Slope vegetation coverage monitoring method based on unmanned aerial vehicle remote sensing

Abstract

The invention discloses a slope vegetation coverage monitoring method based on unmanned aerial vehicle remote sensing. The method comprises the following steps: firstly, acquiring and smoothing a slope topographic data set, and extracting an inclination angle distribution diagram; and analyzing the vegetation coverage condition in combination with the orthoimage to obtain coverage distribution data. And analyzing the relevance between the vegetation coverage and the inclination angle, and calculating vegetation growth difference distribution. And utilizing deep learning to optimize inclination angle features, and if the relevance is lower than a threshold value, introducing micro-topographic feature correction. And finally, generating a microtopography guidance map, determining a vegetation optimization layout, and completing monitoring analysis. According to the invention, accurate monitoring and optimized layout guidance of the slope vegetation restoration condition are realized, and scientific basis and technical support are provided for slope ecological restoration.

Description

Technical Field

[0001] The present invention belongs to the technical field of ecological monitoring and slope engineering, and in particular relates to a slope vegetation coverage monitoring method based on unmanned aerial vehicle (UAV) remote sensing. Background Art

[0002] Monitoring slope vegetation cover is a core research area in ecological restoration and geological disaster prevention. Its importance lies in its direct impact on slope stability, soil erosion control, and sustainable ecosystem development. The widespread application of drone remote sensing technology has enabled unprecedented data acquisition capabilities in this area, providing critical support for analyzing the correlation between slope vegetation and terrain characteristics. However, current research and applications still face significant challenges, and technological breakthroughs are urgently needed to improve monitoring accuracy and practicality.

[0003] Existing methods for monitoring slope vegetation cover often rely on traditional remote sensing imagery or manual measurements. While these methods can provide a certain degree of vegetation distribution, they often lack the ability to accurately depict the microtopographic characteristics of the slope. In particular, in complex terrain, the extraction accuracy of geometric parameters such as inclination angle is insufficient, resulting in unreliable analysis of the correlation between vegetation cover and topography. Furthermore, the data processing process is often cumbersome and the degree of automation is low, making it difficult to meet the needs of large-scale slope monitoring.

[0004] In response to the above problems, it is urgent to propose a slope vegetation cover monitoring method based on UAV remote sensing. Summary of the Invention

[0005] In order to solve the above technical problems, the present invention proposes a slope vegetation cover monitoring method based on UAV remote sensing to solve the problems existing in the above-mentioned prior art.

[0006] To achieve the above objectives, the present invention provides a slope vegetation cover monitoring method based on UAV remote sensing, the method comprising:

[0007] Obtain slope terrain datasets based on UAV remote sensing technology and perform smoothing;

[0008] Based on the smoothed terrain data set, the slope inclination characteristics are extracted to obtain the inclination distribution map;

[0009] According to the inclination distribution map, combined with orthophoto analysis, the vegetation coverage status is extracted to obtain the vegetation coverage distribution data;

[0010] Based on the vegetation cover distribution data and the inclination distribution map, the correlation between vegetation cover and inclination is analyzed to obtain the vegetation cover ratio within different inclination ranges. The vegetation growth difference is calculated using statistical methods to obtain the difference distribution results.

[0011] Based on the difference distribution results, a deep learning algorithm is used to optimize the slope inclination characteristics to obtain enhanced slope inclination characteristic data;

[0012] Based on the enhanced inclination feature data, the correlation between vegetation coverage and inclination is recalculated. If the correlation is lower than the preset threshold, micro-topography features are introduced for correction to obtain the corrected correlation distribution.

[0013] Through the corrected correlation distribution, automated calculation is used to generate a micro-topography guidance map for slope vegetation restoration, determine the optimal vegetation layout in each area, and obtain the final monitoring and analysis results.

[0014] Optionally, the acquiring of a slope terrain dataset based on UAV remote sensing technology and smoothing the dataset may include:

[0015] The original data of the slope is obtained based on UAV remote sensing technology, and the features of the original data are extracted using preset sampling rules to obtain the initial data set;

[0016] Processing the initial data set through a digital elevation model to generate a terrain data set for the slope area;

[0017] The terrain dataset is smoothed using a Gaussian filtering algorithm to obtain a smoothed terrain dataset.

[0018] Optionally, extracting slope inclination features based on the smoothed terrain dataset to obtain an inclination distribution map includes:

[0019] For the smoothed terrain dataset, the grid cells are divided, the slope value of each grid cell is calculated, and the slope distribution information is obtained;

[0020] The inclination angle feature is extracted from the slope distribution information. If the inclination angle feature exceeds the preset threshold, it is marked as a high-relief area, and then a inclination angle distribution map is generated.

[0021] Optionally, the extracting vegetation coverage status based on the inclination distribution map in combination with orthophoto analysis to obtain vegetation coverage distribution data includes:

[0022] According to the inclination distribution map, terrain feature data is obtained, and orthophoto images are fused to generate comprehensive terrain image data;

[0023] By integrating terrain image data and using spectral reflectance characteristics, we can distinguish between vegetation areas and non-vegetation areas and obtain preliminary regional division results.

[0024] Based on the preliminary regional division results, reflectance characteristics are extracted to generate vegetation cover distribution data.

[0025] Optionally, the analysis of the correlation between vegetation coverage and inclination based on the vegetation coverage distribution data and the inclination distribution map to obtain vegetation coverage ratios within different inclination ranges and the use of statistical methods to calculate vegetation growth differences to obtain difference distribution results includes:

[0026] The vegetation coverage distribution data and the inclination distribution map are spatially superimposed so that each pixel contains both vegetation coverage and inclination information;

[0027] Classify the data according to the preset inclination range and count the vegetation coverage ratio within each inclination range;

[0028] By comparing the vegetation coverage ratio within different inclination ranges, the vegetation growth differences are calculated and a difference distribution map is generated.

