Tailing pond beach face water pit identification method and system based on point cloud data

Abstract

The invention provides a tailing pond beach face water pit identification method and system based on point cloud data, and the method comprises the steps: obtaining the point cloud data of a to-be-identified tailing pond, and carrying out the preprocessing of the point cloud data, and obtaining the preprocessed point cloud data; generating a three-dimensional point cloud model according to the preprocessed point cloud data, and determining feature information of each surface voxel corresponding to the to-be-identified tailing pond based on the three-dimensional point cloud model; inputting the feature information of each face voxel into a pre-trained water pit identification model to obtain a water pit identification result of each face voxel; and determining the water pit position of the tailing pond beach face based on the water pit identification result of each face voxel. According to the technical scheme provided by the invention, the tailing pond beach face water pit can be quickly and effectively identified.

Description

Technical Field

[0001] The present application relates to the field of identification and monitoring technology, and in particular to a method and system for identifying water pits on the beach of a tailings pond based on point cloud data. Background Art

[0002] Tailings ponds are places for storing tailings generated during mining and mineral processing. Tailings are solid wastes containing a large amount of useful mineral components, usually mainly fine particles and a certain amount of water. The construction and management of tailings ponds need to take into account a variety of factors such as the physical and chemical properties of tailings, the topography of the reservoir area, and meteorological conditions. The safety of tailings ponds is related to the safety of the surrounding environment and people's lives and property. Therefore, safety monitoring of tailings ponds is particularly important. The presence of water pits on the beach of tailings ponds will cause local wetting of the beach of tailings ponds, reduce the mechanical strength of tailings, and increase the risk of landslides and mudslides. The accumulation of water in the water pits will change the stress distribution inside the tailings pond, which may cause local instability of the dam body and even cause dam breach accidents. The hazards of water pits on the beach of tailings ponds are multifaceted, which not only affect the safe operation of the tailings pond itself, but also pose a serious threat to the surrounding environment and residents. Therefore, timely identification and treatment of water pits on the beach of tailings ponds is of great significance to ensuring the safety of tailings ponds, protecting the environment and promoting sustainable development. Traditional methods for identifying water holes on the tailings pond beach mainly include manual inspection, remote sensing image analysis and ground measurement. Although these methods can identify water holes to a certain extent, they have the disadvantages of low efficiency, limited resolution and accuracy, high cost, long cycle and difficulty in real-time monitoring. Summary of the invention

[0003] The present application provides a method and system for identifying water puddles on the beach of a tailings pond based on point cloud data, so as to at least solve the technical problems of low efficiency, limited resolution and accuracy, high cost, long cycle, and difficulty in real-time monitoring.

[0004] The first embodiment of the present application proposes a method for identifying a puddle on a tailings pond beach based on point cloud data, the method comprising:

[0005] Acquire point cloud data of the tailings pond to be identified, and preprocess the point cloud data to obtain preprocessed point cloud data;

[0006] Generate a three-dimensional point cloud model according to the preprocessed point cloud data, and determine the feature information of each face voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model;

[0007] Inputting the characteristic information of each face voxel into a pre-trained puddle recognition model to obtain a puddle recognition result of each face voxel;

[0008] The location of the puddle on the tailings pond beach is determined based on the puddle identification results of the voxels on each face.

[0009] Preferably, the step of obtaining point cloud data of the tailings pond to be identified includes:

[0010] Using an aerial survey multi-rotor drone to collect point cloud data of the tailings pond to be identified;

[0011] The multi-rotor drone is equipped with a laser radar, a high-definition camera, an RTK module and an inertial measurement unit.

[0012] Furthermore, the preprocessing of the point cloud data to obtain preprocessed point cloud data includes:

[0013] The point cloud data is processed by eliminating erroneous readings, removing flying spots, and filtering noise, and then the point cloud data after the processing of eliminating erroneous readings, removing flying spots, and filtering noise is converted into a rational coordinate system to obtain pre-processed point cloud data.

