Power line target extraction method based on random forest and one-class classification strategy

Abstract

The application provides a power line target extraction method based on a random forest and a one-class classification strategy. The method comprises the following steps: acquiring airborne LiDAR point cloud data and multispectral orthographic images collected by a UAV multi-sensor platform; acquiring a feature set according to the airborne LiDAR point cloud data; training a random forest two-class classifier by using the airborne LiDAR point cloud data, the feature set and the multispectral orthographic images according to a random forest algorithm and a one-class classification strategy, to obtain a PBL-random forest model; acquiring to-be-classified airborne point cloud data; and classifying the to-be-classified airborne point cloud data by using the PBL-random forest model to generate a vector result of the power line. The method solves the problem that the model in the prior art cannot accurately predict the actual existence of the power line.

Description

Technical Field

[0001] This application relates to the technical field of power line target extraction, and more specifically, to a power line target extraction method based on random forest and a single-class classification strategy, a power line target extraction device based on random forest and a single-class classification strategy, a computer-readable storage medium, and an electronic device. Background Technology

[0002] With the advancement of urbanization, the coverage of power lines in cities is gradually expanding, and the threat posed by tree growth to power line safety is becoming increasingly significant. As trees grow, their branches gradually approach power lines, causing short circuits, grounding faults, and other problems, seriously affecting power supply and potentially threatening human safety. The maintenance of power facilities also becomes correspondingly more difficult. Therefore, accurately identifying power lines during urbanization is of great importance for rationally planning power line corridors, regularly maintaining power facilities, promptly identifying and addressing tree-related hazards, improving the security and reliability of power supply, and promoting sustainable urban development.

[0003] Due to the needs of unmanned aerial vehicle (UAV) technology and the field of Earth observation, airborne lidar (LiDAR) technology combines the advantages of high-precision ranging and hyperspectral imaging of lidar, enabling high-precision and high-resolution monitoring of ground targets. It has become a particularly important technology for acquiring accurate three-dimensional (3D) spatial information. Point cloud data acquired by airborne lidar (LiDAR) technology possesses high-precision, high-resolution, and high-dimensional geometric information, which can intuitively represent the shape, surface, and texture of objects in space. Point cloud data can not only obtain high-precision three-dimensional structural information of power lines through high-density three-dimensional point sets, but also, in complex environments such as dense forests or urban high-rise buildings where power lines are often obscured by vegetation and buildings, point cloud data can directly acquire three-dimensional spatial information through laser scanning. This method is less susceptible to environmental interference and can more accurately identify obscured power lines.

[0004] Most existing point cloud datasets primarily focus on the presence information of target objects, i.e., the point cloud data itself, while often neglecting the importance of background or absence information. This imbalance in data composition makes it difficult for models to accurately predict the actual presence of power lines because the models lack learning of background features when objects are absent. Traditional statistical methods, such as the maximum entropy method, typically require both presence and absence data for model training. However, in the actual acquisition of point cloud data, obtaining such paired data is often impractical, thus limiting the effective application of these traditional methods in point cloud classification tasks. Even if some methods can handle point clouds containing only presence data, they often only output a relative suitability index, failing to provide a probability estimate of the actual presence of objects. This lack of probabilistic interpretability makes comparisons between different models difficult and affects the interpretability of prediction results in practical applications. Summary of the Invention

[0005] The main objective of this application is to provide a power line target extraction method based on random forest and a single-class classification strategy, a power line target extraction device based on random forest and a single-class classification strategy, a computer-readable storage medium, and an electronic device, so as to at least solve the problem that existing models are difficult to accurately predict the actual existence of power lines.

[0006] To achieve the above objectives, according to one aspect of this application, a method for extracting power line targets based on random forest and a single-class classification strategy is provided, comprising: acquiring airborne LiDAR point cloud data and multispectral orthophotos collected by a UAV multi-sensor platform; acquiring a feature set based on the airborne LiDAR point cloud data, wherein the feature set includes height-based features, feature-based features, echo-based features, and multi-scale features, the multi-scale features including height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residuals; training a random forest binary classifier using the airborne LiDAR point cloud data, the feature set, and the multispectral orthophotos according to a random forest algorithm and a single-class classification strategy to obtain a PBL-Random Forest model; acquiring airborne point cloud data to be classified, and using the PBL-Random Forest model to classify the airborne point cloud data to be classified as power lines to generate vector results of power lines.

[0007] Optionally, obtaining height-based features from the airborne LiDAR point cloud data includes: identifying ground points in the airborne LiDAR point cloud data using the progressive triangular irregular network densification filtering algorithm in LAStools software to obtain first echo data and ground echo data; performing interpolation processing on the first echo data and the ground echo data respectively using the ordinary Kriging method to obtain a digital surface model and a digital elevation model; determining a normalized digital surface model based on the digital surface model and the digital elevation model; defining a spherical neighborhood for each point on the normalized digital surface model; determining the height variance of all points in the spherical neighborhood; and allocating the height variance to the center point of the spherical neighborhood to obtain the height-based features.

[0008] Optionally, obtaining feature-based features from the airborne LiDAR point cloud data includes: defining a spherical neighborhood centered on each point cloud in the airborne LiDAR point cloud data; determining the local covariance matrix of each point cloud based on the spherical neighborhood; performing eigenvalue decomposition on the local covariance matrix to obtain multiple eigenvectors and multiple eigenvalues; and determining the relevant features of the eigenvectors and the relevant features of the eigenvalues ​​to obtain the feature-based features, wherein the relevant features of the eigenvectors include linearity, flatness, sphericity, and curvature variation, and the relevant features of the eigenvalues ​​include perpendicularity and plane fitting residuals.

[0009] Optionally, the echo-based features include the total number of laser pulse echoes and the echo ratio. Obtaining the echo-based features based on the airborne LiDAR point cloud data includes: determining the value of the laser pulse echo received by each point cloud in the airborne LiDAR point cloud data to obtain the total number of laser pulse echoes; determining the echo number of each laser pulse echo; and determining the echo ratio based on the total number of laser pulse echoes and the echo number.

[0010] Optionally, obtaining multi-scale features based on the airborne LiDAR point cloud data includes: obtaining a preset number of preset scales; determining spherical neighborhoods with radii of each preset scale based on the airborne LiDAR point cloud data; and determining the local geometric features of the spherical neighborhoods corresponding to each preset scale to obtain the multi-scale features.

