current point. Parameters representing the water surface.

[0032] Furthermore, the identified water body information includes the water body volume, and the water body volume identification process specifically includes the following steps:

[0033] Downsampling the point cloud below the water surface to a voxel grid;

[0034] Calculate the average of the depth values ​​of all points in a single voxel in the voxel grid as the depth value of the voxel;

[0035] According to the depth value of each voxel, the volume of water in a single voxel is calculated;

[0036] According to the volume of water in each voxel, the total volume of all water bodies is calculated to obtain the water volume.

[0037] The present invention also provides a real-time identification system for water body information of a flowing water body, comprising:

[0038] Depth camera, used to continuously shoot the identified object;

[0039] The water body information real-time identification module is used to execute the water body information real-time identification method of a flowing water body as described above.

[0040] Compared with the prior art, the present invention has the following advantages:

[0041] (1) The present invention collects depth data of flowing water bodies and utilizes three-dimensional reconstruction technology to break through the traditional two-dimensional level, realize real-time digital modeling of flowing water bodies, identify the water surface using a plane fitting algorithm, and segment the water surface; combined with the fitting equation of the horizontal plane, the water body information is intelligently identified to achieve more intuitive observation of the water body and obtain high-precision water body information at the same time. This not only breaks through the deficiency of traditional identification methods being limited to the two-dimensional level, but also realizes non-contact and precise measurement of water body information without a water gauge.

[0042] (2) The present invention uses a random sampling sample consistency algorithm to realize the recognition and segmentation of the water surface, and uses the point-to-plane distance formula to obtain the water depth information of any point in the water body within the shooting range, which is not limited to the acquisition of water level information in a certain area.

[0043] (3) The present invention achieves the acquisition of water volume that is not covered by other methods by downsampling the voxel grid of the three-dimensional point cloud below the water surface, calculating the volume of a single voxel and summing the volumes of all voxels.

[0044] (4) The present invention realizes the acquisition of water body information, including water level information and water body volume, without the need for water gauge installation and model pre-training, thereby improving the efficiency of traditional monitoring methods, enriching the acquisition of hydraulic parameters, and greatly reducing manpower and financial costs.

[0045] (5) The method provided by the present invention can be used in fields that require water monitoring, such as water conservancy and hydropower engineering, river monitoring, and irrigation system design. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 A schematic diagram of the overall process of a method for real-time identification of water body information of a flowing water body provided in an embodiment of the present invention;

[0047] Figure 2 A schematic diagram of a horizontal plane fitting process provided in an embodiment of the present invention;

[0048] Figure 3 A schematic diagram of a depth and volume recognition process provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0049] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0050] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0051] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0052] Example 1

[0053] like Figure 1 As shown, this embodiment provides a method for real-time identification of water body information of a flowing water body, comprising the following steps:

[0054] S1: Use a depth camera to continuously shoot the identification object containing flowing water to obtain depth data;

[0055] Specifically, the implementation system of this solution mainly consists of a data acquisition module and a data analysis module. The data acquisition module is a depth camera. When in use, the resolution and frame rate of the depth camera are first configured; then the identified object is photographed when it does not contain water as reference data; when the identified object contains flowing water, it is continuously photographed to obtain water change data.

[0056] Preferably, after obtaining the depth data, data preprocessing is performed to filter out outliers. Outliers in the data may be caused by various factors such as environmental interference, different reflectivity, and occlusion. These outliers will affect the accuracy and reliability of the depth value. By filtering these outliers, the overall accuracy of the depth data can be improved, thereby providing more accurate input data for subsequent three-dimensional reconstruction.

[0057] The specific steps of data preprocessing are: Calculate the average value of the depth data obtained by the current frame image ; Calculate the standard deviation of the depth data obtained by the current frame image ; Determine whether the data meets If it is not satisfied, it is an outlier and the data will be removed.

