Method for monitoring coverage and thickness of biological crust in mining area based on UAV (Unmanned Aerial Vehicle) multispectral data

Abstract

The invention discloses a method for monitoring the coverage and thickness of a biological crust in a mining area based on UAV multispectral data, and the method comprises the steps: carrying a multispectral sensor through an unmanned plane, and collecting the multispectral image data of the biological crust in an experiment area; synchronously sampling and measuring actual measurement parameters of the biological crust on the ground; the multispectral image data are imported into cartographic software to be preprocessed, and various vegetation indexes are calculated; matching the vegetation indexes with actually measured parameters to form a data set, drawing a correlation heat map, and screening out the vegetation indexes of which the correlation meets conditions as input features of a machine learning model; multiple machine learning models are constructed and trained, and the machine learning model with the highest prediction precision is output as an optimal model; and evaluating the optimal model, and generating a distribution diagram of the biological crust coverage and thickness. According to the method, the unmanned aerial vehicle multispectral remote sensing technology and the advanced machine learning algorithm are fused, the monitoring efficiency and the data accuracy are improved, and a scientific basis is provided for ecological restoration of a mining area.

Description

Technical Field

[0001] The present invention relates to the technical field of ecological environment monitoring, and in particular to a method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data. Background Art

[0002] In the process of ecological restoration and reconstruction in mining areas, biological crust, as a special soil surface cover, plays a vital role. Biological crusts are organic-inorganic complexes composed of microorganisms and microorganisms such as prokaryotic algae, eukaryotic algae, lichens, mosses, bacteria, fungi and actinomycetes. Through their own physiological activities and metabolism, they can significantly improve the soil quality in mining areas, promote the recovery of vegetation, and thus accelerate the ecological reconstruction process in mining areas. However, biological crusts are pioneer species in mine ecological restoration and the beginning of natural environmental succession. How to accurately and efficiently monitor these indicators has always been a difficult problem in the field of ecological environment monitoring.

[0003] Traditional manual monitoring methods have many limitations. First, manual monitoring is time-consuming and labor-intensive, requiring a large amount of human resources, and the monitoring range is limited, making it difficult to meet the needs of large-scale mining area ecological restoration. UAV multispectral remote sensing technology can meet the needs of large-scale site monitoring and save a lot of manpower, material resources and time. Second, manual monitoring is easily affected by subjective factors, resulting in errors and inconsistencies in monitoring results, while UAV multispectral remote sensing technology can avoid this influence to a certain extent.

[0004] In recent years, with the rapid development of unmanned aerial vehicle (UAV) technology, multispectral remote sensing technology has gradually become an important means in the field of ecological environment monitoring. UAVs are equipped with multispectral sensors, which can quickly obtain spectral information of the surface over a large range. By calculating various vegetation indices, such as the normalized difference vegetation index (NDVI), the green normalized difference vegetation index (GNDVI) and the ratio vegetation index (RVI), the growth status and health of biological crusts can be reflected. However, the inversion method of biological crust coverage and thickness in mining areas based on UAV multispectral technology is still in the exploratory stage. Although there have been studies using multispectral remote sensing technology to monitor the spectral characteristics of biological crusts, most of them focus on the qualitative analysis level and lack effective methods for quantitative inversion of biological crust coverage and thickness. In addition, the differences in biological crust types, soil conditions, time and light in different mining areas also pose challenges to the application of multispectral remote sensing technology. At present, UAV multispectral technology is mainly used in agriculture, water bodies and other fields. There are still certain technical difficulties in the accurate monitoring of biological crusts in mining areas. In particular, how to achieve accurate inversion of the coverage and thickness of biological crusts is a hot spot and difficulty in current research.

[0005] In summary, the existing technology has many shortcomings in the monitoring of biological crusts in mining areas, and there is an urgent need for a method that can efficiently and accurately invert the coverage and thickness of biological crusts in mining areas to provide a scientific basis and technical support for ecological restoration in mining areas. Therefore, the present invention aims to combine the multi-spectral remote sensing technology of unmanned aerial vehicles with machine learning algorithms to propose a method for inverting the coverage and thickness of biological crusts in mining areas based on UAV multi-spectral inversion. This method achieves rapid and accurate monitoring of the coverage and thickness of biological crusts in mining areas by optimizing data acquisition and processing procedures, screening the optimal vegetation index, and constructing a high-precision inversion model, providing a scientific basis and decision-making support for ecological restoration in mining areas. Summary of the invention