[0029] Optionally, the step of optimizing the slope inclination angle characteristics based on the difference distribution result using a deep learning algorithm to obtain enhanced slope inclination angle characteristic data includes:

[0030] Convolutional neural networks are used to optimize the resolution of digital elevation models to obtain enhanced high-resolution elevation data.

[0031] Obtain the inclination distribution from the enhanced high-resolution elevation data, divide the high-relief areas into different inclination intervals, and determine the boundaries of each interval;

[0032] Extract characteristic parameters based on the inclination distribution. If the characteristic parameters exceed the preset threshold, adjust the data smoothness through interpolation to obtain smoothed inclination features.

[0033] According to the smoothed inclination characteristics, combined with the vegetation cover data, the coverage ratio of each inclination interval is calculated to determine the trend of the ratio change;

[0034] In view of the trend of proportion change, statistical test methods were used to analyze the difference distribution and obtain the results of significant difference;

[0035] The convolutional neural network parameters are adjusted based on the significant difference results, the digital elevation model resolution enhancement process is optimized, and the enhanced slope inclination characteristic data is obtained.

[0036] Optionally, the corrected correlation distribution is used to generate a micro-topography guidance map for slope vegetation restoration using automated calculations to determine the optimal vegetation layout for each area and obtain the final monitoring and analysis results, including:

[0037] Based on the corrected correlation distribution data, a clustering algorithm is used to classify the slope areas to obtain preliminary zoning results; through a terrain feature extraction algorithm, micro-topography units are identified from the preliminary zoning results to obtain a micro-topography distribution map; for each micro-topography unit, the vegetation type suitable for planting is calculated based on the vegetation growth condition model to determine the vegetation configuration plan; if the vegetation configuration plan differs from the surrounding ecosystem, the vegetation type combination is adjusted to obtain an optimized vegetation layout; a geographic information system is used to draw a micro-topography guidance map for slope vegetation restoration, showing the optimized vegetation layout of each area; based on the micro-topography guidance map for slope vegetation restoration, the slope vegetation coverage and species diversity are calculated to form the final monitoring and analysis results.

[0038] The present invention also provides a computer device, comprising a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method.

[0039] The present invention also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method when executed by a processor.

[0040] The present invention also provides a computer program product, comprising a computer program, which implements the steps of the method when executed by a processor.

[0041] Compared with the prior art, the present invention has the following advantages and technical effects:

[0042] The present invention discloses a slope vegetation restoration monitoring method based on UAV remote sensing technology. The method generates three-dimensional terrain information by collecting raw data, calculates the slope value and extracts the inclination angle characteristics after filtering, and determines the vegetation growth differences within different inclination angle ranges in combination with vegetation cover analysis. For high-undulation areas, the present invention uses a deep learning algorithm to optimize the resolution of the digital elevation model and improve the accuracy of inclination angle extraction. By introducing micro-topography features to correct the correlation between vegetation cover and inclination angle, a micro-topography guidance map for slope vegetation restoration is finally generated. This method realizes the precise monitoring of slope vegetation restoration status and optimized layout guidance, providing a scientific basis and technical support for slope ecological restoration. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] The accompanying drawings, which constitute part of this application, are intended to provide a further understanding of this application. The exemplary embodiments and descriptions of this application are intended to explain this application and do not constitute an improper limitation on this application. In the accompanying drawings:

[0044] Figure 1 Schematic diagram of a method flow in an embodiment of the present invention;

[0045] Figure 2 A flow chart for generating an evaluation report according to an embodiment of the present invention;

[0046] Figure 3 This is a flow chart of vegetation coverage data generation according to an embodiment of the present invention. DETAILED DESCRIPTION

[0047] It should be noted that, in the absence of conflict, the embodiments and features of the embodiments in this application can be combined with each other. The present application will be described in detail below with reference to the accompanying drawings and in combination with the embodiments.

[0048] It should be noted that the steps shown in the flowcharts of the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions, and that, although a logical order is shown in the flowcharts, in some cases, the steps shown or described can be executed in an order different from that shown here.

[0049] Example 1

[0050] like Figure 1-3 As shown, this embodiment provides a slope vegetation coverage monitoring method based on UAV remote sensing, the method comprising:

[0051] Obtain slope terrain datasets based on UAV remote sensing technology and perform smoothing;

[0052] Based on the smoothed terrain data set, the slope inclination characteristics are extracted to obtain the inclination distribution map;

[0053] According to the inclination distribution map, combined with orthophoto analysis, the vegetation coverage status is extracted to obtain the vegetation coverage distribution data;

[0054] Based on the vegetation cover distribution data and the inclination distribution map, the correlation between vegetation cover and inclination is analyzed to obtain the vegetation cover ratio within different inclination ranges. The vegetation growth difference is calculated using statistical methods to obtain the difference distribution results.

[0055] Based on the difference distribution results, a deep learning algorithm is used to optimize the slope inclination characteristics to obtain enhanced slope inclination characteristic data;

[0056] Based on the enhanced inclination feature data, the correlation between vegetation coverage and inclination is recalculated. If the correlation is lower than the preset threshold, micro-topography features are introduced for correction to obtain the corrected correlation distribution.

[0057] Through the corrected correlation distribution, automated calculation is used to generate a micro-topography guidance map for slope vegetation restoration, determine the optimal vegetation layout in each area, and obtain the final monitoring and analysis results.

