Forest Area Monitoring Methods and Systems Based on Unmanned Aerial Vehicle (UAV) Inspection

Abstract

This invention provides a method and system for forest area monitoring based on unmanned aerial vehicle (UAV) inspection. First, a historical inspection data set containing geographic location identifiers and terrain feature parameters of the target forest area is acquired. Based on this, an initial inspection route is generated, indicating the flight path and image acquisition nodes. Then, a UAV equipped with a multispectral sensor is invoked to dynamically inspect the forest area along the route, acquiring a real-time monitoring image set composed of vegetation cover images of multiple monitoring areas at different timestamps. Next, feature extraction and anomaly detection are performed on this real-time monitoring image set to determine an image anomaly feature set containing anomaly indicators of vegetation and surface structure. Finally, based on this image anomaly feature set, a forest area monitoring optimization strategy is generated to adjust the UAV inspection frequency and the spatial distribution of image acquisition nodes, achieving more efficient and accurate monitoring of the forest area.

Description

Technical Field

[0001] This invention relates to the field of unmanned aerial vehicle (UAV) technology, and more specifically, to a method and system for monitoring forest areas based on UAV inspection. Background Technology

[0002] In today's society, forest area monitoring is of paramount importance for forest resource protection, ecological environment maintenance, and forest disaster prevention. Traditional forest area monitoring methods mainly rely on manual patrols, which have many significant drawbacks. Manual patrols have a limited scope, making it difficult to cover large areas of forest, especially in areas with complex terrain and inconvenient transportation, resulting in numerous blind spots. Furthermore, manual patrols are inefficient, requiring substantial manpower, resources, and time, and cannot provide real-time monitoring, making it difficult to promptly detect anomalies within the forest area.

[0003] With the development of technology, some regions have begun to use fixed monitoring equipment for forest area monitoring. While these fixed monitoring devices can compensate for the limitations of manual inspections in terms of range and real-time performance to some extent, their fixed locations and limited monitoring range prevent flexible adjustments based on the actual conditions of the forest area. Furthermore, fixed monitoring equipment is susceptible to natural environmental factors such as severe weather and tree obstruction, which can reduce monitoring effectiveness. Moreover, fixed monitoring equipment can only acquire images and video information from a fixed perspective, making it difficult to comprehensively and accurately reflect the growth status of forest vegetation and changes in surface structure.

[0004] Furthermore, existing drone inspection methods also have shortcomings in forest area monitoring applications. Most only perform simple aerial photography, lacking effective utilization of historical forest data. Inspection route planning lacks specificity and scientific rigor, failing to fully consider the terrain features and actual monitoring needs of different forest areas. In terms of data processing, they can only perform basic image acquisition, without in-depth feature extraction and anomaly detection, making it impossible to promptly and accurately identify problems such as abnormal vegetation conditions and surface structure anomalies within the forest area. Simultaneously, existing drone inspection methods cannot dynamically adjust inspection strategies according to the actual conditions of the forest area, making it difficult to achieve efficient and precise monitoring of the forest. Summary of the Invention

[0005] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide a forest area monitoring method based on unmanned aerial vehicle (UAV) patrol, the method comprising:

[0006] Acquire a set of historical inspection data for the target forest area, which includes the geographic location identifiers and corresponding terrain feature parameters of multiple monitoring areas;

[0007] Initial inspection routes for multiple monitoring areas are generated based on the historical inspection data set. These initial inspection routes are used to indicate the flight path and image acquisition nodes of the UAV in the target forest area.

[0008] The drone equipped with a multispectral sensor is invoked to perform dynamic inspection operations according to the initial inspection route to obtain a set of real-time monitoring images of the target forest area. The set of real-time monitoring images includes vegetation cover images of multiple monitoring areas at different timestamps.

[0009] Feature extraction and anomaly detection processing are performed on the real-time monitoring image set to determine the image anomaly feature set of the target forest area. The image anomaly feature set includes vegetation state anomaly indicators and surface structure anomaly indicators.

[0010] Based on the set of abnormal image features, an optimized strategy for forest area monitoring is generated.

[0011] In another aspect, embodiments of the present invention also provide a forest area monitoring system based on drone inspection, including a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions, or code. The processor is used to execute the programs, instructions, or code in the machine-readable storage medium to implement the above-described method.

[0012] Based on the above, this embodiment of the invention generates an initial inspection route by acquiring historical inspection data sets of the target forest area, calls upon a drone equipped with a multispectral sensor to perform dynamic inspection operations to acquire a set of real-time monitoring images, performs feature extraction and anomaly detection processing to determine the set of image anomaly features, and then generates a forest area monitoring optimization strategy, thereby improving the accuracy and efficiency of forest area monitoring. Using historical inspection data to generate the initial inspection route allows the drone's flight path and image acquisition nodes to better match the actual terrain of the forest area. The drone equipped with a multispectral sensor performs dynamic inspections, acquiring vegetation cover images of multiple monitoring areas at different time stamps, enriching the dimensions of the monitoring data. Processing the real-time monitoring image set can accurately determine the set of image anomaly features, providing crucial evidence for discovering potential problems in the forest area. Furthermore, the forest area monitoring optimization strategy generated based on the set of image anomaly features can dynamically adjust the drone's inspection frequency and the spatial distribution of image acquisition nodes, achieving a reasonable allocation of forest area monitoring resources and effectively improving the timeliness and targeting of forest area monitoring. Compared with traditional forest area monitoring methods, it can better meet the needs of refined forest area management and real-time monitoring. Attached Figure Description

[0013] Figure 1 This is a schematic diagram of the execution flow of the forest area monitoring method based on drone inspection provided in an embodiment of the present invention.

[0014] Figure 2This is a schematic diagram of exemplary hardware and software components of a forest area monitoring system based on drone inspection provided in an embodiment of the present invention. Detailed Implementation

[0015] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 This is a flowchart illustrating a forest area monitoring method based on drone inspection, as provided in one embodiment of the present invention. The following is a detailed description of this forest area monitoring method based on drone inspection.

[0016] Step S110: Obtain the historical inspection data set of the target forest area, which includes the geographical location identifiers of multiple monitoring areas and their corresponding terrain feature parameters.

[0017] In this embodiment, the historical inspection data set comes from a wide range of sources, including past forest area inspection records, Geographic Information Systems (GIS), and related topographic mapping data. The geographical locations of multiple monitoring areas are represented using a common geographic coordinate system, such as latitude and longitude coordinates (x, y) to accurately determine the location of each monitoring area on Earth. The corresponding topographic feature parameters encompass various terrain-related information, such as elevation data, slope, and aspect. Elevation data reflects the altitude of the monitoring area and is represented by h; slope represents the degree of ground inclination and is represented by θ; aspect indicates the orientation of the slope and is represented by α.

[0018] Taking a large target forest area as an example, the forest area is divided into n monitoring areas, each with its unique geographical location identifier and topographic feature parameters. For the i-th monitoring area (i=1,2,…,n), its geographical location identifier is (xi,yi), and its topographic feature parameters include elevation hi, slope θi, and aspect αi, which constitute the historical inspection data set.

[0019] Step S120: Generate initial inspection routes for multiple monitoring areas based on the historical inspection data set. The initial inspection routes are used to indicate the flight path and image acquisition nodes of the UAV in the target forest area.

[0020] In this embodiment, after acquiring the historical inspection data set of the target forest area, it is necessary to generate initial inspection routes for multiple monitoring areas based on this historical inspection data. The generation of initial inspection routes requires comprehensive consideration of multiple factors to ensure that the UAV can complete the inspection task efficiently and safely.

[0021] Step S121: Extract elevation gradient data and obstacle distribution data from the terrain feature parameters, and determine the vertical flight altitude adjustment threshold of the UAV based on the elevation gradient data.

[0022] In this embodiment, to generate a reasonable initial inspection route, elevation gradient data and obstacle distribution data must first be extracted from terrain feature parameters. Elevation gradient data reflects the vertical changes in terrain and can be obtained by calculating the ratio of the elevation difference between adjacent monitoring areas to the horizontal distance. Let the elevations of two adjacent monitoring areas i and i+1 be hi and hi+1, respectively, and the horizontal distance be d. Then, the elevation gradient g can be expressed as g = (hi+1 - hi) / d.