[0011] Optionally, the random forest binary classifier is trained using the airborne LiDAR point cloud data, the feature set, and the multispectral orthophoto, including: classifying the airborne LiDAR point cloud data to obtain labeled samples and unlabeled samples, wherein the labeled samples represent samples with labeled power lines, and the unlabeled samples represent samples without labeled power lines; training the forest binary classifier using the labeled and unlabeled samples and the feature set, and verifying the output results of the random forest binary classifier using the multispectral orthophoto during the training process.

[0012] Optionally, after generating the vector result of the power line, the method further includes: determining the position and shape of the power line based on the vector result of the power line, so as to realize the inspection work of the power line.

[0013] According to another aspect of this application, a power line target extraction device based on random forest and a single-class classification strategy is provided, comprising: a first acquisition unit for acquiring airborne LiDAR point cloud data and multispectral orthophotos collected by a UAV multi-sensor platform; a second acquisition unit for acquiring a feature set based on the airborne LiDAR point cloud data, wherein the feature set includes height-based features, feature-based features, echo-based features, and multi-scale features, the multi-scale features including height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual; a training unit for training a random forest binary classifier using the airborne LiDAR point cloud data, the feature set, and the multispectral orthophotos according to a random forest algorithm and a single-class classification strategy to obtain a PBL-Random Forest model; and a classification unit for acquiring airborne point cloud data to be classified, and using the PBL-Random Forest model to classify the airborne point cloud data to be classified as power lines to generate vector results of power lines.

[0014] According to another aspect of this application, a computer-readable storage medium is provided, the computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform any of the described power line target extraction methods based on random forest and a class of classification strategies.

[0015] According to another aspect of this application, an electronic device is provided, comprising: one or more processors, a memory, and one or more programs, wherein the one or more programs are stored in the memory and configured to be executed by the one or more processors, the one or more programs including methods for performing any of the described power line target extraction methods based on random forest and a class classification strategy.

[0016] This application utilizes the technical solution to acquire airborne LiDAR point cloud data and multispectral orthophotos collected by a UAV multi-sensor platform. A feature set is obtained from the airborne LiDAR point cloud data, including altitude-based features, feature-based features, echo-based features, and multi-scale features. The multi-scale features include altitude variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residuals. A random forest binary classifier is trained using the airborne LiDAR point cloud data, the feature set, and the multispectral orthophotos, resulting in a PBL-Random Forest model. Airborne point cloud data to be classified is acquired, and the PBL-Random Forest model is used to classify power lines, generating vector results for the power lines. By combining the random forest method and a single-class classification strategy with airborne LiDAR point cloud data and the feature set to obtain the PBL-Random Forest model, and then using the PBL-Random Forest model to predict and identify power lines, the accuracy of power line identification is improved, solving the problem that existing models struggle to accurately predict the actual existence of power lines. Attached Figure Description

[0017] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings:

[0018] Figure 1 A hardware structure block diagram of a mobile terminal that performs a power line target extraction method based on random forest and a classification strategy according to an embodiment of this application is shown.

[0019] Figure 2 A flowchart illustrating a power line target extraction method based on random forest and a classification strategy according to an embodiment of this application is shown.

[0020] Figure 3 A flowchart illustrating a specific power line target extraction method based on random forest and a classification strategy according to an embodiment of this application is shown.

[0021] Figure 4 A flowchart illustrating the calculation of a feature set provided according to an embodiment of this application is shown;

[0022] Figure 5 A flowchart illustrating the generation of a PBL-Random Forest model according to an embodiment of this application is shown;

[0023] Figure 6A flowchart illustrating power line classification using the PBL-Random Forest model provided according to embodiments of this application is shown;

[0024] Figure 7 A structural block diagram of a power line target extraction device based on random forest and a classification strategy according to an embodiment of this application is shown.

[0025] The above figures include the following reference numerals:

[0026] 102. Processor; 104. Memory; 106. Transmission device; 108. Input / output device. Detailed Implementation

[0027] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0028] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0029] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0030] As described in the background section, existing models struggle to accurately predict the actual existence of power lines. To address this issue, embodiments of this application provide a power line target extraction method based on random forest and a classification strategy, a power line target extraction device based on random forest and a classification strategy, a computer-readable storage medium, and an electronic device.

[0031] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention.

[0032] The methods and embodiments provided in this application can be executed on a mobile terminal, computer terminal, or similar computing device. Taking running on a mobile terminal as an example, Figure 1 This is a hardware structure block diagram of a mobile terminal for a power line target extraction method based on random forest and a classification strategy according to an embodiment of the present invention. Figure 1 As shown, a mobile terminal may include one or more ( Figure 1 Only one is shown in the diagram. A processor 102 (which may include, but is not limited to, a microprocessor MCU or a programmable logic device FPGA, etc.) and a memory 104 for storing data are also shown. The mobile terminal may further include a transmission device 106 for communication functions and an input / output device 108. Those skilled in the art will understand that... Figure 1 The structure shown is for illustrative purposes only and does not limit the structure of the mobile terminal described above. For example, the mobile terminal may also include components that are more... Figure 1 The more or fewer components shown, or having the same Figure 1 The different configurations shown.

[0033] Memory 104 can be used to store computer programs, such as application software programs and modules, like the computer program corresponding to the power line target extraction method based on random forest and a classification strategy in this embodiment of the invention. Processor 102 executes various functional applications and data processing by running the computer program stored in memory 104, thereby implementing the above-described method. Memory 104 may include high-speed random access memory and may also include non-volatile memory, such as one or more magnetic storage devices, flash memory, or other non-volatile solid-state memory. In some instances, memory 104 may further include memory remotely located relative to processor 102, and these remote memories can be connected to the mobile terminal via a network. Examples of the aforementioned networks include, but are not limited to, the Internet, corporate intranets, local area networks, mobile communication networks, and combinations thereof. Transmission device 106 is used to receive or send data via a network. Specific examples of the aforementioned networks may include wireless networks provided by the mobile terminal's communication provider. In one example, transmission device 106 includes a Network Interface Controller (NIC), which can be connected to other network devices via a base station to communicate with the Internet. In one example, the transmission device 106 may be a radio frequency (RF) module, which is used to communicate with the Internet wirelessly.

[0034] This embodiment provides a power line target extraction method based on random forest and a classification strategy, which runs on a mobile terminal, computer terminal or similar computing device. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Also, although the logical order is shown in the flowchart, in some cases the steps shown or described may be executed in a different order than that shown here.