[0058] The method of obtaining a slope terrain dataset based on UAV remote sensing technology and performing smoothing processing may include:

[0059] The original data of the slope is obtained based on UAV remote sensing technology, and the features of the original data are extracted using preset sampling rules to obtain an initial data set; the initial data set is processed using a digital elevation model to generate a terrain data set of the slope area; the terrain data set is smoothed using a Gaussian filter algorithm to obtain a smoothed terrain data set.

[0060] As a specific implementation method, obtaining raw data obtained through remote sensing using drone technology involves using sensors carried by drones, such as optical cameras and lidar, to collect information such as the height and spectrum of the ground surface. For example, in a slope monitoring scenario, a drone flies over the target area, and the lidar emits tens of thousands of pulses per second, recording the time difference of the reflected signal from the ground surface, thereby generating point cloud data. Assuming a slope area of 5,000 square meters and a drone flying at an altitude of 100 meters, the raw point cloud collected may contain millions of coordinate points. These data reflect the preliminary outline of the terrain, but may introduce noise due to wind speed, equipment vibration, etc.

[0061] Data features are extracted using preset sampling rules, with the goal of filtering out representative information from massive amounts of raw data. Understandably, sampling rules may be based on spatial intervals or feature significance. For example, in point cloud data, a sampling point is taken every 0.5 meters, and points with sudden changes in height are extracted as characteristic points of the slope. Assuming the original point cloud has 2 million points, after sampling, this is reduced to 200,000 points, preserving key information about the slope contour. This approach reduces the computational burden of subsequent processing while highlighting key areas of terrain change, effectively improving data processing efficiency.

[0062] Processing the initial data set through a digital elevation model to generate three-dimensional terrain information of the slope area is the core step in converting two-dimensional sampling data into a three-dimensional description. Specifically, the digital elevation model uses interpolation algorithms, such as triangulation interpolation, to connect discrete sampling points into a continuous terrain surface. In one possible implementation, for the slope area, the three-dimensional grid with a resolution of 1 meter generated after interpolation clearly displays information such as slope and height difference. For example, for a slope with a slope of 30 degrees, the three-dimensional terrain description can intuitively show the process of the height gradually changing from 50 meters to 80 meters. This three-dimensional information provides an intuitive visual basis for subsequent analysis and provides reliable data support for engineering design.

[0063] In order to effectively remove outliers by using a Gaussian filter algorithm to smooth the data noise in the three-dimensional terrain description, it is necessary to use a weighted average method to make each data point be affected by the surrounding points, and the weight decays with distance. In one embodiment, assuming that the height of a certain point suddenly changes abnormally to 85 meters, while the average of the surrounding points is 60 meters, the point may be smoothed to 62 meters after Gaussian filtering. More specifically, the filter window is set to a 3x3 grid with a standard deviation of 1 meter. After processing, the terrain surface is smoother and more natural. This denoising method not only improves the credibility of the data, but also enhances the visualization of the three-dimensional terrain and avoids misjudgment due to noise.

[0064] Furthermore, denoised terrain information can be used for slope stability analysis. For example, by analyzing the smoothed three-dimensional terrain, engineers can identify potential landslide areas with slopes exceeding 45 degrees and assess the risks by combining it with geological data. In one possible scenario, a slope area, after processing, reveals a steep section 50 meters long with a slope of 50 degrees, indicating the need for reinforcement measures. This technical effect significantly improves the accuracy of slope monitoring and provides a scientific basis for safety management. In addition, the denoised data can be used to generate high-precision engineering drawings to further support construction planning.

[0065] In one embodiment, if the slope area involves vegetation cover, the sampling rules can be adjusted to prioritize the extraction of exposed surface points, and Gaussian filtering can be combined with multi-scale analysis to process surface and vegetation noise separately. For example, if the point cloud of a certain area is mixed with interference points with trees 10 meters high, the influence of the tree height is eliminated after filtering, and the terrain information is closer to the actual surface. This diversified processing method not only ensures the integrity of the solution, but also enriches the application scenarios through optional implementations, reflecting the flexibility and practicality of the technology.

[0066] The extracting of slope inclination features based on the smoothed terrain data set to obtain an inclination distribution map may include:

[0067] The smoothed terrain dataset is divided into grid cells, and the slope value of each grid cell is calculated to obtain the slope distribution information. The inclination angle feature is extracted from the slope distribution information. If the inclination angle feature exceeds the preset threshold, it is marked as a high-relief area, and then a inclination angle distribution map is generated.

[0068] As a specific implementation method, the process of dividing the smoothed data into grid units and calculating the slope value can be understood as decomposing the continuous terrain surface into discrete analysis units in order to describe the terrain changes more finely. For example, assuming that the smoothed slope area is 5,000 square meters, it can be divided into square grids with a side length of 5 meters, generating a total of 200 grid units. Each grid calculates the slope value based on the height difference and horizontal distance of its internal data points. For example, the height within a grid changes from 55 meters to 60 meters, and the horizontal distance is 5 meters. Through simple geometric relationships, it can be seen that the slope is about 45 degrees. This division method facilitates subsequent feature extraction and can reflect the changing laws of local terrain features.

[0069] In one possible implementation, when extracting inclination features from slope distribution information, inclination description data can be generated by counting the slope values of each grid. Specifically, the inclination feature can be the average value, maximum value or rate of change of the slope. For example, among the 200 grids in a certain area, the slope value range is between 10 degrees and 50 degrees, and the maximum slope of 50 degrees is extracted as the key inclination feature. This method highlights the high undulations of the terrain and provides a data basis for subsequent analysis. It should be noted that the generation of inclination description data can also reflect the overall trend of the slope and help identify potential unstable areas. The process of setting a preset threshold for the inclination description data and marking the high undulation areas.