[0006] The technical problem to be solved by the present invention is to address the following defects in the prior art: In the process of ecological restoration in mining areas, biological crusts, as pioneer species, are crucial to improving soil conditions, preventing soil erosion, and promoting vegetation recovery. However, traditional manual monitoring methods have many shortcomings in monitoring the coverage and thickness of biological crusts in mining areas. First, manual monitoring is inefficient, requires a lot of manpower input, and has a limited monitoring range, which is difficult to meet the monitoring needs of large-scale mining areas. Secondly, the data accuracy of manual monitoring is insufficient, and it is easily interfered by subjective judgment and external factors, resulting in inaccurate monitoring results. In addition, traditional methods lack real-time and dynamic monitoring capabilities, and cannot promptly reflect changes in the growth status of biological crusts. A method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data is provided. The method integrates unmanned aerial vehicle (UAV) multispectral remote sensing technology and advanced machine learning algorithms, aiming to improve monitoring efficiency and data accuracy, and provide a scientific basis for ecological restoration in mining areas.

[0007] The technical solution adopted by the present invention to solve its technical problem is:

[0008] The present invention provides a method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data, the method comprising the following steps:

[0009] S1. Use a drone equipped with a multispectral sensor to collect multispectral image data of the biocrust in the experimental area at a preset flight altitude and route; and simultaneously sample and measure the measured parameters of the biocrust on the ground;

[0010] S2, importing the multispectral image data into mapping software for preprocessing, and calculating various vegetation indices on the preprocessed multispectral image data;

[0011] S3. Match and integrate the calculated vegetation index with the measured parameters of ground sampling to form a complete data set; draw a correlation heat map of the data set, quantitatively analyze the correlation between the vegetation index and the measured parameters, compare the correlation between the vegetation index and the coverage and thickness of the biological crust, and select the vegetation index with the qualified correlation as the input feature of the machine learning model;

[0012] S4. Build multiple machine learning models, divide the input features into training sets and test sets, use the training sets to train the machine learning models, adjust the model parameters, and output the machine learning model with the highest prediction accuracy as the optimal model; evaluate the optimal model using the test set and multiple evaluation indicators;

[0013] S5. For the biological crusts in the mining area to be monitored, the multispectral image data of UAV is combined with the prediction results of the optimal model to generate the distribution map of the coverage and thickness of the biological crusts.

[0014] Furthermore, the method for collecting multispectral image data in step S1 of the present invention includes:

[0015] Select drones with emergency braking and obstacle avoidance, strong wind resistance, and light intensity correction functions, equipped with multispectral lenses covering multiple key spectral bands including visible light and near-infrared light;

[0016] Multiple sampling points were evenly arranged in the experimental area according to the topography and distribution of biocrusts, and the location information of each sampling point was recorded;

[0017] At the preset flight altitude and route, the multispectral lens carried by the drone conducts a comprehensive scan and data collection of the experimental area.

[0018] Furthermore, the method of sampling on the ground and measuring the measured parameters in step S1 of the present invention includes: using ground sampling equipment, including a ring knife soil sampler, a tape measure, a vernier caliper, an insulated box, and a GPS locator, to collect biological crust samples and measure their parameters, including: chlorophyll content and thickness.

[0019] Furthermore, the preprocessing method in step S2 of the present invention includes: importing the multispectral image data collected by the drone into mapping software for preprocessing, including image stitching, cropping, and denoising, to obtain a multi-band map of the experimental site.

[0020] Furthermore, the method for calculating the vegetation index on the pre-processed multispectral image data in step S2 of the present invention includes:

[0021] ArcGIS Pro and ENVI software were used to calculate the vegetation index of the preprocessed multispectral image data. The coverage, growth status and chlorophyll content of the biological crust were quantitatively evaluated by calculating the vegetation indices including NDVI and GNDVI.

[0022] Furthermore, the method in step S3 of the present invention includes:

[0023] The calculated vegetation index is matched and integrated with the measured parameters of ground sampling to form a complete data set;

[0024] Pycharm software was used to draw the Pearson correlation heat map, and the correlation between vegetation index and key parameters of biological crust was quantitatively analyzed. The vegetation index that was highly correlated with the coverage and thickness of biological crust was screened out as the input feature of the subsequent machine learning model. According to the P value of Pearson correlation, the vegetation index with a P value less than a certain threshold was screened out.

[0025] Furthermore, the method in step S4 of the present invention includes:

[0026] Construct a variety of machine learning models, including chlorophyll BP model, GBDT model, KNN model, RFR model, RR model, SVR model, and XGB model; divide the input features obtained by feature selection into training sets and test sets, input the training set into the machine learning model, and finally select the model with the highest prediction accuracy as the optimal model by continuously adjusting the model parameters and optimizing the algorithm structure.