[0014] Furthermore, determining the feature information of each voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model includes:

[0015] Generate a three-dimensional point cloud model based on the preprocessed point cloud data;

[0016] Projecting the three-dimensional point cloud model onto a two-dimensional plane to obtain a two-dimensional point cloud data set;

[0017] The point cloud data set is divided to obtain each face voxel, and feature information of each face voxel is determined.

[0018] Furthermore, the feature information of the face voxel includes:

[0019] Local density, surface normal, curvature, depth, surface roughness, color distribution.

[0020] Furthermore, the training process of the puddle recognition model includes:

[0021] Acquire feature information of each face voxel of the tailings pond that has been identified and has a water pit and actual recognition results corresponding to the feature information of each face voxel, wherein the recognition results include: water pits and non-water pits;

[0022] The characteristic information of each face voxel is used as input, the actual recognition result corresponding to the characteristic information of each face voxel is used as output, and the SMO algorithm is used to optimize and train the initial SVM classifier to obtain a trained puddle recognition model.

[0023] Furthermore, the determining of the location of the puddle on the beach of the tailings pond based on the puddle identification results of the voxels on each face includes:

[0024] Obtain the three-dimensional coordinates of each point in each voxel of each face of the puddle identified as a water puddle;

[0025] The location of the water pit on the tailings pond beach is determined based on the three-dimensional coordinates of each point in each voxel of the water pit that is the recognition result.

[0026] The second embodiment of the present application proposes a tailings pond beach puddle identification system based on point cloud data, including:

[0027] An acquisition module is used to acquire point cloud data of the tailings pond to be identified, and preprocess the point cloud data to obtain preprocessed point cloud data;

[0028] A first determination module is used to generate a three-dimensional point cloud model according to the preprocessed point cloud data, and determine the feature information of each face voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model;

[0029] An identification module, used to input the feature information of each face voxel into a pre-trained puddle identification model to obtain a puddle identification result of each face voxel;

[0030] The second determination module is used to determine the location of the water puddle on the beach of the tailings pond based on the water puddle identification results of the voxels on each face.

[0031] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in the first aspect is implemented.

[0032] A fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the method described in the first aspect of the present application.

[0033] The technical solution provided by the embodiments of the present application brings at least the following beneficial effects:

[0034] The present application proposes a method and system for identifying puddles on the beach of a tailings pond based on point cloud data, the method comprising: obtaining point cloud data of the tailings pond to be identified, and preprocessing the point cloud data to obtain preprocessed point cloud data; generating a three-dimensional point cloud model according to the preprocessed point cloud data, and determining the feature information of each face voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model; inputting the feature information of each face voxel into a pre-trained puddle recognition model to obtain the puddle recognition result of each face voxel; and determining the location of the puddle on the beach of the tailings pond based on the puddle recognition result of each face voxel. The technical solution proposed in the present application can quickly and effectively identify puddle on the beach of a tailings pond.

[0035] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through the practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0037] Figure 1 A flowchart of a method for identifying a puddle on a tailings pond beach based on point cloud data according to an embodiment of the present application;

[0038] Figure 2 A flowchart of point cloud data collection provided according to an embodiment of the present application;

[0039] Figure 3 A schematic diagram of a linear classifier classification according to an embodiment of the present application

[0040] Figure 4 The present invention is a structural diagram of a tailings pond beach puddle identification system based on point cloud data according to an embodiment of the present application. DETAILED DESCRIPTION

[0041] Embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0042] The present application proposes a method and system for identifying puddles on the beach of a tailings pond based on point cloud data, the method comprising: obtaining point cloud data of the tailings pond to be identified, and preprocessing the point cloud data to obtain preprocessed point cloud data; generating a three-dimensional point cloud model according to the preprocessed point cloud data, and determining the feature information of each face voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model; inputting the feature information of each face voxel into a pre-trained puddle recognition model to obtain the puddle recognition result of each face voxel; and determining the location of the puddle on the beach of the tailings pond based on the puddle recognition result of each face voxel. The technical solution proposed in the present application can quickly and effectively identify puddle on the beach of a tailings pond.