[0070] Furthermore, thresholds can be defined according to engineering requirements. For example, assuming that the preset inclination threshold is 40 degrees, all grids with a slope of more than 40 degrees are marked as high-relief areas. In one embodiment, there may be 20 grids out of 200 with a slope of more than 40 degrees, and these grids are marked. This marking method intuitively distinguishes between the flat areas and high-risk areas of the terrain, laying the foundation for further analysis. The marking results can also provide a reference for subsequent spatial positioning. A high-relief area distribution map is generated based on the marking results. When determining the range, it can be understood as visualizing the marked grids in space. For example, in a slope monitoring scenario, 20 high-relief grids may be concentrated in the middle of the slope, forming an area 50 meters long and 20 meters wide. For example, these grids are marked with eye-catching colors on the distribution map to intuitively show the concentration of high-relief areas.

[0071] In one possible implementation, the distribution map can also superimpose height information, such as the average height of the high-relief area is 70 meters, to further enrich the description content. This distribution map provides convenience for engineering personnel to judge key areas. Specifically, during the generation of the distribution map of the high-relief area, the spatial relationship of the terrain data can be combined for optimization. For example, adjacent high-relief grids can be merged into a larger area to facilitate the analysis of their coherence. In one embodiment, if two marked grids are adjacent, they can be regarded as a whole, and the range is expanded to 10 meters × 5 meters. This processing method enhances the practicality of the distribution map and avoids the influence of overly scattered marks on judgment.

[0072] Furthermore, the distribution map can be compared with the original point cloud data to verify the accuracy of the marking and ensure the reliability of the results. It should be noted that the entire process from slope distribution information to the scope of high-relief areas forms a complete chain from data processing to spatial analysis. For example, from initial grid division to inclination feature extraction, to threshold marking and distribution map generation, each step is closely connected. This logical progression ensures the systematic nature of the solution, while enriching the application possibilities through diverse implementation methods and providing more comprehensive technical support for slope monitoring.

[0073] The extraction of vegetation coverage status based on the inclination distribution map in combination with orthophoto analysis to obtain vegetation coverage distribution data may be implemented as follows:

[0074] Based on the inclination distribution map, terrain feature data is obtained, and orthophotos are fused to generate comprehensive terrain image data. Through the comprehensive terrain image data, spectral reflectance characteristics are used to distinguish vegetation areas from non-vegetation areas to obtain preliminary regional division results. Based on the preliminary regional division results, reflectance characteristics are extracted to generate vegetation cover distribution data.

[0075] As a specific implementation, obtaining terrain feature data based on a dip distribution map can be understood as extracting key parameters describing terrain undulations from slope information. For example, if a dip distribution map for a slope area shows slope values ranging from 5 to 60 degrees, characteristic data such as an average slope of 30 degrees and a maximum slope of 60 degrees can be extracted. This data provides the basis for subsequent fusion.

[0076] In one possible implementation, when fusing orthophotos, terrain feature data can be associated with image pixels. Specifically, orthophotos provide visual information about the surface, such as color and texture, while inclination data adds details about elevation changes. For example, consider a 50m x 50m slope area whose orthophoto shows green vegetation and exposed rock. By overlaying the inclination data, a comprehensive terrain image can be generated, visually demonstrating the relationship between slope and surface features. After generating the comprehensive terrain image data, spectral reflectance characteristics can be used to distinguish between vegetated and non-vegetated areas. This analysis can be based on the reflectance characteristics of different objects in the image. It should be noted that spectral reflectance characteristics refer to the difference in reflectance intensity of surface objects to light in different wavelengths. For example, vegetation typically has high reflectance in the near-infrared band, while rock or soil has lower reflectance. In one embodiment, a comprehensive terrain image data set covers 200 grids, 100 of which exhibit high near-infrared reflectance and are preliminarily identified as vegetation areas. The remaining grids have lower reflectance and are classified as non-vegetated areas. This classification provides a preliminary basis for subsequent analysis.

[0077] When extracting the reflection characteristics based on the preliminary regional division results, the descriptive ability of the data can be further refined. Preferably, specific indicators can be extracted from the spectral data of each grid, such as the average reflection value of the vegetation area or the reflection fluctuation range of the non-vegetation area. For example, the near-infrared reflection value of a vegetation grid is 0.8, while that of a non-vegetation grid is only 0.2. By comparing these values, the rationality of the division can be verified. In one possible implementation method, the division can also be adjusted in combination with the terrain characteristics. For example, even if the reflection value of a grid with a slope of more than 50 degrees is high, it may not be suitable for vegetation growth due to terrain restrictions and needs to be reclassified. This method improves the accuracy of regional division.

[0078] When generating vegetation coverage distribution data, the extracted reflection characteristics can be converted into spatial distribution information. Specifically, by counting the reflection characteristics of each grid, a distribution map covering the entire area can be generated. For example, of the 200 grids in a slope area, 120 grids are identified as vegetation areas, which are concentrated at the bottom with a gentler slope, forming an area of 60 meters long and 20 meters wide. It can be understood that these data not only reflect the spatial range of vegetation, but also provide support for ecological monitoring. In one embodiment, the distribution map can mark the density changes of vegetation areas, such as the vegetation coverage rate of the bottom grid is 80%, and the top is only 20%, which intuitively shows the distribution trend.