[0027] Furthermore, the method for evaluating in step S4 of the present invention includes: using the test set data to verify the optimal model by including R 2 Evaluation indicators including , slope, and RMSE are used to comprehensively evaluate the prediction accuracy of the optimal model.

[0028] Furthermore, the method for generating a distribution map in step S5 of the present invention includes:

[0029] ArcGIS Pro geographic information system software was used to visualize the drone multispectral image data and the prediction results of the machine learning model to generate a distribution map of the coverage and thickness of the biological crust.

[0030] The present invention provides a system for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data, comprising:

[0031] a memory for storing executable computer programs;

[0032] The processor is used to implement the above-mentioned method of monitoring the coverage and thickness of biological crust in mining areas based on UAV multispectral data when executing the executable computer program stored in the memory.

[0033] The beneficial effects produced by the present invention are:

[0034] 1. Application of UAV multispectral remote sensing technology:

[0035] ①Large-scale rapid monitoring: For the first time, drone multispectral remote sensing technology was applied to the monitoring of biological crusts in mining areas, breaking through the limitations of traditional ground monitoring methods that are restricted by factors such as terrain and weather, and achieving large-scale, rapid, and non-destructive monitoring of biological crusts in mining areas. This technology not only improves monitoring efficiency, but also ensures the continuity and real-time nature of data.

[0036] ② Impact on data acquisition: The multispectral sensor carried by the drone can capture the reflectance characteristics of the surface biological crust in different spectral bands, providing rich spectral information for accurate estimation of the coverage and thickness of the biological crust. Compared with traditional methods, this technology can more accurately reflect the growth and health of the biological crust.

[0037] By using drones equipped with multi-spectral lenses, large-scale, rapid, and non-destructive monitoring of biological crusts in mining areas can be achieved, improving monitoring efficiency and accuracy.

[0038] 2. Introduction of machine learning algorithms:

[0039] ① Intelligent processing: By building a machine learning model based on multispectral image data, the intelligent prediction of the coverage and thickness of biological crusts is realized. This algorithm can not only automatically extract key features in the image, but also accurately classify and estimate these features, significantly improving the efficiency of data processing and analysis.

[0040] ②Model optimization and verification: By comparing and verifying a variety of machine learning algorithms, the most suitable model for this application scenario was selected, and the prediction accuracy and stability of the model were improved by continuously adjusting parameters and optimizing the algorithm structure. This model optimization process ensures the reliability and accuracy of the prediction results.

[0041] By using machine learning algorithms to perform nonlinear fitting on multispectral image data, a model for estimating the coverage and thickness of biological crusts was constructed, which solved the problem that traditional methods are difficult to accurately estimate the coverage and thickness of biological crusts.

[0042] 3. Establishment of a comprehensive evaluation system:

[0043] ① Multi-dimensional evaluation: A comprehensive evaluation system for the restoration effect of biological crusts was constructed by combining drone image data and ground sampling measured values. This system not only takes into account key parameters such as the coverage and thickness of biological crusts, but also comprehensively considers factors such as the health status and growth environment of biological crusts, providing comprehensive data support and decision-making basis for ecological restoration in mining areas.

[0044] ② Visualization: Using ArcGIS Pro geographic information system software, the drone image data and the prediction results of the machine learning model were visualized to generate a distribution map of the coverage and thickness of the biological crust. This intuitive and easy-to-understand display method helps decision makers quickly understand the restoration effect and distribution of biological crusts in the mining area, and provides strong data support for subsequent ecological restoration work.

[0045] By combining UAV image data and ground sampling measurements, a comprehensive evaluation system for the restoration effect of biological crusts was constructed, providing a scientific basis for ecological restoration in mining areas. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] The present invention will be further described below with reference to the accompanying drawings and embodiments, in which:

[0047] Figure 1 It is a technical roadmap of an embodiment of the present invention;

[0048] Figure 2 is a Pearson correlation heat map of an embodiment of the present invention;

[0049] Figure 3 is an RGB image of a phosphogypsum mine according to an embodiment of the present invention;

[0050] Figure 4 is the chlorophyll value of the phosphogypsum mine in an embodiment of the present invention;

[0051] Figure 5 is a phosphogypsum site classification diagram of an embodiment of the present invention;

[0052] Figure 6 is the thickness classification of the embodiment of the present invention;

[0053] Figure 7 is the thickness of the biological crust according to the embodiment of the present invention;

[0054] Figure 8 This is a real shot of the biological crust in the embodiment of the present invention Figure 1 ;