[0035] Figure 2 This is a flowchart of a power line target extraction method based on random forest and a classification strategy according to an embodiment of this application. Figure 2 As shown, the method includes the following steps:

[0036] Step S201: Acquire airborne LiDAR point cloud data and multispectral orthophotos collected by the UAV multi-sensor platform;

[0037] Specifically, airborne sensors are devices installed on UAV platforms for remote sensing during flight. These include common multispectral sensors and lidar point cloud detectors. Airborne sensors can collect multi-source remote sensing data, such as multispectral images and airborne lidar point clouds. In this embodiment, a multispectral sensor is used to acquire high-resolution orthophotos of the Earth, and an airborne lidar detector is used to acquire airborne lidar point clouds at different scales of 1.5 meters, 2.5 meters, and 3.5 meters.

[0038] Step S202: Obtain a feature set based on the airborne LiDAR point cloud data. The feature set includes height-based features, feature-based features, echo-based features, and multi-scale features. The multi-scale features include height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual.

[0039] Feature-based analysis refers to defining one feature or attribute based on a specific characteristic or attribute. This method is commonly used in data analysis and machine learning to identify and extract key feature information. For example, suppose we have a dataset containing students' academic performance and various personal characteristics, such as age, gender, and family background. We can analyze the relationship between these personal characteristics and academic performance to determine which characteristics have a significant impact on academic performance. Then, based on these important features, we can build a model to predict students' academic performance. Through feature-based analysis, we can better understand the relationships between data, discover patterns and regularities hidden behind the data, and thus improve the accuracy and efficiency of data analysis and prediction.

[0040] Step S203: Based on the random forest algorithm and a single-class classification strategy, the random forest binary classifier is trained using the above-mentioned airborne LiDAR point cloud data, the above-mentioned feature set, and the above-mentioned multispectral orthophoto to obtain the PBL-RandomForest model.

[0041] Specifically, the training model is obtained by training the airborne LiDAR point cloud dataset and the four extracted features using the random forest algorithm and a classification strategy. Then, based on the quantitative relationship between the training model and the target model, the model is calibrated using a validation dataset to finally obtain the target model, the PBL-Random Forest model.

[0042] Step S204: Obtain the airborne point cloud data to be classified, and use the above-mentioned PBL-Random Forest model to classify the airborne point cloud data to be classified into power lines, generating vector results of power lines.

[0043] Specifically, vectorization refers to the process of converting raster-format image data into vector data. After classifying airborne point cloud data into power lines using the PBL-Random Forest model, vector lines are generated by connecting adjacent boundary points in the point cloud, ultimately yielding the vector results of the power lines.

[0044] This embodiment acquires airborne LiDAR point cloud data and multispectral orthophotos collected by a UAV multi-sensor platform. A feature set is obtained from the airborne LiDAR point cloud data, including altitude-based features, feature-based features, echo-based features, and multi-scale features. Multi-scale features include altitude variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residuals. A random forest binary classifier is trained using the airborne LiDAR point cloud data, the feature set, and the multispectral orthophotos, resulting in the PBL-RandomForest model. Airborne point cloud data to be classified is acquired, and the PBL-RandomForest model is used to classify power lines, generating vector results for the power lines. By combining the random forest method and a single-class classification strategy with airborne LiDAR point cloud data and the feature set to obtain the PBL-RandomForest model, and using the PBL-RandomForest model to predict and identify power lines, the accuracy of power line identification is improved, solving the problem that existing models struggle to accurately predict the actual existence of power lines.

[0045] In the specific implementation process, height-based features are obtained based on the aforementioned airborne LiDAR point cloud data, including: using the progressive triangular irregular network densification filtering algorithm in LAStools software to identify ground points in the airborne LiDAR point cloud data, obtaining the first echo data and ground echo data; performing interpolation processing on the first echo data and the ground echo data respectively using the ordinary kriging method to obtain a digital surface model and a digital elevation model; determining a normalized digital surface model based on the aforementioned digital surface model and the aforementioned digital elevation model; defining a spherical neighborhood for each point on the aforementioned normalized digital surface model; determining the height variance of all points in the aforementioned spherical neighborhood; and allocating the aforementioned height variance to the center point of the aforementioned spherical neighborhood to obtain the aforementioned height-based features.

[0046] This method first removes discrete outliers from airborne LiDAR point cloud data, and then uses the Progressive Triangular Irregular Network (TIN) densification filtering algorithm in LAStools software to identify ground points. It then uses ordinary kriging to interpolate from the first echo data and ground echo data to obtain a 1-meter resolution digital surface model (DSM) and digital elevation model (DEM). The difference between the DSM and DEM is calculated, and this difference is the normalized digital surface model (nDSM). The normalized digital surface model (nDSM) is output as the normalized height (Hn). A spherical neighborhood is defined for each point on the normalized digital surface model (nDSM), and the variance of the height values ​​of all points within the neighborhood is calculated. This variance is then assigned to the center point to obtain the height variance (Hv).

[0047] Specifically, the feature-based features obtained from the aforementioned airborne LiDAR point cloud data include: defining a spherical neighborhood centered on each point cloud in the airborne LiDAR point cloud data; determining the local covariance matrix of each point cloud based on the spherical neighborhood; performing eigenvalue decomposition on the local covariance matrix to obtain multiple eigenvectors and multiple eigenvalues; and determining the relevant features of the eigenvectors and the relevant features of the eigenvalues ​​to obtain the aforementioned feature-based features. The relevant features of the eigenvectors include linearity, flatness, sphericity, and curvature variation, while the relevant features of the eigenvalues ​​include perpendicularity and plane fitting residuals.

[0048] This method first removes discrete outliers from the airborne LiDAR point cloud data. Since the point cloud data consists of n 3D points, it can be represented as... For each point, define a spherical neighborhood centered at that point, calculate the local covariance matrix of that point, and then calculate the covariance matrix M. cov Perform eigenvalue decomposition to obtain eigenvectors (V1, V2, V3) and eigenvalues ​​(λ1≥λ2≥λ3≥0). Calculate the eigenvalue correlation features and linearity L. λFlatness P λ Spherical degree S λ Given the curvature change k, calculate the eigenvector related features, calculate the angle between the local normal vector and the vertical direction, and obtain the perpendicularity N. z Calculate the residuals of the local plane fitting to obtain the plane fitting residual R. z .

[0049] More specifically, the echo-based features mentioned above include the total number of laser pulse echoes and the echo ratio. Obtaining the echo-based features based on the airborne LiDAR point cloud data includes: determining the value of the laser pulse echo received by each point cloud in the airborne LiDAR point cloud data to obtain the total number of laser pulse echoes; determining the echo number of each of the laser pulse echoes; and determining the echo ratio based on the total number of laser pulse echoes and the echo number.