[0043] The following describes a method and system for identifying water puddles on the beach of a tailings pond based on point cloud data in an embodiment of the present application with reference to the accompanying drawings.

[0044] Embodiment 1

[0045] Figure 1The flowchart of a method for identifying a water hole on a tailings pond beach based on point cloud data is provided according to an embodiment of the present application. Figure 1 As shown, the method includes:

[0046] Step 1: Obtain point cloud data of the tailings pond to be identified, and preprocess the point cloud data to obtain preprocessed point cloud data.

[0047] In the embodiment of the present disclosure, the step of obtaining point cloud data of the tailings pond to be identified includes:

[0048] Using an aerial survey multi-rotor drone to collect point cloud data of the tailings pond to be identified;

[0049] The multi-rotor drone is equipped with a laser radar, a high-definition camera, an RTK module and an inertial measurement unit.

[0050] It should be noted that if Figure 2 As shown in the figure, drones are one of the most commonly used devices for point cloud data collection in tailings ponds. They have flexible flight capabilities, can cover a wide area, and can collect data at different heights and angles. To collect point cloud data for tailings ponds, an aerial survey multi-rotor drone is used, equipped with high-precision laser radar, high-definition camera, high-precision RTK module, IMU (inertial measurement unit) and other equipment. High-precision laser radar can quickly measure the three-dimensional coordinates of a large number of points. It determines the position of the point by emitting laser beams and measuring the reflected signals. When the drone performs the measurement task, the picture taken by the high-definition camera mounted on the drone can be viewed in real time on the drone management platform or remote control, and the data collection situation can be timely understood. Through the image recognition algorithm of the high-definition camera, the drone can autonomously avoid obstacles when encountering obstacles. The RTK module can provide real-time centimeter-level positioning data for the drone, reducing errors caused by factors such as signal delay and multipath effect. The IMU can provide the attitude information of the drone, including pitch angle, roll angle and yaw angle, and can detect and compensate for the movement of the drone, such as bumps, rotations, etc., so as to reduce the impact of these movements on the quality of the point cloud data. In this embodiment, the collected data and the LS format data generated by the image processing software can also be stored in the database.

[0051] Furthermore, the preprocessing of the point cloud data to obtain preprocessed point cloud data includes:

[0052] The point cloud data is processed by eliminating erroneous readings, removing flying spots, and filtering noise, and then the point cloud data after the processing of eliminating erroneous readings, removing flying spots, and filtering noise is converted into a rational coordinate system to obtain pre-processed point cloud data.

[0053] It should be noted that point cloud data preprocessing is a key step in the tailings pond beach puddle identification algorithm, which directly affects the accuracy of subsequent data analysis and puddle detection. In order to control data quality, a series of operations such as removing erroneous readings, removing flying points, and filtering noise will be performed on the data. During the flight, the drone may collect data from multiple perspectives. These data need to be aligned to the same coordinate system through a registration algorithm. Generally, the iterative closest point (ICP) algorithm is used to accurately estimate the relative positions between multiple point cloud data sets. The point cloud data is then converted from the drone coordinate system to the geographic coordinate system, and the accuracy of the geographic reference of the point cloud data is ensured by combining the ground control points (GCP) and high-precision RTK module data.

[0054] Step 2: Generate a three-dimensional point cloud model according to the preprocessed point cloud data, and determine the feature information of each face voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model.

[0055] In the disclosed embodiment, determining the feature information of each voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model includes:

[0056] Generate a three-dimensional point cloud model based on the preprocessed point cloud data;

[0057] Projecting the three-dimensional point cloud model onto a two-dimensional plane to obtain a two-dimensional point cloud data set;

[0058] The point cloud data set is divided to obtain each face voxel, and feature information of each face voxel is determined.

[0059] The feature information of the face voxel includes:

[0060] Local density, surface normal, curvature, depth, surface roughness, color distribution.