[0023] Obstacle distribution data describes the location and extent of obstacles in the target forest area that may affect the drone's flight, such as trees and buildings. The distribution of obstacles can be determined by analyzing image data from historical inspection data and combining it with terrain data from a geographic information system.

[0024] Based on the extracted elevation gradient data, the next step is to determine the vertical flight altitude adjustment threshold for the UAV. This process includes the following steps:

[0025] Step S1211: Perform piecewise linear fitting on the elevation gradient data to generate a topographic relief trend curve of the target forest area. The topographic relief trend curve contains the location coordinates of multiple elevation change points.

[0026] In this embodiment, to better analyze the terrain undulations, piecewise linear fitting is required for the elevation gradient data. Piecewise linear fitting divides the entire terrain area into several small segments, and uses a linear function to approximate the terrain changes within each segment. This method generates a terrain undulation trend curve for the target forest area. During the fitting process, points with significant elevation changes will be identified; these points are called elevation abrupt change points. Let the coordinates of the elevation abrupt change points be (xj, hj), where xj represents the horizontal position and hj represents the corresponding elevation.

[0027] Step S1212: Calculate the vertical height difference and horizontal distance ratio between adjacent elevation change points, and determine the maximum allowable climb angle of the UAV in the corresponding interval by combining the maximum thrust parameters and power efficiency model of the UAV.

[0028] In this embodiment, after obtaining the terrain undulation trend curve and the position coordinates of the elevation change points, the ratio of the vertical height difference to the horizontal distance between adjacent elevation change points is calculated. Let the elevations of two adjacent elevation change points j and j+1 be hj and hj+1, respectively, and the horizontal distance be dj. Then the ratio of the vertical height difference to the horizontal distance is rj = (hj+1 - hj) / dj, which is the elevation gradient. To convert the elevation gradient into an angle, a trigonometric function relationship is used, and the unit is converted using the arctangent function to obtain the corresponding angle θj = arctan(rj).

[0029] Simultaneously, by combining the UAV's maximum thrust parameter Tmax and the power efficiency model, the maximum permissible climb angle of the UAV in the corresponding range is determined. The power efficiency model describes the relationship between the UAV's energy consumption and thrust output under different flight conditions. By analyzing this model and the maximum thrust parameter, the maximum permissible climb angle θmax that the UAV can safely fly under different slopes can be obtained. For example, within a certain range, based on the power efficiency model and the maximum thrust parameter, combined with the transformed angle θj, the maximum permissible climb angle is calculated to be θmax1.

[0030] Step S1213: Dynamically adjust the flight altitude adjustment threshold according to the maximum allowable climb angle, so that the minimum safe altitude between the UAV and the ground surface during the climb is always greater than the preset terrain margin.

[0031] In this embodiment, after determining the maximum permissible climb angle of the UAV in different intervals, the flight altitude adjustment threshold is dynamically adjusted based on these angles. A preset terrain margin is used to ensure that the UAV maintains a safe distance from the ground during flight, avoiding collisions with obstacles. Let the preset terrain margin be δ, and the flight altitude adjustment threshold be Δh. Within each interval, the flight altitude adjustment threshold is adjusted according to the maximum permissible climb angle and terrain conditions, ensuring that the minimum safe altitude hmin between the UAV and the ground always satisfies hmin > δ during the climb. For example, in a certain interval, the flight altitude adjustment threshold is calculated as Δh1 based on the maximum permissible climb angle and terrain. When the UAV flies in this interval, its flight altitude should be adjusted according to this threshold to ensure that the minimum safe altitude between the UAV and the ground is greater than the preset terrain margin.

[0032] Step S1214: If the deviation between the current flight altitude and the flight altitude adjustment threshold is detected to exceed the second preset threshold, the hovering operation of the UAV is triggered and the obstacle avoidance path is replanned.

[0033] In this embodiment, the deviation between the current flight altitude and the flight altitude adjustment threshold is monitored in real time during the drone's flight. Let the current flight altitude be hcurrent, the flight altitude adjustment threshold be Δh, and the second preset threshold be ε. When |hcurrent-Δh|>ε, it indicates that the deviation between the current flight altitude and the adjustment threshold is too large, which may affect the drone's flight safety. At this time, the drone's hovering operation is triggered, keeping the drone stationary in the air. Then, based on the current position and obstacle distribution, the obstacle avoidance path is replanned to ensure the drone can continue flying safely.

[0034] Step S122: Construct a three-dimensional spatial obstacle avoidance model for multiple monitoring areas based on the obstacle distribution data. The three-dimensional spatial obstacle avoidance model is used to mark the dynamic obstacle areas in the UAV's flight path.

[0035] In this embodiment, after acquiring obstacle distribution data, a three-dimensional spatial obstacle avoidance model for multiple monitoring areas is constructed using this data. This model combines the three-dimensional spatial information of the target forest area with the location and extent of obstacles to visually display the dynamic obstacle areas in the UAV's flight path.

[0036] First, the three-dimensional space of the target forest area is meshed, dividing it into several smaller spatial units. Each spatial unit is represented by three-dimensional coordinates (x, y, z), where x and y represent the horizontal position and z represents the vertical height. Then, based on obstacle distribution data, the spatial units where obstacles are located are marked in the three-dimensional spatial grid. These marked spatial units are the dynamic obstacle areas.

[0037] For example, within a monitoring area, the locations and heights of trees and buildings are determined using obstacle distribution data. In a three-dimensional obstacle avoidance model, the spatial units containing these trees and buildings are marked as dynamic obstacle areas. When planning its flight path, the drone can refer to this three-dimensional obstacle avoidance model to avoid these dynamic obstacle areas, thereby ensuring flight safety.

[0038] Step S123: Generate path constraints for the initial inspection route based on the flight altitude adjustment threshold and the three-dimensional obstacle avoidance model. The path constraints include a maximum climb angle limit and a minimum safe distance threshold.

[0039] In this embodiment, after determining the flight altitude adjustment threshold and constructing a three-dimensional obstacle avoidance model, path constraints for the initial inspection route are generated based on this information. Path constraints are rules that restrict the flight path of the UAV to ensure its safety and effectiveness during flight.

[0040] The maximum climb angle limit is set based on the previously determined maximum permissible climb angle of the UAV in different intervals. When planning the flight path, the UAV's climb angle must not exceed the maximum permissible climb angle to ensure that the UAV has sufficient power and stability. Let the maximum permissible climb angle be θmax, then the maximum climb angle limit in the path constraint is θ≤θmax.

[0041] The minimum safe distance threshold is used to ensure a safe distance between the drone and obstacles. A minimum safe distance, dmin, is set based on the dynamically marked obstacle areas in the 3D obstacle avoidance model. When planning the flight path, the distance between the drone and the dynamic obstacle area must always be greater than dmin to avoid collisions.

[0042] For example, within a certain monitoring area, the maximum allowable climb angle is determined as θmax2, and the minimum safe distance threshold is determined as dmin2, based on the flight altitude adjustment threshold and the three-dimensional obstacle avoidance model. When generating the initial inspection route, the path constraints require that the UAV's climb angle cannot exceed θmax2, and the distance to the dynamic obstacle area must be greater than dmin2.

[0043] Step S124: Combine the geographical location identifiers of the multiple monitoring areas to perform waypoint optimization processing on the path constraints, and generate an initial inspection route covering all monitoring areas. The image acquisition node density of the initial inspection route is positively correlated with the vegetation cover density of the monitoring area.

[0044] In this embodiment, after obtaining the path constraints of the initial inspection route, waypoint optimization is performed on these constraints by combining the geographical location identifiers of multiple monitoring areas to generate an initial inspection route covering all monitoring areas.

[0045] Waypoint optimization involves selecting appropriate waypoints for UAVs while satisfying path constraints, ensuring efficient coverage of all monitoring areas. Heuristic algorithms, such as genetic algorithms and ant colony optimization, can be used for waypoint optimization.