[0079] It's important to note that the processes of extracting reflectance characteristics and generating distribution data logically support each other, ensuring coherence from image fusion to regional delineation. For example, a high-angle, non-vegetated grid can show significant differences in reflectance when compared to adjacent vegetated grids, further confirming the influence of topography on vegetation distribution. This analysis can also provide a reference for slope stability assessment, as sparsely vegetated, high-angle areas may indicate potential risk points. This multifaceted approach enhances the practicality of the data.

[0080] The method can be implemented by analyzing the correlation between vegetation coverage and inclination based on vegetation coverage distribution data and inclination distribution map, obtaining vegetation coverage ratios within different inclination ranges, and calculating vegetation growth differences using statistical methods to obtain difference distribution results, including:

[0081] The vegetation coverage distribution data and the inclination distribution map are spatially superimposed so that each pixel contains both vegetation coverage and inclination information; the data are classified according to the preset inclination range, and the vegetation coverage ratio within each inclination range is counted; by comparing the vegetation coverage ratios within different inclination ranges, the vegetation growth differences are calculated and a difference distribution map is generated.

[0082] As a specific implementation method, when the vegetation cover distribution data and the inclination distribution map are spatially superimposed, it can be understood as spatially aligning the two types of data through a geographic information system. For example, in an area of 100 meters by 100 meters, the vegetation cover distribution data records the coverage of each pixel in the form of a grid, such as a value from 0 to 1, and the inclination distribution map provides the slope information of the corresponding grid, such as 0 degrees to 70 degrees. After spatial superposition, each pixel has both coverage and inclination attributes. For example, the coverage of a pixel is 0.7 and the inclination is 25 degrees. This method lays the foundation for subsequent classification.

[0083] In one possible implementation, data can be classified according to preset inclination angle ranges, with slopes divided into several intervals, such as 0-15 degrees, 15-30 degrees, 30-45 degrees, and above 45 degrees. Specifically, among 1,000 pixels in a slope area, 400 pixels fall within the 0-15 degree range, with an average vegetation coverage of 0.8. Meanwhile, 300 pixels fall within the 30-45 degree range, with coverage dropping to 0.3. This classification intuitively reflects the preliminary relationship between vegetation coverage and slope.

[0084] It should be noted that when counting the vegetation coverage ratio within each inclination angle range, the average coverage of each interval can be calculated. For example, the total number of pixels in the range of 15-30 degrees is 200, of which 150 pixels have a coverage greater than 0.5, accounting for 75%. In one embodiment, if the pixel coverage in the range above 45 degrees is generally less than 0.2, it indicates that the vegetation in the high-inclination area is sparse. This statistical method provides data support for subsequent difference analysis.

[0085] Furthermore, when comparing the vegetation coverage ratios within different inclination ranges, the differences can be inferred from multiple aspects. It is understandable that the coverage in low-inclination areas (such as 0-15 degrees) is high, which may be due to excellent water and soil conditions, while the coverage in high-inclination areas (such as above 45 degrees) is low, which may be affected by soil erosion. For example, the coverage ratio of the 0-15 degree interval in an area is 80%, and that above 45 degrees is only 10%, a difference of 70%. This comparison reveals the impact of terrain on vegetation.

[0086] In one possible implementation, statistical methods are used to test the significance of differences, such as simple mean comparisons or hypothesis testing. For example, consider two slope angle ranges: 15-30 degrees and 30-45 degrees, with coverage rates of 60% and 35%, respectively. Statistical analysis can be used to determine whether the differences are due to random factors. This approach enhances the confidence of the analysis.

[0087] Specifically, visualizing the difference results can generate a difference distribution map. For example, within a 200-grid area, the 0-15 degree area is marked green, indicating high coverage, while the area above 45 degrees is marked red, indicating low coverage. This visualization clearly demonstrates the spatial patterns of vegetation changes with inclination angle.

[0088] In one embodiment, the difference map shows that vegetation is concentrated at the bottom of the slope and sparse at the top, reflecting the correlation between water distribution and topography. It is understandable that this analysis process gradually reveals the relationship between vegetation and topography through spatial overlay, classification statistics, and visualization. Preferably, the difference distribution map can also assist in ecological planning, such as identifying high-angle exposed areas that need restoration. This multi-level deliberation approach ensures the rigor and practicality of the analysis.

[0089] The method of optimizing the slope inclination angle characteristics based on the difference distribution results by using a deep learning algorithm to obtain enhanced slope inclination angle characteristic data may include:

[0090] A convolutional neural network is used to optimize the resolution of the digital elevation model to obtain enhanced high-resolution elevation data; the inclination distribution is obtained from the enhanced high-resolution elevation data, and different inclination intervals are divided for high-relief areas, and the boundaries of each interval are determined; characteristic parameters are extracted based on the inclination distribution. If the characteristic parameters exceed the preset threshold, the data smoothness is adjusted through interpolation to obtain smoothed inclination characteristics; based on the smoothed inclination characteristics, the coverage ratio of each inclination interval is calculated in combination with vegetation cover data to determine the trend of proportion change; based on the proportion change trend, the difference distribution is analyzed using statistical test methods to obtain the difference significance results; the convolutional neural network parameters are adjusted based on the difference significance results, the digital elevation model resolution enhancement process is optimized, and enhanced slope inclination characteristic data are obtained.

[0091] As a specific implementation method, when extracting inclination data from high-relief areas, it can be understood as identifying areas with significant terrain changes from the digital elevation model. For example, the elevation data of a mountainous terrain covers an area of 1000 meters by 1000 meters with a resolution of 10 meters. High-relief areas may appear as steep slopes with elevation values rising rapidly from 200 meters to 500 meters. When extracting inclination data, the slope is first calculated by the elevation difference between adjacent pixels to obtain a preliminary inclination distribution.