[0055] Fig. 9 This is a real shot of the biological crust in the embodiment of the present invention Figure 2 ;

[0056] Fig.10 is the distribution of different years of the embodiments of the present invention;

[0057] Fig.11 is a thickness BP model of an embodiment of the present invention;

[0058] Fig.12 is a thickness KNN model of an embodiment of the present invention;

[0059] Fig.13 is a thickness RFR model of an embodiment of the present invention;

[0060] Fig.14 is the thickness RR model of the embodiment of the present invention;

[0061] Fig.15 is a thickness SVR model of an embodiment of the present invention;

[0062] Fig.16 is the thickness XGB model of an embodiment of the present invention;

[0063] Fig.17 is the chlorophyll BP model of an embodiment of the present invention;

[0064] Fig.18 is the chlorophyll GBDT model of an embodiment of the present invention;

[0065] Fig.19 is the chlorophyll KNN model of an embodiment of the present invention;

[0066] Fig. 20 is the chlorophyll RFR model of an embodiment of the present invention;

[0067] Fig.21 is the chlorophyll RR model of an embodiment of the present invention;

[0068] Fig. 22 is the chlorophyll SVR model of an embodiment of the present invention;

[0069] Fig.23 It is the chlorophyll XGB model of the embodiment of the present invention. DETAILED DESCRIPTION

[0070] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0071] Example 1

[0072] like Figure 1 As shown, the method for monitoring the coverage and thickness of biological crust in mining areas based on UAV multispectral data in an embodiment of the present invention includes the following steps:

[0073] S1. Use a drone equipped with a multispectral sensor to collect multispectral image data of the biocrust in the experimental area at a preset flight altitude and route; and simultaneously sample and measure the measured parameters of the biocrust on the ground;

[0074] S2, importing the multispectral image data into mapping software for preprocessing, and calculating the vegetation index for the preprocessed multispectral image data;

[0075] S3. Match and integrate the calculated vegetation index with the measured parameters of ground sampling to form a complete data set; draw a correlation heat map of the data set, quantitatively analyze the correlation between the vegetation index and the measured parameters, compare the correlation between the vegetation index and the coverage and thickness of the biological crust, and select the vegetation index with the qualified correlation as the input feature of the machine learning model;

[0076] S4. Build multiple machine learning models, divide the input features into training sets and test sets, use the training sets to train the machine learning models, adjust the model parameters, and output the machine learning model with the highest prediction accuracy as the optimal model; evaluate the optimal model using the test set and multiple evaluation indicators;

[0077] S5. For the biological crusts in the mining area to be monitored, the multispectral image data of UAV is combined with the prediction results of the optimal model to generate the distribution map of the coverage and thickness of the biological crusts.

[0078] The present invention proposes a method for multi-spectral inversion of biological crust coverage and thickness in mining areas based on UAV (unmanned aerial vehicle). The method uses unmanned aerial vehicles equipped with multi-spectral sensors to conduct rapid and large-scale data collection, which significantly improves the monitoring efficiency. At the same time, by obtaining rich spectral information through multi-spectral remote sensing technology and combining advanced machine learning algorithms for data analysis, accurate inversion of biological crust coverage and thickness is achieved, improving the objectivity and accuracy of the data. In addition, the high-altitude operation capability of unmanned aerial vehicles means that the monitoring range is no longer limited to ground conditions, and comprehensive monitoring of the entire mining area can be achieved. More importantly, combined with real-time data processing and analysis technology, this method can achieve real-time and dynamic monitoring of the growth process of biological crusts, providing strong technical support for ecological restoration work in mining areas.

[0079] The present invention effectively solves the shortcomings of traditional monitoring methods in monitoring biological crust in mining areas, such as low efficiency, insufficient data accuracy, limited monitoring range, and lack of real-time and dynamic monitoring capabilities, by integrating unmanned aerial vehicle multispectral remote sensing technology and machine learning algorithms. In addition, mines lack a mature monitoring system, especially for biological crust. Therefore, the present invention provides a strong guarantee for scientific decision-making and precise implementation of ecological restoration work in mining areas.

[0080] Example 2

[0081] Based on Example 1, this embodiment of the present invention provides a specific device selection and model and software selection method.