[0046] Meanwhile, considering the vegetation cover density of the monitoring area, the density of image acquisition nodes along the initial inspection route is set to be positively correlated with the vegetation cover density of the monitoring area. Vegetation cover density can be obtained through analysis of vegetation images in historical inspection data, denoted by ρ. Let the vegetation cover density of monitoring area i be ρi, and the image acquisition node density be ni, then ni = k * ρi, where k is a proportionality coefficient.

[0047] For example, in a target forest area, there are three monitoring areas A, B, and C, with vegetation cover densities ρA, ρB, and ρC, respectively, where ρA > ρB > ρC. Through waypoint optimization and taking vegetation cover density into account, the generated initial inspection route has the highest image acquisition node density in monitoring area A, followed by monitoring area B, and the lowest in monitoring area C. This ensures that more images are collected in areas with high vegetation cover density, enabling more accurate monitoring of the forest's vegetation condition.

[0048] Step S130: Call the UAV equipped with a multispectral sensor to perform dynamic inspection operation according to the initial inspection route, and obtain a set of real-time monitoring images of the target forest area. The set of real-time monitoring images includes vegetation cover images of multiple monitoring areas at different timestamps.

[0049] In this embodiment, after generating an initial inspection route, a drone equipped with a multispectral sensor is invoked to perform dynamic inspection operations along the route to obtain a set of real-time monitoring images of the target forest area. The multispectral sensor can simultaneously acquire images in multiple bands, including visible and near-infrared bands, providing richer vegetation information.

[0050] Step S131: Acquire current wind speed data and light intensity data in real time during the flight of the UAV, and adjust the flight speed and attitude stability parameters of the UAV based on the current wind speed data.

[0051] In this embodiment, real-time wind speed and light intensity data are acquired during the drone's flight. Wind speed data, denoted by v, can be obtained using a wind speed sensor mounted on the drone; light intensity data, denoted by I, can be obtained using a light sensor.

[0052] The specific steps for adjusting the drone's flight speed and attitude stability parameters based on the current wind speed data are as follows:

[0053] Step S1311: Establish a correlation model between wind speed data and UAV aerodynamic drag in advance. The correlation model is used to predict the energy consumption rate of the UAV under different wind speeds.

[0054] In this embodiment, to accurately adjust the UAV's flight parameters based on wind speed, a correlation model between wind speed data and UAV aerodynamic drag is pre-established. This correlation model describes the relationship between wind speed and UAV aerodynamic drag and can be established through experimental data and theoretical analysis. Let the wind speed be v and the UAV's aerodynamic drag be Fd, then the correlation model can be expressed as Fd = f(v), where f(v) is a function of wind speed.

[0055] This correlation model can be used to predict the energy consumption rate of a drone under different wind speeds. The energy consumption rate is related to aerodynamic drag and flight speed. Let the energy consumption rate be P and the flight speed be v'. Considering that the formula for calculating power is that power equals the product of force and velocity, and the unit of force is Newton (N), the unit of velocity is meters per second (m / s), and the unit of power is watt (W), then the energy consumption rate P = Fd * v'.

[0056] Step S1312: Dynamically adjust the ground speed control parameters of the UAV according to the energy consumption rate, so that the UAV maintains the preset airspeed and increases the propulsion power to offset the wind speed when flying against the wind, and reduces the propulsion power and maintains the consistency between the ground speed and the flight path when flying with the wind.

[0057] In this embodiment, after obtaining the energy consumption rate at different wind speeds, the ground speed control parameters of the UAV are dynamically adjusted based on these rates. The preset airspeed is the desired flight speed of the UAV in a windless state, denoted by v0.

[0058] When a drone flies against the wind, the wind speed *v* is opposite to the flight direction. To maintain the preset airspeed *v0*, the propulsion power needs to be increased to counteract the wind speed's effect. Let the ground speed control parameter for headwind be *vheadwind*, then *vheadwind* = *v0* + *v*. Simultaneously, the energy consumption rate is calculated using a correlation model, and the propulsion power is adjusted to meet the energy demand.

[0059] When a drone flies with the wind, the wind speed *v* is in the same direction as the flight. In this case, propulsion power can be reduced while maintaining consistency between ground speed and flight path planning. Let the ground speed control parameter for a tailwind be *vtailwind*, then *vtailwind* = *v0* - *v*. Similarly, propulsion power is adjusted based on the correlation model.

[0060] Step S1313: Based on the inertial measurement unit, monitor the pitch angle and roll angle of the UAV in real time. When the pitch angle or roll angle exceeds the third preset threshold, activate the attitude stabilization algorithm to compensate for the attitude deviation caused by wind speed.

[0061] In this embodiment, during the flight of the UAV, the pitch and roll angles of the UAV are monitored in real time using an inertial measurement unit (IMU). The pitch angle represents the rotation angle of the UAV about the horizontal axis, denoted by φ; the roll angle represents the rotation angle of the UAV about the vertical axis, denoted by ψ.

[0062] When the pitch or roll angle exceeds a third preset threshold, it indicates that wind speed has caused an attitude deviation in the UAV, requiring the attitude stabilization algorithm to be activated to compensate for this deviation. Let the third preset thresholds be φ0 and ψ0. The attitude stabilization algorithm is activated when |φ|>φ0 or |ψ|>ψ0. The attitude stabilization algorithm can adjust the UAV's propeller speed to change its attitude and restore it to a stable state.

[0063] Step S1314: Feed back the adjusted flight speed and attitude stability parameters to the flight control system to control the image acquisition stability of the multispectral sensor in dynamic environments.

[0064] In this embodiment, after adjusting the UAV's flight speed and attitude stability parameters, these adjusted parameters are fed back to the flight control system. The flight control system controls the UAV's flight based on these parameters to ensure the stability of image acquisition by the multispectral sensor in dynamic environments.

[0065] For example, in a windy environment, the drone acquires real-time wind speed and light intensity data. Based on the wind speed data, it adjusts its flight speed and attitude stabilization parameters. When the pitch or roll angle exceeds a third preset threshold, the attitude stabilization algorithm is activated. Finally, the adjusted parameters are fed back to the flight control system, enabling the multispectral sensor to stably acquire images.

[0066] Step S132: Dynamically adjust the exposure time and sensitivity of the multispectral sensor according to the light intensity data so that the brightness uniformity of the acquired vegetation cover image meets the preset threshold.

[0067] In this embodiment, after acquiring the light intensity data, the exposure time and sensitivity of the multispectral sensor are dynamically adjusted based on this data. Exposure time refers to the duration for which the sensor receives light, denoted by t; sensitivity refers to the sensor's sensitivity to light, denoted by ISO.

[0068] The preset threshold is used to ensure that the brightness uniformity of the acquired vegetation cover image meets the set standard. Let the preset threshold be L0. By adjusting the exposure time and ISO, the brightness uniformity L of the acquired image is made to satisfy L≥L0.

[0069] To improve the robustness of subsequent image segmentation, in addition to adjusting exposure time and ISO, illumination-invariant processing, such as color correction, is added. Color correction eliminates the influence of illumination on image color, ensuring that images acquired under different lighting conditions have similar color characteristics. Color correction can be performed using methods based on color space conversion and histogram matching. For example, the image can be converted from the RGB color space to another color space (such as HSV), the luminance component adjusted, and then converted back to the RGB color space. Through these processes, the acquired vegetation cover images maintain good quality under different lighting conditions.

[0070] Step S133: When the UAV arrives at the image acquisition node, it triggers the multispectral sensor to acquire multi-angle images of the target monitoring area and obtain a composite spectral image containing visible light and near-infrared bands.

[0071] In this embodiment, when the UAV flies along the initial inspection route and arrives at the image acquisition node, it triggers the multispectral sensor to acquire multi-angle images of the target monitoring area. Multi-angle image acquisition can obtain information about the target monitoring area from different angles, improving the amount of information and accuracy of the images.

[0072] Images acquired by multispectral sensors contain information from both the visible and near-infrared bands, forming a composite spectral image. The visible band can provide color and texture information about the target monitoring area, while the near-infrared band can reflect information such as the health status and moisture content of vegetation.

[0073] For example, at an image acquisition node in a monitoring area, the UAV triggers the multispectral sensor to acquire images of the monitoring area from different angles (such as directly above, obliquely above, etc.), and obtains a composite spectral image containing the visible light band and the near-infrared band.