[0050] This method first removes discrete outliers from the airborne LiDAR point cloud data. For each point in the point cloud, the total number of laser pulse echoes received by that point is calculated to obtain the total number of echoes N. E For each echo, calculate the echo number, i.e., how many times the echo returned, and divide the echo number by the total number of echoes N. E The echo ratio R is obtained. E .

[0051] Furthermore, the acquisition of multi-scale features based on the aforementioned airborne LiDAR point cloud data includes: acquiring a preset number of preset scales; determining spherical neighborhoods with radii of each preset scale based on the aforementioned airborne LiDAR point cloud data; and determining the local geometric features of the spherical neighborhoods corresponding to each preset scale to obtain the aforementioned multi-scale features.

[0052] The preset number of preset dimensions can be set to 1.5m, 2.5m, and 3.5m;

[0053] This method repeatedly calculates the local geometric features of the neighborhood for each scale, including the height variance H. V Linearity L λ Flatness P λ Spherical degree S λ Curvature change k, perpendicularity N z and plane fitting residual R z .

[0054] Furthermore, the random forest binary classifier is trained using the aforementioned airborne LiDAR point cloud data, the aforementioned feature set, and the aforementioned multispectral orthophoto. This includes: classifying the aforementioned airborne LiDAR point cloud data to obtain labeled samples and unlabeled samples, wherein the labeled samples represent samples with labeled power lines, and the unlabeled samples represent samples without labeled power lines; training the random forest binary classifier using the aforementioned labeled and unlabeled samples and the aforementioned feature set; and verifying the output results of the random forest binary classifier using the aforementioned multispectral orthophoto during the training process.

[0055] This method is based on the PBL (Presence and Background Learning) method. It models the probability (denoted as P(y=1|x)) of a single sample x being a positive sample, i.e. the target model P(y=1|x), by classifying the collected airborne LiDAR point cloud dataset according to whether it is labeled or not. Labeled data is regarded as positive samples and unlabeled data is regarded as unlabeled samples.

[0056] First, positive and unlabeled samples are randomly selected in a 1:5 ratio. Features are extracted and normalized to the range [0,1]. Then, 75% of the samples are used as the training dataset and 25% as the validation dataset. Features in the training process have equal weights. Let s=1 represent that x is collected as a positive sample and s=0 represent that x is collected as an unlabeled sample. Then, a random forest binary classifier model P(s=1|x,η=1) is trained using a random forest model, where η=1 indicates the presence of a background policy.

[0057] Specifically, after generating the vector results of the power lines, the method further includes: determining the position and shape of the power lines based on the vector results, so as to realize the inspection work of the power lines.

[0058] This embodiment employs the PBL (Presence and Background Learning) method. This method thoroughly addresses the often-overlooked issue of absence information in traditional point cloud datasets, achieving comprehensiveness and depth in model training by cleverly fusing presence and background data. Furthermore, the PBL method significantly improves the prediction accuracy of object presence states by precisely utilizing these two complementary information sources. Notably, this method also overcomes the limitation of traditional techniques that can only output relative suitability scores; it can accurately estimate the actual probability of an object's existence under specific environmental covariates. This probabilistic prediction output not only greatly enhances the interpretability of the prediction results but also provides a more objective and scientific benchmark for comparing the performance of different models.

[0059] To enable those skilled in the art to better understand the technical solution of this application, the implementation process of the power line target extraction method based on random forest and a classification strategy will be described in detail below with reference to specific embodiments.

[0060] This embodiment relates to a specific method for extracting power line targets based on random forest and a classification strategy, such as... Figure 3 As shown, it includes the following steps:

[0061] Step S1: Acquire multispectral high-resolution orthophotos and airborne LiDAR point cloud data using a UAV multi-sensor platform. Airborne sensors are devices mounted on the UAV platform for remote sensing during flight, including common multispectral sensors and LiDAR point cloud detectors. These sensors can acquire multi-source remote sensing data, such as multispectral images and airborne LiDAR point clouds. In this embodiment, multispectral high-resolution orthophotos are acquired using a multispectral sensor for Earth observation, and airborne LiDAR point clouds at different scales (1.5m, 2.5m, and 3.5m) are acquired using an airborne LiDAR detector for Earth observation.

[0062] Step S2: Calculate the altitude features of the airborne LiDAR point cloud data using its lidar altitude data; evaluate its local spatial distribution to obtain feature-based features; obtain echo features based on the number of laser pulse returns; and repeatedly calculate its local geometric features to obtain multi-scale features.

[0063] like Figure 4 As shown, height-based features are generated using airborne LiDAR point cloud data to obtain normalized height (Hn) and height variance (Hv); feature-based features are generated to obtain eigenvalue-related features and eigenvector-related features; echo-based features are generated to obtain total echo number NE and echo ratio RE; and multi-scale features are generated.

[0064] The extraction of altitude-based features from airborne LiDAR point cloud data specifically includes the following steps:

[0065] 1) Remove discrete outliers from the airborne lidar point cloud;

[0066] 2) Use the Progressive Triangular Irregular Network (TIN) densification filtering algorithm in LAStools software to identify ground points;

[0067] 3) Using the ordinary kriging method, interpolation was performed from the first echo data and the ground echo data to obtain a 1-meter resolution digital surface model (DSM) and digital elevation model (DEM);

[0068] 4) Calculate the difference between the digital surface model (DSM) and the digital elevation model (DEM). This difference is the normalized digital surface model (nDSM).

[0069] 5) Output the normalized digital surface model (nDSM) as the normalized height (Hn);

[0070] 6) Define a spherical neighborhood for each point on the normalized digital surface model (nDSM), calculate the variance of the height values ​​of all points in the neighborhood, and assign the variance values ​​to the center point to obtain the height variance (Hv).

[0071] The specific steps involved in extracting feature-based features from airborne LiDAR point cloud data are as follows:

[0072] 1) Remove discrete outliers from the airborne lidar point cloud;

[0073] 2) Since point cloud data consists of n 3D points, it can be represented as For each point, define a spherical neighborhood centered at that point, and calculate the local covariance matrix M of that point. cov Local covariance matrix M cov The calculation formula is as follows:

[0074]

[0075] In the formula, Let be the mean of the points, and its calculation formula is as follows:

[0076]

[0077] Among them, O i Let be the i-th point in the point cloud dataset, T be the transpose operation, n be the total number of points in the point cloud dataset, and i be the index of the point.