[0082] 1. Equipment and Materials

[0083] 1. Core equipment: This solution uses the DJI M3 multispectral drone, which is known for its excellent flight stability and data collection capabilities. The DJI M3 integrates an emergency brake and obstacle avoidance system, which allows it to fly safely in complex environments; the twelve-level wind resistance ensures the stable operation of the drone under strong wind interference in the mining area; the light intensity correction function effectively reduces the impact of light changes on image quality. In terms of sensor configuration, the multispectral lens carried by the drone covers multiple key spectral bands such as visible light (such as GREEN, RED) and near-infrared (NIR), which can accurately capture the spectral reflectance characteristics of surface biological crusts. At the same time, the addition of a photometric sensor further improves the accuracy and reliability of the data.

[0084] 2. Ground sampling equipment: including knife ring soil sampler, tape measure, vernier caliper, incubator, etc., used to collect biological crust samples and determine their chlorophyll content, thickness and other parameters. The knife ring soil sampler is a special tool for collecting biological crust samples. It can accurately cut and collect surface biological crust samples for subsequent laboratory analysis. The chlorophyll meter can directly measure the chlorophyll content in biological crusts, providing an important basis for evaluating the health status of biological crusts. The level of chlorophyll content directly reflects the photosynthesis capacity and growth status of biological crusts. Other auxiliary tools include GPS locators, rulers, etc., which are used to accurately determine the location of sampling points and measure key parameters such as the thickness of biological crusts.

[0085] 3. Data processing software: Use professional software such as Pycharm, ArcGIS Pro, ENVI, etc. for image data processing, analysis, and the construction and verification of machine learning models. As a professional programming software, Pycharm provides strong support for the construction, training, and verification of machine learning models. Its rich libraries and tool sets make data processing and analysis more efficient and convenient; ArcGIS Pro is a powerful geographic information system software that excels in spatial analysis and visualization of image data. With ArcGIS Pro, multispectral image data collected by drones can be accurately spatially positioned and attribute analyzed; ENVI is professional software in the field of remote sensing image processing, and is good at tasks such as vegetation index calculation and image classification. In this plan, ENVI will be used to calculate vegetation indices such as NDVI and GNDVI, and provide key data support for the extraction of biological crust features (such as Figure 1 shown).

[0086] 2. Methods and Steps

[0087] 1. Experimental Design

[0088] ① Site selection: This plan selects the Zhongxiang phosphogypsum field in Hubei Province as the experimental site. The surface vegetation of the site has been severely damaged due to long-term mining activities, but in recent years, through ecological restoration measures, the distribution of biological crusts in different years (2006-2016) has gradually formed, such as Fig.10 This makes the site an ideal location for studying the growth patterns and monitoring techniques of biocrusts (e.g. Figures 8 to 9 shown).

[0089] ② Sampling point setting: In the experimental area, sampling points were evenly arranged according to the topography and distribution of biocrusts. Detailed location information was recorded for each sampling point to ensure the accuracy and traceability of the data.

[0090] ③Data collection: At the preset flight altitude and route, the drone equipped with a multispectral lens conducted a comprehensive scan and data collection of the experimental area. At the same time, the ground team simultaneously collected samples of biological crusts and measured parameters to obtain the measured values ​​of biological crusts as a benchmark for subsequent analysis.

[0091] 2. Image data processing:

[0092] ① Data import and preprocessing: The multispectral image data collected by the drone is imported into the DJI Terra mapping software for preprocessing, including image stitching, cropping, denoising and other steps to ensure the quality and availability of the image data and obtain multi-band images (GREEN, NIR, RED, etc.) under the field experiment.

[0093] ② Calculation of vegetation index: ArcGIS Pro and ENVI software are used to calculate the vegetation index of the preprocessed image data. The embodiment of the present invention can quantitatively evaluate the key parameters such as the coverage, growth status and chlorophyll content of the biological crust by calculating vegetation indices such as NDVI and GNDVI (as shown in Tables 1 to 3). After an in-depth comparative analysis of the soluble organic matter content (SOC) and the chlorophyll content in the leaves, a comprehensive and detailed consideration was carried out. Although both can reveal the physiological status of plants and their adaptation strategies to environmental changes from different dimensions, after carefully weighing various factors, including the operability of the detection technology, the direct correlation of the indicators to the core process of plant photosynthesis, and the complex and subtle interactions between them and environmental variables, the embodiment of the present invention selected chlorophyll as the key parameter for subsequent research.

[0094] Chlorophyll, as an indispensable pigment component for plant photosynthesis, has a slight fluctuation in its content that is directly related to the efficiency of plants in capturing light energy and converting energy, which in turn profoundly affects the growth rate, health status and even the evolution of the entire life cycle of plants. More importantly, the technology for measuring chlorophyll content is mature and stable, and can provide an accurate and sensitive window through which we can understand how plants respond to changes in light intensity, the supply of nutrients, and the subtle effects of various environmental pressures.