[0061] It should be noted that after the point cloud data is preprocessed, the data is imported into the model making software to generate a three-dimensional point cloud model.

[0062] The purpose of point cloud feature extraction is to identify the puddles on the tailings pond beach. To achieve this goal, we can use the method of surface voxel mapping. First, we need to project the point cloud data onto a two-dimensional plane. This can be achieved by projecting each point in the point cloud onto a horizontal plane. In this way, we can get a two-dimensional point cloud dataset.

[0063] Next, the 2D point cloud dataset is divided into many small voxels. Each voxel records some information, such as the number of points in the voxel, the average value, etc. This information can be used to analyze the characteristics of the puddle.

[0064] It should be noted that local density: the number of points in each voxel is calculated as a basic feature of the voxel. Voxels in the puddle area usually have a higher local density.

[0065] Surface normal: The surface normal of a voxel is estimated by the distribution of points within the voxel. For a puddle, its surface normal is usually parallel to the ground.

[0066] Curvature: Based on the change in curvature of the points in the voxel, circular or nearly circular puddle features can be identified. The curvature of the edge of the puddle is larger, while the bottom is relatively flat.

[0067] Depth: Determine the depth of a puddle by comparing the relative height of the bottom voxel of the puddle with other voxels. The depth of a puddle can serve as an important distinguishing feature.

[0068] Surface roughness: Analyzing the point distribution characteristics of the puddle surface, such as the surface roughness, can assist in identifying puddles and non-puddle areas.

[0069] Color distribution: If the point cloud data contains color information, features can be extracted by analyzing the color distribution of the puddle area.

[0070] Step 3: Input the feature information of each face voxel into a pre-trained puddle recognition model to obtain a puddle recognition result of each face voxel.

[0071] It should be noted that the training process of the puddle recognition model includes:

[0072] Acquire feature information of each face voxel of the tailings pond that has been identified and has a water pit and actual recognition results corresponding to the feature information of each face voxel, wherein the recognition results include: water pits and non-water pits;

[0073] The characteristic information of each face voxel is used as input, the actual recognition result corresponding to the characteristic information of each face voxel is used as output, and the SMO algorithm is used to optimize and train the initial SVM classifier to obtain a trained puddle recognition model.

[0074] It should be noted that, in the detection of water puddles, the characteristic information of the extracted face voxels can be used to identify the water puddles on the beach of the tailings pond through a classification algorithm. In this embodiment, a classification algorithm based on machine learning, support vector machine (SVM), is used. SVM is an effective two-class classification algorithm that separates data by finding a decision boundary (or "hyperplane") that can maximize the "interval" between two categories.

[0075] Support vector machine (SVM) is a type of linear classifier that aims to separate different data points through a decision boundary. In a two-dimensional plane, this decision boundary is a straight line; in three-dimensional space, it is a plane, and so on. In N-dimensional space, this decision boundary is called a "hyperplane". There are red and blue points on a two-dimensional plane. A linear classifier (such as SVM) will look for a straight line that separates the green and blue points as much as possible, such as Figure 3 shown.

[0076] First, we need to take the extracted puddle features and non-puddle sample data as input, output (the output is a puddle or non-puddle), build a training set, and train an SVM classifier. This classifier can learn how to distinguish between puddle and non-puddle areas based on these features.

[0077] The support vector machine is to find the partitioning hyperplane with the "maximum margin", that is, to establish the following basic model:

[0078]

[0079] Where ~γ is the geometric interval, is the maximum geometric interval, x i is the i-th feature information, y i is the identification category corresponding to the i-th feature information, ω is a vector perpendicular to the hyperplane, defined as the normal vector, is the function interval corresponding to the i-th feature information, To define the function interval threshold, b is the bias term, which determines the vertical distance between the hyperplane and the origin. |ω T x i +b| is the distance from point x to the hyperplane, y i (ω T x i +b) indicates the specified function interval.