[0078] 3) For the covariance matrix M cov Eigenvalue decomposition yields eigenvectors (V1, V2, V3) and eigenvalues ​​(λ1≥λ2≥λ3≥0). The formulas for calculating eigenvectors (V1, V2, V3) and eigenvalues ​​(λ1≥λ2≥λ3≥0) are as follows:

[0079]

[0080] 4) Calculate the eigenvalue correlation characteristics and linearity L. λ Flatness P λ Spherical degree S λ The formula for calculating the curvature change k is as follows:

[0081]

[0082] In the formula, λ1, λ2 and λ3 represent the three eigenvalues ​​of the covariance matrix.

[0083] 5) Calculate the relevant features of the feature vector, calculate the angle between the local normal vector and the vertical direction, and obtain the perpendicularity N. z Calculate the residuals of the local plane fitting to obtain the plane fitting residual R. z .

[0084] The specific steps involved in extracting echo-based features from airborne LiDAR point cloud data are as follows:

[0085] 1) Remove discrete outliers from the airborne lidar point cloud;

[0086] 2) For each point in the point cloud, calculate the total number of laser pulse echoes received by that point to obtain the total number of echoes N. E ;

[0087] 3) For each echo, calculate the echo number, i.e., how many times the echo returned, and divide the echo number by the total number of echoes N. E The echo ratio R is obtained. E .

[0088] The extraction of multi-scale features from airborne LiDAR point cloud data specifically includes the following steps:

[0089] 1) Determine the scale of interest: 1.5m, 2.5m, and 3.5m;

[0090] 2) For each scale of interest, repeatedly calculate the local geometric features of that neighborhood, including: height variance H. V Linearity L λ Flatness P λ Spherical degree S λ Curvature change k, perpendicularity N z and plane fitting residual R z .

[0091] Step S3: The training model is obtained by training the airborne LiDAR point cloud dataset and the four extracted features using the random forest algorithm and a classification strategy. Then, based on the quantitative relationship between the training model and the target model, the model is calibrated using the validation dataset to finally obtain the target model PBL-Random Forest model.

[0092] like Figure 5As shown, this invention, based on the PBL (Presence and Background Learning) method, models the probability (denoted as P(y=1|x)) of a single sample x being a positive sample by classifying the collected airborne LiDAR point cloud dataset according to whether it is labeled or not, treating labeled data as positive samples and unlabeled data as unlabeled samples, i.e., the target model P(y=1|x).

[0093] First, positive and unlabeled samples are randomly selected in a 1:5 ratio. Features are extracted and normalized to the range [0,1]. Then, 75% of the samples are used as the training dataset and 25% as the validation dataset. Features in the training process have equal weights. Let s=1 represent that x is collected as a positive sample and s=0 represent that x is collected as an unlabeled sample. Then, a random forest binary classifier model P(s=1|x,η=1) is trained using a random forest model, where η=1 indicates the presence of a background policy.

[0094] The specific steps for training a random forest binary classifier are as follows:

[0095] 1) Use the Bootstraping method to randomly draw multiple subsets of samples with replacement from the training set. Each subset is used to train a decision tree. The Bootstraping method is a method to estimate the distribution of a statistic or perform hypothesis testing by repeatedly sampling (drawing samples with replacement from the original dataset).

[0096] 2) When constructing each decision tree, a subset of features are randomly selected as candidate features, and the optimal feature is chosen from these for node splitting;

[0097] 3) Each tree grows to its maximum depth without pruning;

[0098] 4) Use majority voting to aggregate the prediction results of multiple decision trees;

[0099] 5) Optimize the trained model, including adjusting hyperparameters (such as the number of decision trees, maximum number of features, maximum depth, etc.) and optimizing the model structure, in order to improve the model's performance and generalization ability.

[0100] 6) Finally, the trained model P(s=1|x,η=1) is obtained.

[0101] Based on the conditional probability method and random sampling strategy, the quantitative relationship between the training model P(s=1|x,η=1) and the target model P(y=1|x) can be expressed as:

[0102]

[0103] In the formula, c is a constant used for calibration. c is estimated from the validation dataset O and is calculated as follows:

[0104]

[0105] In the formula, n is the cardinality of the dataset O.

[0106] Finally, the target model P(y=1|x) is obtained, which is the PBL-Random Forest model.

[0107] Step S4: Use the PBL-Random Forest model to classify power lines in the airborne point cloud data, then vectorize the classification results to finally generate vector results of power lines.

[0108] like Figure 6 As shown, in this embodiment, vectorization refers to the process of converting raster-format image data into vector data. After classifying the airborne point cloud data into power lines using the PBL-Random Forest model, vector lines are generated by connecting adjacent boundary points in the point cloud, and finally, the vector result of the power lines is obtained.

[0109] This embodiment addresses the problem that most existing point cloud datasets focus solely on the point cloud data itself, neglecting the importance of background or absence information in object detection tasks. It employs the Presence and Background Learning (PBL) method, utilizing both presence and background data for model training, effectively solving the problem of missing absence information in point cloud data. PBL is a general uniclass classification scheme that models the probability (denoted as P(y=1|x)) that a single sample x is a positive sample based on the input positive and unlabeled samples. Compared to traditional methods, PBL can estimate the actual probability of an object's presence under given environmental covariates.

[0110] In this embodiment, to improve the accuracy of the Random Forest algorithm in point cloud classification tasks, Random Forest is combined with the Probabilistic Learning (PBL) method to form a PBL-Random Forest model. Random Forest improves the robustness and prediction accuracy of the model by constructing multiple decision trees and outputting the average or mode of their predictions. Using Random Forest in PBL can further improve the prediction accuracy of the PBL model. Since PBL relies on accurate posterior probability estimation, and the ensemble nature of Random Forest makes it better suited to handling complex and noisy datasets, it can provide more reliable probability estimates.

[0111] This invention employs the PBL (Presence and Background Learning) method. This method completely solves the problem of absence information often overlooked in traditional point cloud datasets. By cleverly fusing presence data and background data, it achieves comprehensiveness and depth in model training. Furthermore, the PBL method significantly improves the prediction accuracy of object presence states by accurately utilizing these two complementary information sources. Notably, this method also overcomes the limitation of traditional techniques that can only output relative suitability scores; it can accurately estimate the actual probability of an object's existence under specific environmental covariates. This probabilistic prediction output not only greatly enhances the interpretability of the prediction results but also provides a more objective and scientific benchmark for comparing the performance of different models. In addition, the PBL-Random Forest model, obtained by employing random forest and a single-class classification strategy, does not use an existing classification model but combines random forest with the PBL method, enabling the PBL-Random Forest model to maintain high classification accuracy even when there is no missing information in the point cloud data.