[0095] Therefore, using chlorophyll as a core analytical indicator will undoubtedly reveal the deep mechanisms of plant growth and the intricate and interdependent dynamic balance between them and the environment, providing a clearer and more accurate perspective.

[0096]

[0097] Table 1 Measured values ​​of biological crust data

[0098]

[0099] Table 2 UAV impact data extraction

[0100]

[0101] Table 3 Arc GIS Pro & ENVI calculation index

[0102] 3.Pearson correlation analysis:

[0103] ① Data preparation: Match and integrate the calculated vegetation index with the chlorophyll content, thickness and other parameters of the biological crust measured by ground sampling to form a complete data set.

[0104] ② Correlation analysis: Pycharm software was used to draw the Pearson correlation heat map (such as Figure 2 The correlation between vegetation index and key parameters of BSCs was quantitatively analyzed (as shown in Table 4). Through correlation analysis, vegetation indices that are highly correlated with BSC coverage and thickness can be selected as input features of subsequent machine learning models. According to the P value in the Pearson correlation table, indices with a P value less than 0.001 (extremely correlated) can be selected.

[0105] 4. Machine Learning Model Construction:

[0106] ① Feature selection: Based on the results of Pearson correlation analysis, vegetation indices that are highly correlated with the coverage and thickness of biological crusts were selected as input features of the machine learning model. These features can fully reflect the spectral reflectance characteristics and growth status of biological crusts.

[0107] ② Model training and verification: We built a variety of machine learning models (such as random forest RFR, support vector machine SVR, neural network BP, etc.), and used training set data and test sets to train the models. By continuously adjusting model parameters and optimizing algorithm structure, we finally selected the model with the highest prediction accuracy as the optimal model. At the same time, we used test set data to verify the model to ensure its reliability and stability in practical applications.

[0108] 5. Result verification and evaluation:

[0109] ① Model evaluation: through R 2 The prediction accuracy of the optimal model was comprehensively evaluated using indicators such as slope and RMSE. The results showed that the model had high accuracy and reliability in predicting the coverage and thickness of biocrusts.

[0110] ②Results display: The distribution map of biocrust coverage and thickness was generated by combining drone image data with the prediction results of the machine learning model (such as Figures 5 to 7 The distribution map intuitively shows the restoration effect and distribution of biological crust in the mining area, providing strong data support for ecological restoration in the mining area.

[0111] Example 3

[0112] The embodiment of the present invention provides the following alternative solutions based on embodiments 1 and 2:

[0113] 1. Replacement of UAV platforms: In addition to the DJI M3 multispectral drone, other UAV platforms with multispectral remote sensing capabilities can also be selected as alternatives. For example, the DJI M600Pro equipped with a multispectral camera. These platforms also have excellent flight stability and data collection capabilities, which can meet the needs of biological crust monitoring in mining areas. When selecting an alternative UAV platform, factors such as its flight performance, sensor configuration, data processing compatibility, and cost-effectiveness should be considered comprehensively. Ensure that the selected platform can meet the actual monitoring needs and has a high cost-effectiveness.

[0114] 2. Alternatives to machine learning algorithms: When building machine learning models, in addition to common algorithms such as random forests, support vector machines, and neural networks, other new algorithms can also be tried as alternatives. For example, convolutional neural networks (CNNs) in deep learning. These new algorithms perform well in processing complex image data and may further improve the prediction accuracy and stability of the model. When selecting alternative algorithms, factors such as the characteristics of the data, the availability of computing resources, and the prediction performance of the model should be fully considered. Ensure that the selected algorithm can fully utilize the data advantages and meet the actual application needs.

[0115] 3. Alternatives to data processing software: In terms of data processing, in addition to professional software such as Pycharm, ArcGIS Pro, and ENVI, other software or platforms with similar functions can also be considered as alternatives. Such as ERDASImagine, QGIS, R language, etc. These software or platforms also have powerful data processing and analysis capabilities and can meet the needs of biological crust monitoring in mining areas. When selecting alternative software, attention should be paid to factors such as its functionality, ease of use, data processing efficiency, and compatibility with existing workflows. Ensure that the selected software can efficiently process multispectral image data collected by drones and provide strong support for subsequent analysis.

[0116] In summary, this technical solution proposes an innovative solution based on drone multispectral remote sensing technology and machine learning algorithms to monitor biological crusts in mining areas. By integrating advanced drone platforms, multispectral sensors, ground sampling equipment, and professional data processing software, this solution achieves large-scale, rapid, and non-destructive monitoring of biological crusts in mining areas, providing strong data support for ecological restoration in mining areas.