[0080] The minimum function interval is set to 1, and the original problem is transformed into a convex optimization problem. In order to facilitate the solution and introduce the kernel function, the original problem is dualized. By adding a Lagrange multiplier α to each constraint condition, the Lagrange function is constructed:

[0081]

[0082] Where w is the weight vector, which contains the parameters of the support vector machine model.

[0083] α is the Lagrange multiplier, which is a vector whose elements α i For each training sample x i .y iis the identification category or label corresponding to the i-th feature information. It is a binary variable, usually taking the value of +1 or -1, indicating that the sample belongs to the positive class or the negative class. i is the feature vector of the i-th sample, which contains all the features of the sample. i is the total number of samples in the training set, ||w|| is the second-order norm of w, which is a concept similar to the modulus to represent length. According to the duality of Lagrange and after a certain solution, the original problem is transformed into:

[0084]

[0085] Then use the KKT condition to solve it, which is a necessary and sufficient condition for a nonlinear programming problem to have an optimal solution. The original problem has been transformed into a dual problem by satisfying the KKT condition. Solving this dual learning problem is divided into three steps: first, minimize L(w,b,a) with respect to w and b, then maximize α, and finally use the SMO algorithm to solve the Lagrange multiplier in the dual problem.

[0086] Sequential Minimal Optimization (SMO) algorithm:

[0087] The core of SVM is to find an optimal hyperplane that maximizes the interval between data points of different categories. This optimal hyperplane can be obtained by solving a convex quadratic programming problem. However, when the training set is large, the traditional quadratic programming method has huge computational overhead and low efficiency. The SMO algorithm decomposes the original quadratic programming problem and transforms it into a series of minimization problems, each of which involves two Lagrange multipliers, thereby significantly reducing the complexity of the calculation.

[0088] The steps of the SMO algorithm can be summarized into the following main parts:

[0089] 1. Initialization

[0090] First, the SMO algorithm needs to initialize a series of parameters, including Lagrange multiplier vectors, hyperplane parameters, and bias terms (b). Usually, the vectors are initialized to zero or initialized according to specific heuristic rules.

[0091] 2. Select the working set

[0092] The core idea of ​​the SMO algorithm is to iteratively select a pair of Lagrange multipliers and optimize them. This pair of multipliers is called a working set. There are many strategies for selecting working sets, the most common ones are heuristic selection, random selection, and interval-based selection.

[0093] 3. Solving subproblems

[0094] Once the working sets and are chosen, the SMO algorithm fixes all other values ​​and then solves the subproblem involving only and. This subproblem is a quadratic programming problem with only two variables and can be solved analytically.

[0095] 4. Update parameters

[0096] Based on the solved subproblems, the SMO algorithm will update the values ​​of and and then recalculate and based on the updated values.

[0097] 5. Iteration

[0098] Repeat steps 2 to 4 until a stopping condition is met, for example, until the change in value is less than a certain threshold, or until a predetermined number of iterations is reached.

[0099] Characteristics of the SMO algorithm: 1. High efficiency. Since the SMO algorithm only updates two Lagrange multipliers each time, its computational complexity is greatly reduced. This makes the SMO algorithm particularly suitable for large-scale SVM training problems. 2. Simplicity. The steps of the SMO algorithm are relatively simple and easy to implement. It does not require complex mathematical libraries or optimization tools, only some basic linear algebra operations. 3. Convergence. The SMO algorithm has good convergence. Although it is an iterative algorithm, it can usually obtain satisfactory results within a limited number of iterations. 4. Flexibility. The SMO algorithm can be easily combined with other optimization techniques, such as the use of kernel functions, so that it can handle nonlinear SVM problems.

[0100] Step 4: Determine the location of the water puddle on the tailings pond beach based on the water puddle identification results of the voxels on each face.

[0101] In the embodiment of the present disclosure, step 4 specifically includes:

[0102] Obtain the three-dimensional coordinates of each point in each voxel of each face of the puddle identified as a water puddle;

[0103] The location of the water pit on the tailings pond beach is determined based on the three-dimensional coordinates of each point in each voxel of the water pit that is the recognition result.