[0112] This application also provides a power line target extraction device based on random forest and a single-class classification strategy. It should be noted that this device can be used to execute the power line target extraction method based on random forest and a single-class classification strategy provided in this application. This device is used to implement the above embodiments and preferred embodiments; details already described will not be repeated. As used below, the term "module" can refer to a combination of software and / or hardware that performs a predetermined function. Although the device described in the following embodiments is preferably implemented in software, hardware implementation, or a combination of software and hardware, is also possible and contemplated.

[0113] The following describes the power line target extraction device based on random forest and a classification strategy provided in the embodiments of this application.

[0114] Figure 7 This is a schematic diagram of a power line target extraction device based on random forest and a classification strategy according to an embodiment of this application. Figure 7 As shown, the device includes:

[0115] The first acquisition unit 71 is used to acquire airborne LiDAR point cloud data and multispectral orthophotos collected by the UAV multi-sensor platform.

[0116] The second acquisition unit 72 is used to acquire a feature set based on the airborne LiDAR point cloud data. The feature set includes height-based features, feature-based features, echo-based features, and multi-scale features. The multi-scale features include height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual.

[0117] Training unit 73 is used to train the random forest binary classifier using the above-mentioned airborne LiDAR point cloud data, the above-mentioned feature set, and the above-mentioned multispectral orthophotos based on the random forest algorithm and a classification strategy, to obtain the PBL-RandomForest model.

[0118] Classification unit 74 is used to acquire airborne point cloud data to be classified, and to classify power lines using the above-mentioned PBL-Random Forest model to generate vector results of power lines.

[0119] In this embodiment, the first acquisition unit is used to acquire airborne LiDAR point cloud data and multispectral orthophotos collected by the UAV multi-sensor platform; the second acquisition unit is used to acquire a feature set based on the airborne LiDAR point cloud data, wherein the feature set includes altitude-based features, feature-based features, echo-based features, and multi-scale features, wherein the multi-scale features include altitude variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual; the training unit is used to train a random forest binary classifier using the airborne LiDAR point cloud data, the feature set, and the multispectral orthophotos according to the random forest algorithm and a single-class classification strategy to obtain a PBL-Random Forest model; the classification unit is used to acquire airborne point cloud data to be classified, and use the PBL-Random Forest model to classify the airborne point cloud data to be classified as electric field lines to generate vector results of electric field lines. By combining airborne LiDAR point cloud data and feature sets with the random forest method and a single-class classification strategy, the PBL-Random Forest model was obtained. The PBL-Random Forest model was used to predict and identify power lines, which improved the accuracy of power line identification and solved the problem that existing models could not accurately predict the actual existence of power lines.

[0120] As an optional scheme, the second acquisition unit includes an identification module, an interpolation processing module, and a first definition module. The identification module is used to identify ground points in the airborne LiDAR point cloud data using the progressive triangular irregular network densification filtering algorithm in LAStools software, to obtain the first echo data and ground echo data. The interpolation processing module is used to perform interpolation processing on the first echo data and the ground echo data according to the ordinary kriging method, to obtain a digital surface model and a digital elevation model. The first definition module is used to determine a normalized digital surface model based on the digital surface model and the digital elevation model, define a spherical neighborhood for each point on the normalized digital surface model, determine the height variance of all points in the spherical neighborhood, and allocate the height variance to the center point of the spherical neighborhood to obtain the height-based features.

[0121] In one optional scheme, the second acquisition unit includes a second definition module, a first determination module, a decomposition module, and a second determination module. The second definition module is used to define a spherical neighborhood centered on each point cloud in the airborne LiDAR point cloud data. The first determination module is used to determine the local covariance matrix of each point cloud based on the spherical neighborhood. The decomposition module is used to perform eigenvalue decomposition on the local covariance matrix to obtain multiple eigenvectors and multiple eigenvalues. The second determination module is used to determine the relevant features of the eigenvectors and the relevant features of the eigenvalues ​​respectively to obtain the feature-based features. The relevant features of the eigenvectors include linearity, flatness, sphericity, and curvature variation, and the relevant features of the eigenvalues ​​include perpendicularity and plane fitting residuals.

[0122] In one optional scheme, the second acquisition unit includes a third determining module and a fourth determining module. The third determining module is used to determine the value of the laser pulse echo received by each point cloud in the airborne LiDAR point cloud data to obtain the total number of laser pulse echoes. The fourth determining module is used to determine the echo number of each of the laser pulse echoes and determine the echo ratio based on the total number of laser pulse echoes and the echo number.

[0123] In one optional scheme, the second acquisition unit includes a fifth determining module and a sixth determining module. The fifth determining module is used to acquire a preset number of preset scales and determine spherical neighborhoods with the above-mentioned preset scales as radii based on the airborne LiDAR point cloud data. The sixth determining module is used to determine the local geometric features of the above-mentioned spherical neighborhoods corresponding to each preset scale, thereby obtaining the above-mentioned multi-scale features.

[0124] In one optional scheme, the training unit includes a classification processing module and a training module. The classification processing module is used to classify the airborne LiDAR point cloud data to obtain labeled samples and unlabeled samples. The labeled samples represent samples with labeled power lines, and the unlabeled samples represent samples without labeled power lines. The training module is used to train the forest binary classifier using the labeled and unlabeled samples and the feature set. During the training of the random forest binary classifier, the output results of the random forest binary classifier are verified using the multispectral orthophoto.

[0125] In one alternative, the apparatus further includes a determining unit, which, after generating the vector result of the power line, determines the position and shape of the power line based on the vector result, so as to realize the inspection work of the power line.

[0126] The aforementioned power line target extraction device based on random forest and a single-class classification strategy includes a processor and a memory. The first acquisition unit, second acquisition unit, training unit, and classification unit are all stored as program units in the memory, and the processor executes these program units to achieve their respective functions. All of the above modules reside in the same processor; alternatively, the modules may be located in different processors in any combination.

[0127] The processor contains a kernel, which retrieves the corresponding program units from memory. One or more kernels can be configured, and adjusting kernel parameters can address the problem that existing models struggle to accurately predict the actual existence of power lines.

[0128] The memory may include non-permanent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM, and the memory includes at least one memory chip.

[0129] This invention provides a computer-readable storage medium including a stored program, wherein, when the program is executed, it controls the device where the computer-readable storage medium is located to perform the power line target extraction method based on random forest and a classification strategy.