[0074] Step S134: Associatively store the composite spectral image with the corresponding geographical location identifier and acquisition timestamp to form a real-time monitoring image set.

[0075] In this embodiment, after obtaining the composite spectral images, these composite spectral images are associatively stored with the corresponding geographical location identifiers and acquisition timestamps. The geographical location identifier can accurately locate the image acquisition position, and the acquisition timestamp can record the image acquisition time.

[0076] These information are stored in the database to form a real-time monitoring image set. The real-time monitoring image set includes vegetation coverage images of multiple monitoring areas at different timestamps.

[0077] For example, for the composite spectral image of a certain monitoring area, it is associatively stored in the database with the geographical location identifier (x, y) of the monitoring area and the acquisition timestamp t. In this way, a complete real-time monitoring image set is formed.

[0078] Step S135: If it is detected that the remaining power of the UAV is lower than the first preset threshold, estimate the power required for returning based on the distance between the current flight position and the return path. If the remaining power is sufficient for returning, re-plan the shortest return path and interrupt the uncompleted image acquisition nodes. If the remaining power is not enough for returning, preferentially execute the image acquisition nodes in the key monitoring areas and then force the UAV to return.

[0079] In this embodiment, during the flight of the UAV, the remaining power of the UAV is monitored in real time. Let the remaining power be E, and the first preset threshold be E0. When E < E0, it means that the remaining power of the UAV is low and corresponding processing is required.

[0080] First, estimate the power required for returning based on the distance between the current flight position and the return path. Considering the real-time wind speed during flight and the adjusted power model, integrate the dynamic ground speed control parameter in step S1312 into the power estimation. According to the real-time wind speed and the power efficiency model of the UAV, calculate the energy consumption per unit distance of the UAV flight at the current wind speed. Let the current flight position be (xcurrent, ycurrent), and the end position of the return path be (xreturn, yreturn), calculate the distance d between the current position and the return point. Combining the energy consumption per unit distance of flight at the current wind speed, estimate the power Ereturn required for returning.

[0081] If E ≥ Ereturn, it means the remaining battery power is sufficient for a return trip. In this case, the shortest return path is replanned, and unfinished image acquisition nodes are interrupted. A path planning algorithm is used, comprehensively considering factors such as terrain, obstacles, and battery consumption, to find the shortest path from the current location to the return point. After planning the shortest path, the drone stops the unfinished image acquisition task and returns along the newly planned path. If E is less than Ereturn, it means the remaining battery power is insufficient to support a direct return trip. In this case, image acquisition nodes for key monitoring areas are prioritized. Key monitoring areas are pre-defined based on factors such as the importance of the forest area and historical anomalies. The drone acquires images of key monitoring areas in order of priority, and is forced to return before the battery is about to run out. This allows for the acquisition of monitoring data for key areas as much as possible with limited battery power.

[0082] Step S140: Perform feature extraction and anomaly detection processing on the real-time monitoring image set to determine the image anomaly feature set of the target forest area. The image anomaly feature set includes vegetation status anomaly indicators and surface structure anomaly indicators.

[0083] In this embodiment, after obtaining the set of real-time monitoring images of the target forest area, it is necessary to perform feature extraction and anomaly detection processing on these images in order to determine the set of image anomaly features of the target forest area.

[0084] Step S141: Perform preprocessing operations on the vegetation cover image, including noise suppression processing and image enhancement processing, to improve the contrast between the vegetation area and the ground surface area in the vegetation cover image.

[0085] In this embodiment, since the vegetation cover images in the real-time monitoring image set may be affected by various factors, such as sensor noise and uneven illumination, preprocessing is required. First, noise suppression is performed using filtering algorithms to remove noise from the image. Common filtering algorithms include mean filtering and median filtering. Mean filtering replaces the gray value of each pixel in the image with the average gray value of its neighboring pixels; median filtering replaces the gray value of a pixel with the median of its neighboring pixels. These filtering algorithms effectively reduce random noise in the image, making the image smoother.

[0086] Next, image enhancement processing is performed to improve the contrast between vegetation areas and ground surface areas in the vegetation cover image. Methods such as histogram equalization can be used to adjust the image's gray-level histogram, making the gray-level distribution more uniform and thus enhancing image contrast. This makes vegetation areas and ground surface areas more clearly displayed in the image, facilitating subsequent feature extraction.

[0087] For example, in a vegetated image, the boundary between vegetation and the ground may be unclear due to noise. After noise suppression processing, the noise in the image is effectively removed, and the image becomes clearer. Further image enhancement processing significantly improves the contrast between the vegetated and ground areas, making the vegetation outlines more defined.

[0088] Step S142: Extract vegetation texture features and surface edge features from the preprocessed vegetation cover image. The vegetation texture features are used to characterize the density change trend of the vegetation canopy, and the surface edge features are used to identify the boundary position between the bare surface area and the vegetation cover area.

[0089] In this embodiment, after preprocessing the vegetation cover image, vegetation texture features and surface edge features are extracted from the processed image. For vegetation texture feature extraction, methods such as the gray-level co-occurrence matrix (GLCM) can be used. The GLCM describes the spatial distribution of gray values ​​in an image. By calculating various statistics of the GLCM, such as contrast, correlation, and energy, feature values ​​of the vegetation texture can be obtained. These feature values ​​can reflect the density variation trend of the vegetation canopy; for example, a texture with high contrast may indicate a high vegetation canopy density.

[0090] For extracting surface edge features, edge detection algorithms, such as the Canny edge detection algorithm, can be used. The Canny edge detection algorithm accurately detects edge information in vegetation cover images by performing Gaussian smoothing, gradient calculation, non-maximum suppression, and double thresholding. In vegetation cover images, this edge information can identify the boundary between bare land areas and vegetation-covered areas.

[0091] For example, in a preprocessed image of vegetation cover, vegetation texture features were extracted using the gray-level co-occurrence matrix, revealing high texture contrast in certain areas, which may indicate a high density of vegetation canopy in those areas. Simultaneously, the Canny edge detection algorithm was used to extract surface edge features, clearly identifying the boundaries between bare and vegetated areas.

[0092] Step S143: Compare the vegetation texture features with the baseline texture features of the corresponding monitoring area in the historical inspection data set to generate vegetation status anomaly indicators. The vegetation status anomaly indicators include the decrease in canopy density and the proportion of leaf discoloration areas.

[0093] In this embodiment, after extracting the vegetation texture features, it is necessary to compare them with the baseline texture features of the corresponding monitoring area in the historical inspection data set to generate vegetation status anomaly indicators.

[0094] Step S1431: Extract the baseline texture features of the same monitoring area in the historical time period from the historical inspection data set. The baseline texture features include the vegetation canopy grayscale distribution histogram and texture direction gradient statistics.

[0095] In this embodiment, baseline texture features of areas identical to the current monitoring area over a historical time period are identified from the historical inspection data set. These baseline texture features include a vegetation canopy grayscale distribution histogram and texture direction gradient statistics. The vegetation canopy grayscale distribution histogram describes the distribution of grayscale values ​​in the vegetation canopy, reflecting the overall brightness characteristics of the vegetation canopy; the texture direction gradient statistics describe the changes in texture in different directions, reflecting the texture structure characteristics of the vegetation canopy.

[0096] For example, for monitoring area A, the histogram of vegetation canopy grayscale distribution and the statistics of texture direction gradient are extracted from the historical inspection data set for a certain period in the past, and used as the baseline texture features of the monitoring area.

[0097] Step S1432: Normalize the gray-level distribution histograms of the current vegetation texture features and the baseline texture features, calculate the Bach distance as the first difference index, and vectorize the texture direction gradient statistics to calculate the cosine similarity as the second difference index.

[0098] In this embodiment, to accurately compare the differences between the current vegetation texture features and the baseline texture features, their gray-level distribution histograms are first normalized. Normalization makes the gray-level distribution histograms of different images comparable, unifying the value range of the gray-level distribution histograms to the same interval. Then, the Bach distance between the normalized gray-level distribution histograms is calculated. The Bach distance measures the similarity between two probability distributions and is used as the first difference index.