[0092] In one possible implementation, a convolutional neural network (CNN) is used to optimize the resolution of digital elevation models, increasing the original 10-meter resolution to 2 meters. Specifically, the network learns terrain texture features to predict elevation details at high resolution. For example, the original data for a steep slope shows only smooth transitions, but after enhancement, the subtle undulations on the slope are clearly visible. This increased resolution lays the foundation for subsequent, precise analysis.

[0093] It should be noted that when obtaining the precise inclination distribution from the enhanced high-resolution elevation data, the slope can be recalculated based on a 2-meter grid. For example, in a local area of 50 meters by 50 meters, the average inclination before enhancement was 20 degrees, and after enhancement, the local maximum inclination was 40 degrees. When dividing the inclination intervals for high-relief areas, 0-10 degrees, 10-20 degrees, 20-30 degrees, and above 30 degrees can be set as boundaries to clearly distinguish terrain features.

[0094] Specifically, when extracting characteristic parameters based on the precise inclination distribution, the slope change rate can be selected as an indicator. If the slope change rate in the 20-30 degree range in a certain area exceeds 0.5, it exceeds the preset threshold, indicating that the data may be too sharp.

[0095] In one embodiment, the data smoothness is adjusted by an interpolation method, for example, spline interpolation is applied to areas above 30 degrees to make the inclination transition more natural, and the rate of change after smoothing is reduced to below 0.3. After obtaining the smoothed inclination characteristics, the average value of each interval can be calculated when calculating the coverage ratio in combination with the vegetation coverage data. For example, in an area of 100 grids, the coverage ratio of the 10-20 degree interval is 70%, and it is only 20% above 30 degrees. When judging the trend of the ratio change, it can be observed that the coverage ratio gradually decreases with the increase of the inclination angle. This trend reflects the constraints of the terrain on the vegetation distribution.

[0096] In one possible implementation, statistical testing can be used to analyze the distribution of differences. For example, the coverage ratios for the 10-20 degree range and the 20-30 degree range can be compared. For example, if the coverage ratio for the 10-20 degree range is 70% and the coverage ratio for the 20-30 degree range is 50%, the test confirms a significant difference. This significant result provides a basis for optimizing the model.

[0097] It's understandable that adjusting deep learning model parameters can increase the depth of the convolutional layer, further improving the accuracy of resolution enhancement. For example, when the optimized model processes the same area, the 2-meter resolution data reveals more detailed inclination features, with clearer boundaries in areas above 30 degrees. After acquiring enhanced inclination feature data, the vegetation cover distribution in a steep slope area shifts from a smooth transition to a more realistic layered pattern. This optimization enhances data usability and provides more reliable support for subsequent ecological analysis.

[0098] It is feasible to recalculate the correlation between vegetation coverage and inclination based on the enhanced inclination feature data. If the correlation is lower than the preset threshold, the micro-topography feature is introduced for correction to obtain the corrected correlation distribution, including:

[0099] The enhanced inclination feature data is used to extract the correlation between vegetation cover and inclination. A preliminary determination is made as to whether the correlation is below a preset threshold. If this is the case, micro-geomorphological data is retrieved for parameter correction, resulting in a corrected correlation distribution.

[0100] As a specific implementation method, when the correlation value between vegetation coverage and inclination is extracted through enhanced inclination feature data, it can be understood as analyzing the degree of influence of terrain slope on vegetation distribution. For example, in a mountainous scene, based on a digital elevation model with a resolution of 2 meters, the inclination data covers an area of 1000 meters × 1000 meters. The vegetation coverage status is extracted through remote sensing images to obtain the coverage percentage of each grid. The correlation value can be quantified by statistically analyzing the trend of vegetation coverage ratio changing with inclination. For example, the coverage ratio of areas below 10 degrees is 80%, and it drops to 20% above 30 degrees, indicating that there is a negative correlation between the two.

[0101] In one possible implementation, when determining whether the correlation value is below a preset threshold, a threshold of 0.6 can be set, indicating that the correlation between inclination and vegetation cover must reach a certain level of strength. If the calculated correlation value is 0.4, which is lower than 0.6, the preliminary judgment result is insufficient correlation. This judgment helps identify whether terrain characteristics are the primary limiting factor for vegetation distribution, providing a basis for subsequent analysis.

[0102] It should be noted that if the correlation value is lower than the preset threshold, it is necessary to obtain micro-geomorphological data for parameter correction. Micro-geomorphological data includes information such as local surface undulations, slope direction, and small gullies. For example, in a 50m x 50m steep slope area, the enhanced inclination data shows an average inclination of 35 degrees and a vegetation coverage ratio of only 25%. However, through micro-geomorphological analysis, it was found that the coverage ratio of small areas facing south increased to 40% due to sufficient sunlight. This difference suggests that single inclination data may ignore local terrain details. After correction, the correlation value increased from 0.4 to 0.65, exceeding the threshold and reflecting a more realistic relationship.

[0103] Specifically, microtopography data can be acquired through high-precision LiDAR scanning, supplementing the details in the digital elevation model. The original inclination distribution of a steep slope area shows a smooth transition, but the addition of microtopography reveals tiny steps along the slope. Due to soil accumulation in these step areas, the vegetation cover increases from 20% to 35%. The corrected correlation distribution more closely matches field observations. This approach improves the interpretability of the data.

[0104] In one embodiment, when correcting the correlation distribution, water flow data can be combined to further enrich the analysis. For example, the slope bottom has a 15-degree inclination and a 60% coverage ratio, but the microtopography indicates a small water catchment area with abundant water, resulting in an actual coverage ratio of 75%. Through parameter correction, the correlation value is adjusted from 0.5 to 0.7, reflecting the synergistic effect of water and inclination. This adjustment provides a more reliable reference for ecological planning.