[0130] Specifically, methods for extracting power line targets based on random forests and a single-class classification strategy include:

[0131] Step S201: Acquire airborne LiDAR point cloud data and multispectral orthophotos collected by the UAV multi-sensor platform;

[0132] Step S202: Obtain a feature set based on the airborne LiDAR point cloud data. The feature set includes height-based features, feature-based features, echo-based features, and multi-scale features. The multi-scale features include height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual.

[0133] Step S203: Based on the random forest algorithm and a single-class classification strategy, the random forest binary classifier is trained using the above-mentioned airborne LiDAR point cloud data, the above-mentioned feature set, and the above-mentioned multispectral orthophoto to obtain the PBL-RandomForest model.

[0134] Step S204: Obtain the airborne point cloud data to be classified, and use the above-mentioned PBL-Random Forest model to classify the airborne point cloud data to be classified into power lines, generating vector results of power lines.

[0135] This invention provides a processor for running a program, wherein the program executes the power line target extraction method based on random forest and a classification strategy.

[0136] Specifically, methods for extracting power line targets based on random forests and a single-class classification strategy include:

[0137] Step S201: Acquire airborne LiDAR point cloud data and multispectral orthophotos collected by the UAV multi-sensor platform;

[0138] Step S202: Obtain a feature set based on the airborne LiDAR point cloud data. The feature set includes height-based features, feature-based features, echo-based features, and multi-scale features. The multi-scale features include height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual.

[0139] Step S203: Based on the random forest algorithm and a single-class classification strategy, the random forest binary classifier is trained using the above-mentioned airborne LiDAR point cloud data, the above-mentioned feature set, and the above-mentioned multispectral orthophoto to obtain the PBL-RandomForest model.

[0140] Step S204: Obtain the airborne point cloud data to be classified, and use the above-mentioned PBL-Random Forest model to classify the airborne point cloud data to be classified into power lines, generating vector results of power lines.

[0141] This invention provides an electronic device, which includes a processor, a memory, and a program stored in the memory and executable on the processor. When the processor executes the program, it performs at least the following steps:

[0142] Step S201: Acquire airborne LiDAR point cloud data and multispectral orthophotos collected by the UAV multi-sensor platform;

[0143] Step S202: Obtain a feature set based on the airborne LiDAR point cloud data. The feature set includes height-based features, feature-based features, echo-based features, and multi-scale features. The multi-scale features include height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual.

[0144] Step S203: Based on the random forest algorithm and a single-class classification strategy, the random forest binary classifier is trained using the above-mentioned airborne LiDAR point cloud data, the above-mentioned feature set, and the above-mentioned multispectral orthophoto to obtain the PBL-RandomForest model.

[0145] Step S204: Obtain the airborne point cloud data to be classified, and use the above-mentioned PBL-Random Forest model to classify the airborne point cloud data to be classified into power lines, generating vector results of power lines.

[0146] The devices mentioned in this article can be servers, PCs, tablets, mobile phones, etc.

[0147] This application also provides a computer program product, which, when executed on a data processing device, is suitable for executing an initialization program having at least the following method steps:

[0148] Step S201: Acquire airborne LiDAR point cloud data and multispectral orthophotos collected by the UAV multi-sensor platform;

[0149] Step S202: Obtain a feature set based on the airborne LiDAR point cloud data. The feature set includes height-based features, feature-based features, echo-based features, and multi-scale features. The multi-scale features include height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual.

[0150] Step S203: Based on the random forest algorithm and a single-class classification strategy, the random forest binary classifier is trained using the above-mentioned airborne LiDAR point cloud data, the above-mentioned feature set, and the above-mentioned multispectral orthophoto to obtain the PBL-RandomForest model.

[0151] Step S204: Obtain the airborne point cloud data to be classified, and use the above-mentioned PBL-Random Forest model to classify the airborne point cloud data to be classified into power lines, generating vector results of power lines.

[0152] It is obvious to those skilled in the art that the modules or steps of the present invention described above can be implemented using general-purpose computing devices. They can be centralized on a single computing device or distributed across a network of multiple computing devices. They can be implemented using computer-executable program code, and thus can be stored in a storage device for execution by a computing device. In some cases, the steps shown or described can be performed in a different order than those described herein, or they can be fabricated as separate integrated circuit modules, or multiple modules or steps can be fabricated as a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0153] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0154] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0155] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0156] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0157] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0158] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, like read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0159] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0160] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0161] As can be seen from the above description, the embodiments of this application achieve the following technical effects:

[0162] 1) A method for extracting power line targets based on random forest and a single-class classification strategy according to this application includes: acquiring airborne LiDAR point cloud data and multispectral orthophotos collected by a UAV multi-sensor platform; acquiring a feature set based on the airborne LiDAR point cloud data, wherein the feature set includes height-based features, feature-based features, echo-based features, and multi-scale features, and the multi-scale features include height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residuals; training a random forest binary classifier using the airborne LiDAR point cloud data, the feature set, and the multispectral orthophotos according to the random forest algorithm and a single-class classification strategy to obtain a PBL-Random Forest model; acquiring the airborne point cloud data to be classified, and using the PBL-Random Forest model to classify the airborne point cloud data to be classified as power lines to generate vector results of power lines. By combining airborne LiDAR point cloud data and feature sets with the random forest method and a single-class classification strategy, the PBL-Random Forest model was obtained. The PBL-Random Forest model was used to predict and identify power lines, which improved the accuracy of power line identification and solved the problem that existing models could not accurately predict the actual existence of power lines.

[0163] 2) A power line target extraction device based on random forest and a class-aspect classification strategy according to this application includes: a first acquisition unit for acquiring airborne LiDAR point cloud data and multispectral orthophotos collected by a UAV multi-sensor platform; a second acquisition unit for acquiring a feature set based on the airborne LiDAR point cloud data, wherein the feature set includes height-based features, feature-based features, echo-based features, and multi-scale features, and the multi-scale features include height variance, linearity, flatness, sphericity, curvature variation, verticality, and plane fitting residual; a training unit for training a random forest binary classifier using airborne LiDAR point cloud data, feature set, and multispectral orthophotos according to the random forest algorithm and a class-aspect classification strategy to obtain a PBL-Random Forest model; and a classification unit for acquiring airborne point cloud data to be classified, using the PBL-Random Forest model to classify power lines in the airborne point cloud data to be classified, and generating vector results of power lines. By combining airborne LiDAR point cloud data and feature sets with the random forest method and a single-class classification strategy, the PBL-Random Forest model was obtained. The PBL-Random Forest model was used to predict and identify power lines, which improved the accuracy of power line identification and solved the problem that existing models could not accurately predict the actual existence of power lines.