[0099] For texture orientation gradient statistics, they are first vectorized, that is, converted into vector form. Then, the cosine similarity between the vectorized texture orientation gradient statistics is calculated. Cosine similarity measures the cosine of the angle between two vectors, reflecting their directional similarity, and is used as the second difference index.

[0100] For example, for the current vegetation texture features and the baseline texture features of monitoring area A, after normalization, the Bartholomew distance of the gray-level distribution histogram is calculated as d1, and the cosine similarity of the vectorized texture direction gradient statistics is s1.

[0101] Step S1433: Normalize the first difference index and the second difference index to the same dimension interval, then sum them by weight to generate the canopy density decrease rate.

[0102] In this embodiment, since the first difference index (Bach's distance, range [0, 1]) and the second difference index (cosine similarity, range [-1, 1]) have different dimensions, the second difference index needs to be processed in order to perform a comprehensive calculation. First, the range of cosine similarity is corrected to [0, 1], which can be done by taking the absolute value, that is, the corrected cosine similarity is |s|, where s is the original cosine similarity.

[0103] Then, the corrected first and second difference indices are normalized using linear normalization, mapping their values ​​to the interval [0, 1]. Next, based on their influence on the canopy density decrease, different weights are assigned to the first and second difference indices. Let the weight of the first difference index be w1, and the weight of the second difference index be w2, with w1 + w2 = 1. The normalized first and second difference indices are then weighted and summed to obtain the canopy density decrease.

[0104] For example, for monitoring area A, after normalizing the Bach distance d1 and the corrected cosine similarity |s1| to the interval [0, 1], we obtain the normalized d1' and |s1'| respectively. If w1=0.6 and w2=0.4, the decrease in canopy density is 0.6*d1'+0.4*|s1'|.

[0105] Step S1434: Identify the leaf area in the current vegetation cover image using an image segmentation algorithm, extract the color space distribution features of the leaf area and compare them with the baseline color features to determine the proportion of discolored areas on the leaf.

[0106] In this embodiment, to determine the proportion of discolored leaf areas, an image segmentation algorithm is first used to identify the leaf areas in the current vegetation cover image. Image segmentation algorithms can separate different regions in an image, isolating the leaf areas from the background. Then, the color space distribution features of the leaf areas are extracted, such as the mean and variance in the RGB color space. These extracted color space distribution features are compared with the baseline color features of the corresponding monitoring areas in the historical inspection data set to determine which areas have changed color. The proportion of discolored leaf areas is determined by statistically analyzing the ratio of the discolored area to the total area of ​​the leaf area.

[0107] For example, in the current vegetation cover image of monitoring area A, an image segmentation algorithm is used to segment the leaf area and extract its color space distribution features. After comparing it with the baseline color features, it is found that the color of some leaf areas has changed. The proportion of these discolored areas to the total leaf area is calculated to obtain the proportion of discolored leaf areas.

[0108] Step S1435: If the decrease in canopy density or the proportion of leaf discoloration area exceeds the fourth preset threshold, then mark the corresponding monitoring area as an area with abnormal vegetation status.

[0109] In this embodiment, a fourth preset threshold is set to determine whether the vegetation status of the monitoring area is abnormal. When the decrease in canopy density or the proportion of leaf discoloration areas exceeds this threshold, it indicates that the vegetation in the monitoring area may be abnormal, and it is marked as an area with abnormal vegetation status.

[0110] For example, for monitoring area A, if the calculated decrease in canopy density or the proportion of leaf discoloration exceeds the fourth preset threshold, then monitoring area A will be marked as an area with abnormal vegetation status.

[0111] Step S144: Detect surface structure anomaly indicators based on the surface edge features, the surface structure anomaly indicators including the estimated value of surface subsidence depth and the number of newly added illegal paths.

[0112] In this embodiment, after extracting surface edge features, surface structure anomaly indicators are detected based on these features. For the detection of the estimated depth of surface subsidence, stereoscopic information or multi-view images can be used for analysis. By comparing the changes in the position and shape of the surface edges in the current image and historical images, combined with topographic data, the depth of surface subsidence is estimated.

[0113] To detect the number of newly added illegal paths, image recognition technology can be used to identify paths in the current image that are not part of the normal planning. The number of these newly added paths is then counted as the number of newly added illegal paths.

[0114] For example, in the images of monitoring area B, by comparing the current image with historical images, significant changes in the surface edges were found, and the depth of surface subsidence was estimated by combining topographic data. Simultaneously, image recognition technology was used to identify newly added illegal paths in the images and their number was counted.

[0115] Step S145: Perform spatiotemporal correlation analysis on the vegetation state anomaly index and the surface structure anomaly index to determine the set of image anomaly features of the target forest area.

[0116] In this embodiment, after obtaining the vegetation state anomaly indicators and surface structure anomaly indicators, it is necessary to perform spatiotemporal correlation analysis on them. Spatiotemporal correlation analysis considers the interrelationships of these anomaly indicators in time and space. The vegetation state anomaly indicators and surface structure anomaly indicators can be matched according to the geographical location of the monitoring area and the collection time. The analysis examines whether vegetation state anomalies and surface structure anomalies occur simultaneously at the same time and location, and the degree of correlation between them.

[0117] For example, in a monitoring area C, at the same time point, anomaly indicators of vegetation status showed a significant decrease in canopy density, while anomaly indicators of surface structure showed surface subsidence. Through spatiotemporal correlation analysis, it can be determined that there are serious anomalies in this monitoring area, and it can be included in the image anomaly feature set of the target forest area.

[0118] Step S150: Generate a forest area monitoring optimization strategy based on the image anomaly feature set. The forest area monitoring optimization strategy is used to adjust the inspection frequency of the UAV and the spatial distribution of image acquisition nodes.

[0119] In this embodiment, after determining the set of image anomaly features of the target forest area, it is necessary to generate an optimized forest area monitoring strategy based on these anomaly features in order to improve the efficiency and accuracy of forest area monitoring.

[0120] Step S151: Determine the fire risk level of the target monitoring area based on the decrease in canopy density in the abnormal vegetation status indicators, and assess the probability of pest and disease spread based on the proportion of leaf discoloration areas.

[0121] In this embodiment, the fire risk level of the target monitoring area is determined based on the decrease in canopy density, an indicator of abnormal vegetation condition. A greater decrease in canopy density indicates a poorer vegetation health condition, a higher likelihood that the vegetation has become dry and flammable, and thus a greater fire risk. The decrease in canopy density can be divided into different intervals, each corresponding to a fire risk level.

[0122] The assessment of the probability of pest and disease spread is based on the proportion of discolored areas on the leaves. A larger proportion of discolored areas indicates a more severe degree of pest and disease damage, and a higher likelihood of pest and disease spread. A mapping relationship can be established to map the proportion of discolored areas on the leaves to a range of pest and disease spread probabilities.

[0123] For example, for monitoring area D, if the decrease in canopy density is within a certain high range, the fire risk level of the monitoring area is determined to be high; if the proportion of leaf discoloration area is large, the probability of pest and disease spread in the monitoring area is assessed according to the mapping relationship.

[0124] Step S152: Generate a soil erosion early warning signal based on the estimated depth of surface subsidence in the surface structure anomaly index, and identify the activity level of illegal logging based on the number of newly added illegal paths.

[0125] In this embodiment, a soil erosion early warning signal is generated based on the estimated surface subsidence depth from the surface structure anomaly index. The greater the surface subsidence depth, the more severe the soil erosion. Different surface subsidence depth thresholds can be set; when the estimated value exceeds a certain threshold, a corresponding level of soil erosion early warning signal is generated.

[0126] The identification of illegal logging activity is based on the number of newly added illegal paths. A higher number of newly added illegal paths indicates potentially more illegal logging activity. The activity level can be categorized into different levels based on the number of newly added illegal paths.

[0127] For example, in monitoring area E, if the estimated depth of surface subsidence exceeds a certain high threshold, a high-level soil erosion early warning signal is generated; if there are a large number of new illegal paths, the illegal logging activity in the monitoring area is identified as high.

[0128] Step S153: The fire risk level, the probability of pest and disease spread, the soil erosion early warning signal and the activity level of illegal logging are normalized into sub-risk scores of a unified scoring interval, and the sub-risk scores are weighted and summed according to the preset weight coefficient to generate a comprehensive risk score for each monitoring area.