[0117] During the implementation of the program, high-resolution image data of the surface of the mining area was first obtained using drone multispectral remote sensing technology. These data cover multiple key spectral bands such as visible light and near-infrared, and can accurately reflect the spectral reflectance characteristics of biological crusts. Subsequently, the image data was preprocessed and the vegetation index was calculated using professional data processing software, and characteristic information highly correlated with the coverage and thickness of the biological crust was extracted. On this basis, a prediction model based on a machine learning algorithm was constructed, and accurate prediction of the coverage and thickness of the biological crust was achieved by training and optimizing the model parameters. The innovation of this program lies in its unique monitoring methods, advanced prediction models, and comprehensive evaluation systems. These innovations not only improve the efficiency and accuracy of monitoring, but also make important contributions to the ecological restoration and environmental protection of mining areas. In the future, with the continuous advancement of technology and the in-depth expansion of applications, this technical solution is expected to play an important role in more fields.

[0118] Figure 1 The technical roadmap summarizes the implementation path of the entire technical solution. From data collection to result verification, each step is closely linked to ensure the integrity and feasibility of the technical solution. Tables 1 to 3 record the measured values ​​of biological crust data, UAV image data extraction, and ArcGIS Pro&ENVI calculation index results, providing a solid foundation for subsequent data analysis. Figure 2 The Pearson correlation heat map visually displays the correlation between different vegetation indices and key parameters of biological crusts through colors and values, providing an important basis for feature selection of machine learning models.

[0119] Tables 4-1 and 4-2 list the results of Pearson correlation analysis in detail, further verifying the accuracy of the heat map. Table 5 shows the fitting parameters of different machine learning models under different biocrust indicators. Figures 11 to 23 The application of different machine learning models in the inversion of biocrust coverage and thickness was demonstrated. By comparing the prediction accuracy and stability of each model, the optimal model was finally selected.

[0120]

[0121]

[0122] Table 4-1 Pearson chlorophyll correlation analysis

[0123]

[0124] Table 5-2 Pearson chlorophyll thickness correlation analysis

[0125]

[0126]

[0127] Table 5 Fitting parameters of different machine learning models under different biocrust indicators

[0128] Figures 3 to 8 Taking the phosphogypsum mine as an example, the key information of the experimental site, such as RGB images, chlorophyll value distribution, site classification, thickness classification and thickness distribution of biological crusts, is displayed. These images not only provide intuitive visual information of the experimental site, but also show the spatial distribution and thickness changes of biological crusts in the mining area through classification and thickness distribution maps, providing important data support for ecological restoration in mining areas.

[0129] The present invention has shown significant technical advantages and beneficial effects in monitoring the coverage and thickness of biological crusts in mining areas. These advantages are not only reflected in monitoring efficiency, data accuracy, visual classification and calculation, but also have a profound impact on the progress and effect of ecological restoration in mining areas.

[0130] 1. Significant improvement in monitoring efficiency and significant reduction in costs

[0131] Combined with the multispectral remote sensing technology of unmanned aerial vehicles (UAV), the present invention realizes the efficient monitoring of biological crust in mining areas. UAVs, with their fast and flexible characteristics, can cover a large area in a short time, significantly improving the monitoring efficiency. Compared with traditional manual ground monitoring methods, UAV monitoring not only reduces labor costs and time costs, but also avoids the monitoring difficulties caused by complex terrain, bad climate and other factors. According to statistics, the efficiency of monitoring using drones can be increased by more than 5 times compared with manual monitoring. This efficiency improvement directly reduces the monitoring cost. In addition, the multispectral lens of the drone can simultaneously obtain spectral information of multiple bands. This parallel data collection method further improves the monitoring efficiency and provides a strong guarantee for the continuous advancement of ecological restoration projects in mining areas.

[0132] 2. Significant improvement in data accuracy and enhanced reliability

[0133] Through correlation analysis and the construction of machine learning models, the present invention achieves accurate inversion of the coverage and thickness of biological crusts. Pearson correlation can extract key indexes that are highly correlated with the characteristics of biological crusts from massive multispectral data and establish their corresponding functional relationships in combination with machine learning. It effectively avoids errors caused by human factors. Experimental data show that the inversion accuracy of this method can reach more than 90%, which is much higher than the accuracy of traditional manual monitoring methods. This high-precision data inversion not only improves the reliability of monitoring results, but also provides a more accurate and scientific basis for ecological restoration in mining areas.

[0134] It should be understood that the size of the serial numbers of the steps in the above embodiments does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0135] It should be understood that those skilled in the art can make improvements or changes based on the above description, and all these improvements and changes should fall within the scope of protection of the appended claims of the present invention.