[0164] The above description is merely a preferred embodiment of this application and is not intended to limit this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

Claims

1. A power line target extraction method based on random forest and one-class classification strategy, characterized in that, The method comprises the following steps: acquiring airborne LiDAR point cloud data and multi-spectral orthographic images collected by a multi-sensor platform of a UAV; acquiring a feature set from the airborne LiDAR point cloud data, wherein the feature set comprises height-based features, feature-based features, echo-based features, and multi-scale features, the multi-scale features comprising height variance, linearity, planarity, sphericity, curvature variation, perpendicularity, and plane fitting residual; training a random forest binary classifier using the airborne LiDAR point cloud data, the feature set, and the multi-spectral orthographic images according to a random forest algorithm and a one-class classification strategy to obtain a PBL-Random Forest model; acquiring to-be-classified airborne point cloud data and performing power line classification on the to-be-classified airborne point cloud data using the PBL-Random Forest model to generate a vector result of power lines; acquiring height-based features from the airborne LiDAR point cloud data, comprising: identifying ground points in the airborne LiDAR point cloud data using a progressive triangulated irregular network densification filtering algorithm in LAStools software to obtain first echo data and ground echo data; interpolating the first echo data and the ground echo data respectively according to an ordinary Kriging method to obtain a digital surface model and a digital elevation model; determining a normalized digital surface model according to the digital surface model and the digital elevation model, defining a spherical neighborhood for each point on the normalized digital surface model, determining height variance of all points in the spherical neighborhood, and assigning the height variance to the center point of the spherical neighborhood to obtain the height-based features; wherein the echo-based features comprise total laser pulse echo number and echo ratio, and the echo-based features are acquired from the airborne LiDAR point cloud data, comprising: determining the number of laser pulse echoes received by each point cloud in the airborne LiDAR point cloud data to obtain the total laser pulse echo number; determining the echo number of each laser pulse echo, and determining the echo ratio according to the total laser pulse echo number and the echo number.

2. The method of claim 1, wherein, acquiring feature-based features from the airborne LiDAR point cloud data, comprising: defining a spherical neighborhood centered at each point cloud in the airborne LiDAR point cloud data; determining a local covariance matrix of each point cloud according to the spherical neighborhood; performing eigenvalue decomposition on the local covariance matrix to obtain a plurality of eigenvectors and a plurality of eigenvalues; determining relevant features of the eigenvectors and relevant features of the eigenvalues respectively to obtain the feature-based features, wherein the relevant features of the eigenvectors comprise linearity, planarity, sphericity, and curvature variation, and the relevant features of the eigenvalues comprise perpendicularity and plane fitting residual.

3. The method of claim 1, wherein, acquiring multi-scale features from the airborne LiDAR point cloud data, comprising: acquiring a preset number of preset scales, and determining a spherical neighborhood with each preset scale as a radius according to the airborne LiDAR point cloud data; determining local geometric features of the spherical neighborhood corresponding to each preset scale to obtain the multi-scale features.

4. The method of claim 1, wherein, training a random forest binary classifier using the airborne LiDAR point cloud data, the feature set, and the multi-spectral orthoimage pair, including: classifying the airborne LiDAR point cloud data to obtain labeled samples and unlabeled samples, wherein the labeled samples represent samples that have been labeled with the power lines, and the unlabeled samples represent samples that have not been labeled with the power lines; training the random forest binary classifier using the labeled samples and the unlabeled samples, and the feature set, and verifying the output of the random forest binary classifier using the multi-spectral orthoimage pair during the training of the random forest binary classifier.

5. The method of claim 1, wherein, After generating the vector result of the power lines, the method further includes: determining the position and shape of the power lines according to the vector result of the power lines to implement the inspection of the power lines.

6. A power line target extraction device based on a random forest and a one-class classification strategy, characterized by, including: a first obtaining unit configured to obtain airborne LiDAR point cloud data and multi-spectral orthoimage collected by a multi-sensor platform of a UAV; a second obtaining unit configured to obtain a feature set from the airborne LiDAR point cloud data, wherein the feature set includes height-based features, feature-based features, echo-based features, and multi-scale features, and the multi-scale features include height variance, linearity, planarity, sphericity, curvature variation, perpendicularity, and plane fitting residual; a training unit configured to train a random forest binary classifier using the airborne LiDAR point cloud data, the feature set, and the multi-spectral orthoimage pair according to a random forest algorithm and a one-class classification strategy to obtain a PBL-RandomForest model; a classification unit configured to obtain to-be-classified airborne point cloud data, and classify the to-be-classified airborne point cloud data using the PBL-RandomForest model to generate a vector result of power lines; the second obtaining unit includes an identification module, an interpolation processing module, and a first definition module; the identification module is configured to identify ground points in the airborne LiDAR point cloud data using a progressive triangulated irregular network densification filtering algorithm in LAStools software to obtain first echo data and ground echo data; the interpolation processing module is configured to interpolate the first echo data and the ground echo data respectively according to an ordinary Kriging method to obtain a digital surface model and a digital elevation model; and the first definition module is configured to determine a normalized digital surface model according to the digital surface model and the digital elevation model, define a spherical neighborhood for each point on the normalized digital surface model, determine height variance of all points in the spherical neighborhood, and assign the height variance to a center point of the spherical neighborhood to obtain the height-based features. The second acquisition unit comprises a third determination module and a fourth determination module, the echo-based features comprise a total laser pulse echo number and an echo ratio, the third determination module is configured to determine a value of a laser pulse echo received by each point cloud in the airborne LiDAR point cloud data, and obtain the total laser pulse echo number; and the fourth determination module is configured to determine an echo number of each laser pulse echo, and determine the echo ratio according to the total laser pulse echo number and the echo number.

7. A computer readable storage medium characterized in that, The computer readable storage medium comprises a stored program, wherein the program, when executed, controls a device in which the computer readable storage medium is located to perform the power line target extraction method based on the random forest and the one-class classification strategy according to any one of claims 1 to 5.

8. An electronic device, comprising: Comprise: One or more processors, memories, and one or more programs, wherein the one or more programs are stored in the memories and configured to be executed by the one or more processors, and the one or more programs comprise a program for executing the power line target extraction method based on the random forest and the one-class classification strategy according to any one of claims 1 to 5.

Images