[0058] S2: reorganize the depth data according to a fixed format to reconstruct and dynamically update the three-dimensional point cloud of the identified object to obtain a digital twin model of the flowing water body;

[0059] A 3D point cloud is a collection of points distributed in three-dimensional space. Common formats include XYZ, PCD, PLY, etc. The PLY format is flexible and scalable, supporting multiple types of data. The PLY file can be in ASCII or binary format, providing a flexible data exchange method.

[0060] Therefore, combined with the depth value of each pixel obtained in step S1, the data is arranged in .ply format, and the arrangement order is the pixel's x coordinate value, y coordinate value, z depth value, pixel color red R attribute value, green G attribute value, and blue B attribute value. By reconstructing each frame of data in three dimensions, real-time digital modeling of flowing water is achieved.

[0061] S3: Based on the reconstructed 3D point cloud, the water surface is identified and segmented using a plane fitting algorithm;

[0062] Optionally, the plane fitting algorithm is a random sampling consensus algorithm, which is an iterative method that can estimate the parameters of a mathematical model from a set of observation data containing outliers. The algorithm is used to fit and identify the horizontal plane, and its flow chart is as shown in the attached figure. Figure 2 As shown, the following steps are included:

[0063] S31: Set the number of iterations, the minimum number of samples required for fitting, and the fitting error threshold parameters;

[0064] S32: randomly selecting a point cloud with a minimum number of samples from the reconstructed 3D point cloud for initial plane estimation;

[0065] S33: Perform plane fitting on the selected point cloud using Indicates that For the required parameters, the fitting plane is obtained;

[0066] S34: Calculate the distances from all points in the point cloud to the fitting plane, mark the points whose distances are less than or equal to the fitting error threshold as inliers; mark the points whose distances are greater than the fitting error threshold as outliers, and count the number of inliers;

[0067] S35: Return to step S32 and perform loop iteration until the number of iterations is reached;

[0068] S36: Select a fitting plane with a maximum number of inner points as a water surface recognition result.

[0069] That is, the steps of random sampling, plane fitting, and interior point test are repeated; when the number of iterations reaches the set parameters, the plane model with the largest number of interior points is selected, and its fitting parameters are output.

[0070] All the internal points in the final output result are points on the horizontal plane. By extracting these points, the segmentation of the horizontal plane is achieved. These points and the three-dimensional point cloud without water collected in step S1 are further used to identify water body information.

[0071] S4: Combined with the segmentation results of the water surface, the water body information is identified according to the reconstructed 3D point cloud.

[0072] The identified water body information includes the water level depth at each point and the water volume in the current image. The flow chart is shown in the attached Figure 3 , specifically including:

[0073] Step S41: Identify the depth of the water body, combine the three-dimensional point cloud data without water body, and use the point-to-plane distance formula to calculate the distance from each point to the water surface, that is, the depth value of each point. The specific formula is:

[0074]

[0075] in, Indicates the depth of the point. Represents the coordinate value of the point. Parameters representing the horizontal plane.

[0076] Step S42: Identify the water volume and perform voxel grid downsampling on the point cloud below the water level. The grid size is ;

[0077] Step S43: Calculate the average depth value of all points in a single voxel as the depth value of the voxel ;

[0078] Step S44: Calculate the volume of water within a single voxel ;

[0079] Step S45: Calculate the total volume of all water bodies ,in is the total number of voxels.

[0080] Through the above steps, real-time digital modeling and dynamic updating of flowing water bodies can be achieved, a digital twin of flowing water bodies can be established, and on this basis, water body information including water level depth and water volume can be obtained. This method does not require calibration and is simple and efficient. Contents not described in detail in this specification belong to the prior art known to professional and technical personnel in this field.

[0081] Example 2

[0082] This embodiment provides a system for real-time identification of water body information of a flowing water body, including:

[0083] Depth camera, used to continuously shoot the identified object;

[0084] The water body information real-time identification module is used to execute a method for real-time identification of water body information of a flowing water body as described in Example 1.