[0129] In this embodiment, in order to comprehensively assess the risk situation of each monitoring area, it is necessary to normalize the fire risk level, the probability of pest and disease spread, the soil erosion early warning signal, and the activity level of illegal logging into sub-risk scores within a unified scoring interval. A linear normalization method can be used to map these different types of risk indicators to the same scoring interval, such as [0, 100].

[0130] Then, the sub-risk scores are weighted and summed according to preset weighting coefficients. These preset weighting coefficients are pre-set based on the importance of these risk factors to forest area safety. Let w3 be the weight for fire risk level, w4 for pest and disease spread probability, w5 for soil erosion early warning signal, and w6 for illegal logging activity activity, with w3 + w4 + w5 + w6 = 1. The normalized sub-risk scores are multiplied by their corresponding weighting coefficients and then summed to obtain the comprehensive risk score for each monitoring area.

[0131] For example, for monitoring area F, after normalizing the fire risk level, the probability of pest and disease spread, the soil erosion early warning signal, and the activity level of illegal logging, the sub-risk scores are obtained as s2, s3, s4, and s5, respectively. Let w3=0.3, w4=0.2, w5=0.3, and w6=0.2, then the comprehensive risk score of the monitoring area is 0.3*s2+0.2*s3+0.3*s4+0.2*s5.

[0132] Step S154: Adjust the drone inspection frequency of the corresponding monitoring area based on the comprehensive risk score, so that the inspection frequency of the monitoring area with the larger comprehensive risk score is higher than that of the monitoring area with the smaller comprehensive risk score.

[0133] In this embodiment, to more effectively monitor the forest area, the frequency of drone patrols in the corresponding monitoring area is adjusted based on the comprehensive risk score. The higher the comprehensive risk score, the higher the likelihood of anomalies in the monitoring area, requiring more frequent patrols.

[0134] Step S1541: Assign an initial inspection frequency to each monitoring area, the initial inspection frequency being associated with the geographical area and vegetation type of the monitoring area.

[0135] In this embodiment, an initial inspection frequency is assigned to each monitoring area before adjusting the inspection frequency. The initial inspection frequency is related to the geographical area and vegetation type of the monitoring area. Generally, monitoring areas with larger geographical areas may require more inspections to cover the entire area; different vegetation types have different sensitivities to environmental changes, and monitoring areas with vegetation types that are easily affected by pests, diseases, or fires may require a higher initial inspection frequency.

[0136] For example, monitoring area G, which has a large geographical area and flammable coniferous forest vegetation, is assigned a relatively high initial inspection frequency; while monitoring area H, which has a small geographical area and relatively stable broadleaf forest vegetation, is assigned a relatively low initial inspection frequency.

[0137] Step S1542: Establish a mapping table between the comprehensive risk score and the inspection frequency. The mapping table is used to represent the inspection frequency that increases exponentially with the comprehensive risk score.

[0138] In this embodiment, to accurately adjust the inspection frequency based on the comprehensive risk score, a mapping table between the comprehensive risk score and the inspection frequency is established. This mapping table reflects the exponential growth relationship between the comprehensive risk score and the inspection frequency; that is, the higher the comprehensive risk score, the faster the inspection frequency increases.

[0139] For example, when the overall risk score is in a lower range, the increase in inspection frequency is smaller; when the overall risk score is in a higher range, the increase in inspection frequency is larger. Through this exponential growth mapping relationship, inspection resources can be allocated more rationally.

[0140] Step S1543: When the comprehensive risk score of the monitoring area exceeds the fifth preset threshold for multiple consecutive inspection cycles, the emergency inspection mode is triggered and the inspection frequency of the monitoring area is adjusted according to the mapping relationship table.

[0141] In this embodiment, to promptly detect and handle anomalies in high-risk areas, the fifth preset threshold is set as a dynamic value. By calculating the mean and standard deviation of historical comprehensive risk scores, the fifth preset threshold is set as the sum of the historical score mean and standard deviation. Let the set of historical comprehensive risk scores be {score1, score2, ..., scoreren}. Then, the historical score mean_score is the sum of all scores divided by the number of scores, i.e., mean_score = (score1 + score2 + ... + scoreren) / n. The standard deviation std_score can be obtained by taking the square root of the average of the squares of the differences between each score and the mean. The fifth preset threshold is therefore equal to mean_score + std_score.

[0142] When the comprehensive risk score of a monitored area exceeds the dynamic fifth preset threshold for multiple consecutive inspection cycles, it indicates a serious abnormal risk in the monitored area, triggering the emergency inspection mode. In emergency inspection mode, the inspection frequency of the monitored area is quickly adjusted according to the mapping table, increasing the number of inspections.

[0143] For example, for monitoring area I, if its comprehensive risk score exceeds the dynamic fifth preset threshold for three consecutive inspection cycles, an emergency inspection mode is triggered, and its inspection frequency is significantly increased according to the mapping relationship table.

[0144] Step S1544: Synchronize the adjusted inspection frequency to the UAV's task scheduling system, and prioritize the execution of monitoring area nodes whose comprehensive risk score is ranked above the preset percentage percentile in the task queue.

[0145] In this embodiment, after adjusting the inspection frequency of the monitoring area, the adjusted inspection frequency is synchronized to the UAV's task scheduling system. The task scheduling system arranges the UAV's inspection tasks according to the new inspection frequency. Simultaneously, in the task queue, monitoring area nodes whose comprehensive risk scores rank above a preset percentile are prioritized for execution. The preset percentile is set according to actual needs; for example, monitoring area nodes ranked in the top 20% of comprehensive risk scores are prioritized for execution.

[0146] For example, in the task queue, monitoring area nodes ranked in the top 20% of the comprehensive risk scores are prioritized for drone inspections to ensure that high-risk areas are monitored in a timely manner.

[0147] Step S155: Dynamically adjust the density of image acquisition nodes in the monitoring area based on the spatial distribution data of the image acquisition nodes and the comprehensive risk score.

[0148] In this embodiment, to obtain more accurate information about anomalies in the forest area, the density of image acquisition nodes in the monitoring area needs to be dynamically adjusted based on the spatial distribution data of the image acquisition nodes and the comprehensive risk score. This allows the drone to collect more images in areas with higher overall risk, improving the accuracy of anomaly detection.

[0149] For example, step S1551: Map the comprehensive risk score to a risk level coefficient within a preset range.

[0150] In this embodiment, to facilitate subsequent calculations and adjustments, the comprehensive risk score needs to be mapped to a risk level coefficient within a preset interval. First, the preset interval is determined, for example, set to [0, 1]. Then, the mapping relationship between the comprehensive risk score and the risk level coefficient is established. A linear mapping method can be used. Assuming the comprehensive risk score ranges from [min_score, max_score], and the comprehensive risk score for a certain monitoring area is current_score, then its corresponding risk level coefficient risk_coefficient can be calculated as follows: First, calculate the relative position of the comprehensive risk score within its range, i.e., (current_score - min_score) / (max_score - min_score). The resulting value is the risk level coefficient corresponding to that monitoring area. In this way, monitoring areas with different comprehensive risk scores are mapped to different risk level coefficients within the preset interval. For example, if the comprehensive risk score of monitoring area A is at a high level, the risk level coefficient obtained after mapping will be close to 1; while the comprehensive risk score of monitoring area B is low, its risk level coefficient will be close to 0.

[0151] Step S1552: Extract the spatial distribution data of the current image acquisition nodes and calculate the initial value of the node density for each monitoring area. The initial value of the node density is the number of image acquisition nodes per unit area.

[0152] In this embodiment, to adjust the density of image acquisition nodes, it is first necessary to extract the spatial distribution data of the current image acquisition nodes. This data records the specific location of each image acquisition node within the monitoring area. Then, the initial node density value for each monitoring area is calculated. The initial node density value is measured by the number of image acquisition nodes per unit area. For each monitoring area, its area is first determined, and then the number of image acquisition nodes in that area is counted (node_count). The initial node density value (initial_density) is then equal to node_count divided by area. For example, monitoring area C has a larger area and contains relatively fewer image acquisition nodes, so its initial node density value is lower; while monitoring area D has a smaller area but a larger number of image acquisition nodes, so its initial node density value is higher.