Claims

1. A method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data, characterized in that: The method comprises the following steps: S1. Use a drone equipped with a multispectral sensor to collect multispectral image data of the biocrust in the experimental area at a preset flight altitude and route; and simultaneously sample and measure the measured parameters of the biocrust on the ground; S2, importing the multispectral image data into mapping software for preprocessing, and calculating various vegetation indices on the preprocessed multispectral image data; S3. Match and integrate the calculated vegetation index with the measured parameters of ground sampling to form a complete data set; draw a correlation heat map of the data set, quantitatively analyze the correlation between the vegetation index and the measured parameters, compare the correlation between the vegetation index and the coverage and thickness of the biological crust, and select the vegetation index with the qualified correlation as the input feature of the machine learning model; S4. Build multiple machine learning models, divide the input features into training sets and test sets, use the training sets to train the machine learning models, adjust the model parameters, and output the machine learning model with the highest prediction accuracy as the optimal model; evaluate the optimal model using the test set and multiple evaluation indicators; S5. For the biological crusts in the mining area to be monitored, the multispectral image data of UAV is combined with the prediction results of the optimal model to generate the distribution map of the coverage and thickness of the biological crusts.

2. The method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data according to claim 1 is characterized in that: The method for collecting multispectral image data in step S1 includes: Select drones with emergency braking and obstacle avoidance, strong wind resistance, and light intensity correction functions, equipped with multispectral lenses covering multiple key spectral bands including visible light and near-infrared light; Multiple sampling points were evenly arranged in the experimental area according to the topography and distribution of biocrusts, and the location information of each sampling point was recorded; At the preset flight altitude and route, the multispectral lens carried by the drone conducts a comprehensive scan and data collection of the experimental area.

3. The method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data according to claim 1 is characterized in that: The method of sampling on the ground and measuring the measured parameters in step S1 includes: using ground sampling equipment, including a ring knife soil sampler, a tape measure, a vernier caliper, an insulated box, and a GPS locator, to collect biological crust samples and measure their parameters, including: chlorophyll content and thickness.

4. The method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data according to claim 1, characterized in that: The preprocessing method in step S2 includes: importing the multispectral image data collected by the drone into the mapping software for preprocessing, including image stitching, cropping, and denoising, to obtain a multi-band map of the experimental site.

5. The method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data according to claim 4 is characterized in that: The method for calculating the vegetation index on the pre-processed multispectral image data in step S2 includes: ArcGIS Pro and ENVI software were used to calculate the vegetation index of the preprocessed multispectral image data. The coverage, growth status and chlorophyll content of the biological crust were quantitatively evaluated by calculating the vegetation indices including NDVI and GNDVI.

6. The method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data according to claim 1, characterized in that: The method in step S3 includes: The calculated vegetation index is matched and integrated with the measured parameters of ground sampling to form a complete data set; Pycharm software was used to draw the Pearson correlation heat map, and the correlation between vegetation index and key parameters of biological crust was quantitatively analyzed. The vegetation index that was highly correlated with the coverage and thickness of biological crust was screened out as the input feature of the subsequent machine learning model. According to the P value of Pearson correlation, the vegetation index with a P value less than a certain threshold was screened out.

7. The method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data according to claim 1, characterized in that: The method in step S4 includes: Construct a variety of machine learning models, including chlorophyll BP model, GBDT model, KNN model, RFR model, RR model, SVR model, and XGB model; divide the input features obtained by feature selection into training sets and test sets, input the training set into the machine learning model, and finally select the model with the highest prediction accuracy as the optimal model by continuously adjusting the model parameters and optimizing the algorithm structure.

8. The method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data according to claim 1, characterized in that: The evaluation method in step S4 includes: using the test set data to verify the optimal model, including R 2 Evaluation indicators including , slope, and RMSE are used to comprehensively evaluate the prediction accuracy of the optimal model.

9. The method for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data according to claim 1, characterized in that: The method for generating the distribution map in step S5 includes: ArcGIS Pro geographic information system software was used to visualize the drone multispectral image data and the prediction results of the machine learning model to generate a distribution map of the coverage and thickness of the biological crust.

10. A system for monitoring the coverage and thickness of biological crusts in mining areas based on UAV multispectral data, characterized in that: include: a memory for storing executable computer programs; A processor, when used to execute an executable computer program stored in a memory, implements the method of monitoring the coverage and thickness of biological crust in a mining area based on UAV multispectral data as described in any one of claims 1 to 9.

Images