[0085] It should be noted that the specific contents and beneficial effects of the modules of this system can be found in the above method embodiments, which will not be repeated here.

[0086] The preferred specific embodiments of the present invention are described in detail above. It should be understood that a person skilled in the art can make many modifications and changes based on the concept of the present invention without creative work. Therefore, any technical solution that can be obtained by a person skilled in the art through logical analysis, reasoning or limited experiments based on the concept of the present invention on the basis of the prior art should be within the scope of protection determined by the claims.

Claims

1. A method for real-time identification of water body information of a flowing water body, characterized in that: The following steps are involved: Use a depth camera to continuously shoot the identified object containing flowing water to obtain depth data; The depth data is reorganized in a fixed format to reconstruct and dynamically update the three-dimensional point cloud of the identified object to obtain a digital twin model of the flowing water body; According to the reconstructed 3D point cloud, the water surface is identified and segmented using a plane fitting algorithm; Combined with the segmentation results of the water surface, the water body information is identified according to the reconstructed 3D point cloud; The process of acquiring the depth data specifically includes: Using a depth camera, take photos of the identified object when it does not contain water as reference data, and take photos of the identified object when it contains water to obtain water body change data; The identified water body information includes the water body depth, and the identification process of the water body depth is specifically as follows: According to the reference data of the identified object without water body, the distance from each point in the reference data to the water surface is calculated to obtain the depth value of each point; The calculation expression of the depth value is: In the formula, is the depth value, Indicates the coordinate value of the current point. Parameters representing the water surface; The identified water body information includes the water body volume, and the water body volume identification process specifically includes the following steps: Downsampling the point cloud below the water surface to a voxel grid; Calculate the average of the depth values ​​of all points in a single voxel in the voxel grid as the depth value of the voxel; According to the depth value of each voxel, the volume of water in a single voxel is calculated; According to the volume of water in each voxel, the total volume of all water bodies is calculated to obtain the water volume.

2. The method for real-time identification of water body information of a flowing water body according to claim 1, characterized in that: The method further includes: after obtaining the depth data, performing data preprocessing on the depth data, and the data preprocessing includes: filtering outliers.

3. The method for real-time identification of water body information of a flowing water body according to claim 2, characterized in that: The process of filtering outliers includes: Calculate the average value of the depth data obtained by the current frame image and standard deviation ; Determine whether the current depth data Z satisfies If it is not satisfied, it is an outlier and the depth data is removed.

4. The method for real-time identification of water body information of a flowing water body according to claim 1, characterized in that: The reorganization of the depth data according to the fixed format is specifically as follows: The depth data is arranged in a ply format and reorganized in a predetermined order of the pixel's x coordinate value, y coordinate value, z depth value, pixel color red R attribute value, green G attribute value, and blue B attribute value.

5. The method for real-time identification of water body information of a flowing water body according to claim 1, characterized in that: The plane fitting algorithm is a random sampling consistency algorithm, which specifically includes the following steps: S31: Set the number of iterations, the minimum number of samples required for fitting, and the fitting error threshold parameters; S32: randomly selecting a point cloud with a minimum number of samples from the reconstructed 3D point cloud for initial plane estimation; S33: performing plane fitting on the selected point cloud to obtain a fitting plane; S34: Calculate the distances from all points in the point cloud to the fitting plane, mark the points whose distances are less than or equal to the fitting error threshold as inliers; mark the points whose distances are greater than the fitting error threshold as outliers, and count the number of inliers; S35: Return to step S32 and perform loop iteration until the number of iterations is reached; S36: Select a fitting plane with a maximum number of inner points as a water surface recognition result.

6. A real-time water body information identification system for a flowing water body, characterized in that: include: Depth camera, used to continuously shoot the identified object; The water body information real-time identification module is used to execute the water body information real-time identification method of a flowing water body as described in any one of claims 1-5.

Images