[0105] Furthermore, the corrected correlation distribution can be used to guide the design of vegetation restoration plans. For example, in an area with an inclination above 30 degrees, the corrected correlation value was 0.68, indicating that the inclination strongly restricts vegetation. However, micromorphological analysis revealed that localized flat zones are suitable for planting. Targeted restoration measures could increase the coverage ratio to 50%, enhancing ecological stability.

[0106] As can be seen, this analysis process, through the fusion of multidimensional data, gradually reveals the deep connection between inclination and vegetation. For example, for a 100-meter-by-100-meter area of high relief, the initial correlation value was 0.3, below the threshold of 0.6. After adding micro-geomorphology and moisture data, the correlation value rose to 0.72, indicating that the corrected results are more realistic and provide solid support for the study of terrain-vegetation relationships. The application value of this method lies in optimizing resource allocation and improving analytical accuracy.

[0107] It is feasible to generate a micro-topography guidance map for slope vegetation restoration by using automated calculation based on the corrected correlation distribution, determine the optimal vegetation layout for each area, and obtain the final monitoring and analysis results, including:

[0108] Based on the corrected correlation distribution data, a clustering algorithm is used to classify the slope areas to obtain preliminary zoning results; through a terrain feature extraction algorithm, micro-topography units are identified from the preliminary zoning results to obtain a micro-topography distribution map; for each micro-topography unit, the vegetation type suitable for planting is calculated based on the vegetation growth condition model to determine the vegetation configuration plan; if the vegetation configuration plan differs from the surrounding ecosystem, the vegetation type combination is adjusted to obtain an optimized vegetation layout; a geographic information system is used to draw a micro-topography guidance map for slope vegetation restoration, showing the optimized vegetation layout of each area; based on the micro-topography guidance map for slope vegetation restoration, the slope vegetation coverage and species diversity are calculated to form the final monitoring and analysis results.

[0109] As a specific implementation method, when a clustering algorithm is used to classify slope areas based on the corrected correlation distribution data, it can be understood as revealing the potential patterns of terrain and vegetation through data grouping. For example, on a 1000m x 1000m mountain slope, the correlation distribution data shows that the correlation between inclination and vegetation coverage has been corrected to 0.7. Using the K-means clustering algorithm, the slopes are divided into three categories: low inclination area (0-15 degrees), medium inclination area (15-30 degrees) and high inclination area (above 30 degrees), corresponding to coverage rates of 70%, 45% and 20%, respectively. This classification reflects the differentiated effects of terrain gradients on vegetation.

[0110] In one possible implementation, when extracting micro-topography units from the zoning results, local details can be identified through a terrain feature extraction algorithm. Specifically, based on a 2-meter resolution digital elevation model, the elevation change rate and slope direction are analyzed to divide small units of 50 meters by 50 meters. For example, a unit in a low-inclination area is identified as a "gentle platform" in the micro-topography because of its south-facing slope and gentle surface, with a coverage rate of 75%, while the adjacent shady slope unit is only 60%. This distribution map intuitively shows the diversity of micro-topography. For each micro-topography unit, the suitable type is calculated based on the vegetation growth condition model.

[0111] It should be noted that the model can integrate light, water, and soil conditions. For example, gently sloping units with ample light and moderate water are suitable for growing drought-tolerant shrubs such as vitex, while steep slope units with high inclination angles are more suitable for mosses due to their thin soil.

[0112] In one embodiment, if the configuration scheme is inconsistent with the surrounding ecosystem, such as the shrubs conflict with the surrounding grassland ecology, it is adjusted to mixed herbs and low shrubs to form a transition zone to improve ecological compatibility.

[0113] Furthermore, when using a geographic information system to create a guidance map, the slope can be divided into a grid, with the vegetation type of each microtopographic unit labeled. For example, a 200-meter-long steep slope could have shrubs on the southern terrace and mosses on the northern steep slope. Overlaying the slope aspect and coverage data facilitates intuitive planning. This approach, understandably, improves the scientific nature of the layout.

[0114] In one embodiment, coverage and species diversity can be calculated when calculating slope vegetation indicators. For example, after optimizing the layout, coverage increased from the initial 40% to 65%, and the number of species increased from 3 to 6, reflecting the restoration effect. When visualizing the monitoring results, for example, a heat map can be used to display the coverage distribution, with red indicating high coverage areas and blue indicating low coverage areas. The report is supplemented with a bar chart to compare species diversity before and after optimization. This method allows managers to quickly evaluate the results and provides a basis for subsequent adjustments.

[0115] Specifically, when adjusting vegetation combinations, optimization plans can be validated through field trials. For example, mosses were initially used in high-angle areas, but localized water shortages were detected. Drought-tolerant herbs were then incorporated, increasing the coverage from 20% to 35%. This flexibility enhances the adaptability of the plan and provides more reliable support for slope ecological restoration.

[0116] Example 2

[0117] This embodiment further provides a computer device, including a memory, a processor, and a computer program stored in the memory, wherein the processor executes the computer program to implement the steps of the method.

[0118] Example 3

[0119] This embodiment also provides a computer-readable storage medium having a computer program stored thereon, which implements the steps of the method when executed by a processor.

[0120] Example 4

[0121] This embodiment also provides a computer program product, including a computer program, which implements the steps of the method when executed by a processor.