[0153] Step S1553: Normalize the risk level coefficient and the initial value of node density to a preset ratio range, and generate a density adjustment factor based on the ratio of the risk level coefficient to the initial value of node density. The density adjustment factor is positively correlated with the risk level coefficient and negatively correlated with the initial value of node density.

[0154] In this embodiment, to ensure that the risk level coefficient and the initial value of node density can be calculated on the same scale, they need to be normalized to a preset ratio range, such as [0, 1]. For the risk level coefficient, since it is already within the preset range [0, 1], no further normalization is needed. For the initial value of node density, assuming its range is [min_density, max_density], and the initial value of node density for a certain monitoring area is current_initial_density, then the normalized initial value of node density, normalized_initial_density, can be calculated using (current_initial_density - min_density) / (max_density - min_density).

[0155] To avoid the problem of an infinite density adjustment factor caused by the initial normalized node density value being 0, a minimum density threshold ε is set (e.g., ε = 1e-5). When the initial normalized node density value is less than ε, its value is set to ε.

[0156] Next, a density adjustment factor is generated based on the ratio of the risk level coefficient to the processed initial value of the normalized node density. Let the risk level coefficient be `risk_coefficient`, and the processed initial value of the normalized node density be `normalized_initial_density'`. Then, the density adjustment factor `density_adjustment_factor` equals `risk_coefficient` divided by `normalized_initial_density'`. Since a higher risk level coefficient indicates a greater risk in the monitored area, requiring an increase in image acquisition node density, while a higher initial node density indicates that the original node density in the area is already relatively high and may not require a significant increase, the density adjustment factor is positively correlated with the risk level coefficient and negatively correlated with the initial node density. For example, if monitoring area E has a high risk level coefficient but a low initial node density, the calculated density adjustment factor will be large, meaning a significant increase in the image acquisition node density in this area is needed. Conversely, monitoring area F has a low risk level coefficient and a high initial node density, resulting in a smaller density adjustment factor, potentially requiring only a small increase or no increase in node density.

[0157] Step S1554: If the newly added image acquisition node generated by the density adjustment factor overlaps with the obstacle area, then spatial interpolation is used to allocate the newly added image acquisition node in the adjacent area until the conflict-free condition is met.

[0158] In this embodiment, after determining the number of image acquisition nodes to be added based on the density adjustment factor, it is necessary to check whether the positions of these new nodes overlap with obstacle areas. Obstacle areas are marked in the previously constructed 3D spatial obstacle avoidance model and include areas such as trees and buildings that may affect the drone's flight and image acquisition. If it is found that the new image acquisition nodes overlap with obstacle areas, spatial interpolation is used to allocate the new image acquisition nodes to adjacent areas.

[0159] Spatial interpolation is a method for estimating data for unknown points based on data from known points. In this scenario, a new image acquisition node overlapping with an obstacle is selected as the center, and its adjacent regions are chosen. Based on the spatial location of the adjacent regions and the distribution of image acquisition nodes, suitable locations are estimated to assign new nodes. For example, using linear interpolation, a new location in the adjacent region that does not overlap with the obstacle is calculated based on the coordinates and related attributes of adjacent known nodes, and this location is used as the new image acquisition node. This process is repeated until all new image acquisition nodes meet the conflict-free condition, i.e., they do not overlap with obstacle areas.

[0160] Step S1555: Determine the density increment of image acquisition nodes in each monitoring area based on the density adjustment factor, mark the monitoring areas with node density increments greater than 0 as priority adjustment areas, and generate the spatial coordinates of newly added image acquisition nodes.

[0161] In this embodiment, after obtaining the density adjustment factor, the density increment of image acquisition nodes in each monitoring area is determined based on it. The node density increment, *increment*, can be obtained by multiplying the density adjustment factor by a preset coefficient, i.e., *increment* = density_adjustment_factor * coefficient. This preset coefficient is arbitrarily set based on actual conditions and experience, and is used to control the magnitude of the node density increase.

[0162] For monitoring areas where the node density increment is greater than 0, these areas are marked as priority adjustment areas. This is because these areas require increased image acquisition node density to better monitor forest conditions. Then, based on the node density increment and the spatial distribution of the monitoring area, the spatial coordinates of the newly added image acquisition nodes are generated.

[0163] When generating the spatial coordinates of new image acquisition nodes, a path planning algorithm (such as the A* algorithm) is used to ensure flight path feasibility, taking into account the UAV's minimum turning radius and endurance. The A* algorithm is a heuristic search algorithm that finds the optimal path by evaluating the cost of each node (including the actual cost from the starting point to the current node and the estimated cost from the current node to the target node). In this scenario, the location of the new image acquisition node is taken as the target node. Considering constraints such as the UAV's minimum turning radius and endurance, the A* algorithm is used to search for feasible paths and node locations. For example, when generating the coordinates of new nodes, it is ensured that the distance between adjacent nodes meets the UAV's minimum turning radius requirement, while also ensuring that the UAV has sufficient battery power to return after completing the image acquisition tasks for all nodes. This avoids the problem of theoretically generated node coordinates being unattainable in actual flight.

[0164] Step S1556: Correct the initial inspection route according to the spatial coordinates of the newly added image acquisition nodes, so that the adjusted image acquisition node density and the comprehensive risk score of the priority adjustment area are distributed in a non-linear positive correlation.

[0165] In this embodiment, after generating the spatial coordinates of the newly added image acquisition nodes, the initial inspection route needs to be corrected based on these coordinates. The initial inspection route was previously generated based on historical inspection data and terrain factors, but due to the addition of new image acquisition nodes, the initial inspection route needs to be adjusted.

[0166] The goal of the adjustment is to ensure that the adjusted image acquisition node density exhibits a non-linear positive correlation with the overall risk score of the priority adjustment area. This means that areas with higher overall risk scores will see a greater increase in image acquisition node density, but the relationship is not a simple linear one. For example, non-linear functions such as exponential or logarithmic functions can be used to describe this relationship. When revising flight paths, factors such as the location of new image acquisition nodes and the flight performance of the UAV are considered, and the UAV's flight path is replanned so that the UAV can acquire images according to the adjusted image acquisition node density. For example, for priority adjustment areas with high overall risk scores, more stopover points are added to the flight path, allowing the UAV to acquire more images in these areas.

[0167] Step S1557: If the spatial coordinates of the newly added image acquisition node are detected to overlap with the dynamic obstacle area in the obstacle distribution data, the density increment of the image acquisition node in the adjacent monitoring area is reallocated and the density adjustment factor is updated.

[0168] In this embodiment, after correcting the initial inspection route, it is necessary to check again whether the spatial coordinates of the newly added image acquisition nodes overlap with the dynamic obstacle areas in the obstacle distribution data. Dynamic obstacle areas are obstacle areas that may change over time, such as moving vehicles or animals. If overlap is detected, the density increment of image acquisition nodes in adjacent monitoring areas needs to be reallocated.

[0169] The specific approach involves allocating the density increments corresponding to newly added image acquisition nodes that overlap with dynamic obstacle areas to adjacent monitoring areas according to predefined rules. For example, allocation can be based on factors such as the risk level and area of ​​adjacent monitoring areas. After allocation, the density adjustment factors of adjacent monitoring areas are updated. The ratio of the risk level coefficient of these monitoring areas to the normalized initial node density value is recalculated to obtain a new density adjustment factor. Then, based on the new density adjustment factor, the density increments of image acquisition nodes and the spatial coordinates of newly added nodes are determined again until all newly added image acquisition nodes no longer overlap with dynamic obstacle areas, completing the entire dynamic adjustment process of image acquisition node density.

[0170] In summary, this embodiment can effectively monitor forest areas, promptly detect anomalies, and adjust the inspection frequency and image acquisition node density according to the risk situation, thereby improving the efficiency and accuracy of forest area monitoring.

[0171] Figure 2 The illustration shows exemplary hardware and software components of a drone-based forest monitoring system 100 that can implement the ideas of this application, according to some embodiments of this application. For example, a processor 120 can be used in the drone-based forest monitoring system 100 and to perform the functions described in this application.