[0104] It should be noted that the method for identifying water puddles on the beach of a tailings pond based on point cloud data proposed in this embodiment has the following advantages: 1. High precision. Point cloud data can provide accurate three-dimensional coordinates, which helps to accurately identify the location and range of water puddles. 2. High resolution. Point cloud data can capture subtle changes in the surface, which helps to discover water puddles formed early. 3. Real-time monitoring. Point cloud data can be quickly acquired using equipment such as drones and mobile measurement systems to achieve real-time monitoring. 4. Point cloud data has the characteristics of high precision and high resolution, and can provide rich surface information, providing a new technical means for the identification of water puddles on the beach of tailings ponds.

[0105] In summary, this embodiment proposes a method for identifying water puddles on the beach of a tailings pond based on point cloud data. The point cloud-based data processing technology can provide high-precision three-dimensional information, while the voxel technology can be used to quickly process and analyze these data. Combining these two technologies can quickly and effectively identify water puddles on the beach of a tailings pond.

[0106] Embodiment 2

[0107] Figure 4 FIG. 1 is a structural diagram of a tailings pond beach surface puddle identification system based on point cloud data according to an embodiment of the present application. Figure 4 As shown, the system comprises:

[0108] The acquisition module 100 is used to acquire the point cloud data of the tailings pond to be identified, and pre-process the point cloud data to obtain the pre-processed point cloud data;

[0109] A first determination module 200 is used to generate a three-dimensional point cloud model according to the preprocessed point cloud data, and determine the feature information of each face voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model;

[0110] The recognition module 300 is used to input the feature information of each face voxel into a pre-trained puddle recognition model to obtain a puddle recognition result of each face voxel;

[0111] The second determination module 400 is used to determine the location of the water puddle on the beach of the tailings pond based on the water puddle identification results of the voxels on each face.

[0112] In the embodiment of the present disclosure, the acquisition module 100 is further used for:

[0113] Using an aerial survey multi-rotor drone to collect point cloud data of the tailings pond to be identified;

[0114] The multi-rotor drone is equipped with a laser radar, a high-definition camera, an RTK module and an inertial measurement unit.

[0115] Furthermore, the acquisition module 100 is also used for:

[0116] The point cloud data is processed by eliminating erroneous readings, removing flying spots, and filtering noise, and then the point cloud data after the processing of eliminating erroneous readings, removing flying spots, and filtering noise is converted into a rational coordinate system to obtain pre-processed point cloud data.

[0117] In the embodiment of the present disclosure, the first determining module 200 is further configured to:

[0118] Generate a three-dimensional point cloud model based on the preprocessed point cloud data;

[0119] Projecting the three-dimensional point cloud model onto a two-dimensional plane to obtain a two-dimensional point cloud data set;

[0120] The point cloud data set is divided to obtain each face voxel, and feature information of each face voxel is determined.

[0121] The feature information of the face voxel includes:

[0122] Local density, surface normal, curvature, depth, surface roughness, color distribution.

[0123] It should be noted that the identification module 300 is also used to: train the puddle recognition model to obtain a pre-trained puddle recognition model;

[0124] The training process of the puddle recognition model includes:

[0125] Acquire feature information of each face voxel of the tailings pond that has been identified and has a water pit and actual recognition results corresponding to the feature information of each face voxel, wherein the recognition results include: water pits and non-water pits;

[0126] The characteristic information of each face voxel is used as input, the actual recognition result corresponding to the characteristic information of each face voxel is used as output, and the SMO algorithm is used to optimize and train the initial SVM classifier to obtain a trained puddle recognition model.

[0127] In the embodiment of the present disclosure, the second determining module 400 is further configured to:

[0128] Obtain the three-dimensional coordinates of each point in each voxel of each face of the puddle identified as a water puddle;

[0129] The location of the water pit on the tailings pond beach is determined based on the three-dimensional coordinates of each point in each voxel of the water pit that is the recognition result.

[0130] In summary, the tailings pond beach surface water hole identification system based on point cloud data proposed in this embodiment can quickly and effectively identify the tailings pond beach surface water holes.