[0122] The above are merely preferred embodiments of the present application, but the scope of protection of the present application is not limited thereto. Any changes or substitutions that can be easily conceived by a person skilled in the art within the technical scope disclosed in this application should be included in the scope of protection of the present application. Therefore, the scope of protection of the present application should be based on the scope of protection of the claims.

Claims

1. A slope vegetation cover monitoring method based on UAV remote sensing, characterized in that: The method comprises: Obtain slope terrain datasets based on UAV remote sensing technology and perform smoothing; Based on the smoothed terrain data set, the slope inclination characteristics are extracted to obtain the inclination distribution map; According to the inclination distribution map, combined with orthophoto analysis, the vegetation coverage status is extracted to obtain the vegetation coverage distribution data; Based on the vegetation cover distribution data and the inclination distribution map, the correlation between vegetation cover and inclination is analyzed to obtain the vegetation cover ratio within different inclination ranges. The vegetation growth difference is calculated using statistical methods to obtain the difference distribution results. Based on the difference distribution results, a deep learning algorithm is used to optimize the slope inclination characteristics to obtain enhanced slope inclination characteristic data; Based on the enhanced inclination feature data, the correlation between vegetation coverage and inclination is recalculated. If the correlation is lower than the preset threshold, micro-topography features are introduced for correction to obtain the corrected correlation distribution. Through the corrected correlation distribution, automated calculation is used to generate a micro-topography guidance map for slope vegetation restoration, determine the optimal vegetation layout in each area, and obtain the final monitoring and analysis results.

2. The method according to claim 1, characterized in that The method of obtaining a slope terrain dataset based on UAV remote sensing technology and performing smoothing processing includes: The original data of the slope is obtained based on UAV remote sensing technology, and the features of the original data are extracted using preset sampling rules to obtain the initial data set; Processing the initial data set through a digital elevation model to generate a terrain data set for the slope area; The terrain dataset is smoothed using a Gaussian filtering algorithm to obtain a smoothed terrain dataset.

3. The method according to claim 2, characterized in that The method of extracting slope inclination features based on the smoothed terrain data set to obtain an inclination distribution map includes: For the smoothed terrain dataset, the grid cells are divided, the slope value of each grid cell is calculated, and the slope distribution information is obtained; The inclination angle feature is extracted from the slope distribution information. If the inclination angle feature exceeds the preset threshold, it is marked as a high-relief area, and then a inclination angle distribution map is generated.

4. The method according to claim 1, wherein The extraction of vegetation coverage status based on the inclination distribution map and combined with orthophoto analysis to obtain vegetation coverage distribution data includes: According to the inclination distribution map, terrain feature data is obtained, and orthophoto images are fused to generate comprehensive terrain image data; By integrating terrain image data and using spectral reflectance characteristics, we can distinguish between vegetation areas and non-vegetation areas and obtain preliminary regional division results. Based on the preliminary regional division results, reflectance characteristics are extracted to generate vegetation cover distribution data.

5. The method according to claim 1, wherein The method of analyzing the correlation between vegetation coverage and inclination based on vegetation coverage distribution data and inclination distribution map, obtaining vegetation coverage ratios within different inclination ranges, and calculating vegetation growth differences using statistical methods to obtain difference distribution results includes: The vegetation coverage distribution data and the inclination distribution map are spatially superimposed so that each pixel contains both vegetation coverage and inclination information; Classify the data according to the preset inclination range and count the vegetation coverage ratio within each inclination range; By comparing the vegetation coverage ratio within different inclination ranges, the vegetation growth differences are calculated and a difference distribution map is generated.

6. The method according to claim 3, characterized in that Based on the difference distribution results, the deep learning algorithm is used to optimize the slope inclination characteristics to obtain enhanced slope inclination characteristic data, including: Convolutional neural networks are used to optimize the resolution of digital elevation models to obtain enhanced high-resolution elevation data. Obtain the inclination distribution from the enhanced high-resolution elevation data, divide the high-relief areas into different inclination intervals, and determine the boundaries of each interval; Extract characteristic parameters based on the inclination distribution. If the characteristic parameters exceed the preset threshold, adjust the data smoothness through interpolation to obtain smoothed inclination features. According to the smoothed inclination characteristics, combined with the vegetation cover data, the coverage ratio of each inclination interval is calculated to determine the trend of the ratio change; In view of the trend of proportion change, statistical test methods were used to analyze the difference distribution and obtain the results of significant difference; The convolutional neural network parameters are adjusted based on the significant difference results, the digital elevation model resolution enhancement process is optimized, and the enhanced slope inclination characteristic data is obtained.

7. The method according to claim 1, characterized in that The corrected correlation distribution is used to generate a micro-topography guidance map for slope vegetation restoration using automated calculations to determine the optimal vegetation layout for each area and obtain the final monitoring and analysis results, including: Based on the corrected correlation distribution data, a clustering algorithm is used to classify the slope areas to obtain preliminary zoning results; through a terrain feature extraction algorithm, micro-topography units are identified from the preliminary zoning results to obtain a micro-topography distribution map; for each micro-topography unit, the vegetation type suitable for planting is calculated based on the vegetation growth condition model to determine the vegetation configuration plan; if the vegetation configuration plan differs from the surrounding ecosystem, the vegetation type combination is adjusted to obtain an optimized vegetation layout; a geographic information system is used to draw a micro-topography guidance map for slope vegetation restoration, showing the optimized vegetation layout of each area; based on the micro-topography guidance map for slope vegetation restoration, the slope vegetation coverage and species diversity are calculated to form the final monitoring and analysis results.

8. A computer device comprising a memory, a processor, and a computer program stored in the memory, wherein: The processor executes the computer program to implement the steps of the method according to any one of claims 1 to 7.

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

10. A computer program product comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images