[0131] Embodiment 3

[0132] In order to implement the above embodiments, the present disclosure further proposes an electronic device, including: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method described in Embodiment 1 is implemented.

[0133] Embodiment 4

[0134] In order to implement the above embodiments, the present disclosure further proposes a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the method described in the first embodiment is implemented.

[0135] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" etc. means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner. In addition, those skilled in the art may combine and combine the different embodiments or examples described in this specification and the features of the different embodiments or examples, without contradiction.

[0136] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by technicians in the technical field to which the embodiments of the present application belong.

[0137] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A method for identifying water pits on the tailings pond beach based on point cloud data, characterized in that: The method comprises: Acquire point cloud data of the tailings pond to be identified, and preprocess the point cloud data to obtain preprocessed point cloud data; Generate a three-dimensional point cloud model according to the preprocessed point cloud data, and determine the feature information of each face voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model; Inputting the feature information of each face voxel into a pre-trained puddle recognition model to obtain a puddle recognition result of each face voxel; The location of the puddle on the tailings pond beach is determined based on the puddle identification results of the voxels on each face.

2. The method according to claim 1, characterized in that The step of obtaining point cloud data of the tailings pond to be identified includes: Using an aerial survey multi-rotor drone to collect point cloud data of the tailings pond to be identified; The multi-rotor drone is equipped with a laser radar, a high-definition camera, an RTK module and an inertial measurement unit.

3. The method according to claim 2, characterized in that The preprocessing of the point cloud data to obtain preprocessed point cloud data includes: The point cloud data is processed by eliminating erroneous readings, removing flying spots, and filtering noise, and then the point cloud data after the processing of eliminating erroneous readings, removing flying spots, and filtering noise is converted into a rational coordinate system to obtain pre-processed point cloud data.

4. The method according to claim 3, characterized in that The determining, based on the three-dimensional point cloud model, feature information of each face voxel corresponding to the tailings pond to be identified comprises: Generate a three-dimensional point cloud model based on the preprocessed point cloud data; Projecting the three-dimensional point cloud model onto a two-dimensional plane to obtain a two-dimensional point cloud data set; The point cloud data set is divided to obtain each face voxel, and feature information of each face voxel is determined.

5. The method according to claim 4, characterized in that The feature information of the face voxel includes: Local density, surface normal, curvature, depth, surface roughness, color distribution.

6. The method according to claim 5, characterized in that The training process of the puddle recognition model includes: Acquire feature information of each face voxel of the tailings pond that has been identified and has a water pit and actual recognition results corresponding to the feature information of each face voxel, wherein the recognition results include: water pits and non-water pits; The characteristic information of each face voxel is used as input, the actual recognition result corresponding to the characteristic information of each face voxel is used as output, and the SMO algorithm is used to optimize and train the initial SVM classifier to obtain a trained puddle recognition model.

7. The method according to claim 5, characterized in that The determining of the location of the puddle on the beach of the tailings pond based on the puddle identification results of the voxels on each face includes: Obtain the three-dimensional coordinates of each point in each voxel of each face of the puddle identified as a water puddle; The location of the water pit on the tailings pond beach is determined based on the three-dimensional coordinates of each point in each voxel of the water pit that is the recognition result.

8. A tailings pond beach puddle identification system based on point cloud data, characterized in that: The system comprises: An acquisition module is used to acquire point cloud data of the tailings pond to be identified, and preprocess the point cloud data to obtain preprocessed point cloud data; A first determination module is used to generate a three-dimensional point cloud model according to the preprocessed point cloud data, and determine the feature information of each face voxel corresponding to the tailings pond to be identified based on the three-dimensional point cloud model; An identification module, used to input the feature information of each face voxel into a pre-trained puddle identification model to obtain a puddle identification result of each face voxel; The second determination module is used to determine the location of the puddle on the beach of the tailings pond based on the puddle identification results of the voxels on each face.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, the method according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the method according to any one of claims 1 to 7 is implemented.

Images