[0172] The forest area monitoring system 100 based on drone inspection can be a general-purpose server or a special-purpose server; both can be used to implement the drone-based forest area monitoring method of this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the load.

[0173] For example, a forest area monitoring system 100 based on drone inspection may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and various forms of storage media 140, such as a disk, ROM, or RAM, or any combination thereof. Exemplarily, the forest area monitoring system 100 based on drone inspection may also include program instructions stored in ROM, RAM, or other types of non-transitory storage media, or any combination thereof. The methods of this application can be implemented according to these program instructions. The forest area monitoring system 100 based on drone inspection also includes an I / O interface 150 between the computer and other input / output devices.

[0174] For ease of explanation, only one processor is described in the UAV-based forest monitoring system 100. However, it should be noted that the UAV-based forest monitoring system 100 of this application may also include multiple processors. Therefore, the steps performed by one processor as described in this application may also be performed jointly or individually by multiple processors. For example, if the processor of the UAV-based forest monitoring system 100 performs steps A and B, it should be understood that steps A and B may also be performed jointly by two different processors or individually by one processor. For example, the first processor performs step A, the second processor performs step B, or the first processor and the second processor jointly perform steps A and B.

[0175] Furthermore, this embodiment of the invention also provides a readable storage medium, which has computer-executable instructions pre-set in it. When the processor executes the computer-executable instructions, the above-mentioned forest area monitoring method based on UAV inspection is implemented.

[0176] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A forest area monitoring method based on unmanned aerial vehicle (UAV) inspection, characterized in that, The method includes: Acquire a set of historical inspection data for the target forest area, which includes the geographic location identifiers and corresponding terrain feature parameters of multiple monitoring areas; Initial inspection routes for multiple monitoring areas are generated based on the historical inspection data set. These initial inspection routes are used to indicate the flight path and image acquisition nodes of the UAV in the target forest area. The drone equipped with a multispectral sensor is invoked to perform dynamic inspection operations according to the initial inspection route to obtain a set of real-time monitoring images of the target forest area. The set of real-time monitoring images includes vegetation cover images of multiple monitoring areas at different timestamps. Feature extraction and anomaly detection processing are performed on the real-time monitoring image set to determine the image anomaly feature set of the target forest area. The image anomaly feature set includes vegetation state anomaly indicators and surface structure anomaly indicators. A forest area monitoring optimization strategy is generated based on the aforementioned set of image anomaly features; The step of performing feature extraction and anomaly detection processing on the real-time monitoring image set to determine the set of image anomaly features of the target forest area includes: The vegetation cover image is preprocessed, including noise suppression and image enhancement, to improve the contrast between the vegetation area and the ground surface area in the vegetation cover image. Vegetation texture features and surface edge features are extracted from the preprocessed vegetation cover image. The vegetation texture features are used to characterize the density change trend of the vegetation canopy, and the surface edge features are used to identify the boundary position between the bare area and the vegetation cover area. The vegetation texture features are compared with the baseline texture features of the corresponding monitoring area in the historical inspection data set to generate vegetation status anomaly indicators, which include the decrease in canopy density and the proportion of leaf discoloration areas. Surface structure anomaly indicators are detected based on the surface edge features, including the estimated depth of surface subsidence and the number of newly added illegal paths; Spatiotemporal correlation analysis is performed between the vegetation state anomaly index and the surface structure anomaly index to determine the set of image anomaly features of the target forest area. The forest area monitoring optimization strategy generated based on the set of abnormal image features includes: The fire risk level of the target monitoring area is determined based on the decrease in canopy density in the abnormal vegetation status indicators, and the probability of pest and disease spread is assessed based on the proportion of leaf discoloration areas. Soil erosion early warning signals are generated based on the estimated depth of surface subsidence in the surface structure anomaly indicators, and the activity level of illegal logging is identified based on the number of newly added illegal paths. The fire risk level, the probability of pest and disease spread, the soil erosion early warning signal and the activity level of illegal logging are normalized into sub-risk scores within a unified scoring range. The sub-risk scores are then weighted and summed according to preset weight coefficients to generate a comprehensive risk score for each monitoring area. The frequency of drone inspections in the corresponding monitoring areas is adjusted based on the comprehensive risk score, so that the inspection frequency of the monitoring area with the higher comprehensive risk score is higher than that of the monitoring area with the lower comprehensive risk score. The density of image acquisition nodes in the monitoring area is dynamically adjusted based on the spatial distribution data of the image acquisition nodes and the comprehensive risk score.

2. The forest area monitoring method based on UAV inspection according to claim 1, characterized in that, The process of generating initial inspection routes for multiple monitoring areas based on the historical inspection data set includes: Extract elevation gradient data and obstacle distribution data from the terrain feature parameters, and determine the vertical flight altitude adjustment threshold of the UAV based on the elevation gradient data; Based on the obstacle distribution data, a three-dimensional spatial obstacle avoidance model for multiple monitoring areas is constructed. The three-dimensional spatial obstacle avoidance model is used to mark the dynamic obstacle areas in the UAV's flight path. The path constraints for generating the initial inspection route are based on the flight altitude adjustment threshold and the three-dimensional obstacle avoidance model. The path constraints include the maximum climb angle limit and the minimum safe distance threshold. By combining the geographic location identifiers of the multiple monitoring areas, waypoint optimization is performed on the path constraints to generate an initial inspection route covering all monitoring areas. The image acquisition node density of the initial inspection route is positively correlated with the vegetation cover density of the monitoring areas.

3. The forest area monitoring method based on UAV inspection according to claim 2, characterized in that, Determining the vertical flight altitude adjustment threshold of the UAV based on the elevation gradient data includes: The elevation gradient data is subjected to piecewise linear fitting to generate a topographic relief trend curve of the target forest area, which includes the location coordinates of multiple elevation change points. Calculate the vertical height difference and horizontal distance ratio between adjacent elevation change points, and combine the maximum thrust parameters and power efficiency model of the UAV to determine the maximum allowable climb angle of the UAV in the corresponding section. The flight altitude adjustment threshold is dynamically adjusted according to the maximum allowable climb angle, so that the minimum safe altitude between the UAV and the ground surface is always greater than the preset terrain margin during the climb process. If the deviation between the current flight altitude and the flight altitude adjustment threshold is detected to exceed the second preset threshold, the drone will be triggered to hover and replan the obstacle avoidance path.

4. The forest area monitoring method based on UAV inspection according to claim 1, characterized in that, The process of calling upon a drone equipped with a multispectral sensor to perform dynamic inspection operations according to the initial inspection route, and acquiring a set of real-time monitoring images of the target forest area, includes: During the flight of the drone, the current wind speed data and light intensity data are acquired in real time, and the flight speed and attitude stability parameters of the drone are adjusted based on the current wind speed data. The exposure time and sensitivity of the multispectral sensor are dynamically adjusted based on the light intensity data so that the brightness uniformity of the acquired vegetation cover image meets a preset threshold. When the drone arrives at the image acquisition node, it triggers the multispectral sensor to acquire multi-angle images of the target monitoring area, obtaining a composite spectral image containing visible light and near-infrared bands; The composite spectral image is associated with and stored with the corresponding geographic location identifier and acquisition timestamp to form a real-time monitoring image set; If the remaining battery power of the drone is detected to be lower than the first preset threshold, the battery power required for return is estimated based on the distance between the current flight position and the return path. If the remaining battery power is sufficient for return, the shortest return path is replanned and the unfinished image acquisition nodes are interrupted. If the remaining battery power is insufficient for return, the image acquisition nodes of the key monitoring area are executed first, and then the drone is forced to return.

5. The forest area monitoring method based on UAV inspection according to claim 4, characterized in that, The adjustment of the drone's flight speed and attitude stability parameters based on the current wind speed data includes: A correlation model between wind speed data and UAV aerodynamic drag is pre-established, and the correlation model is used to predict the energy consumption rate of the UAV under different wind speeds; The ground speed control parameters of the UAV are dynamically adjusted according to the energy consumption rate, so that the UAV maintains a preset airspeed and increases propulsion power to offset the wind speed when flying against the wind, and reduces propulsion power and maintains the consistency between ground speed and flight path planning when flying with the wind. The pitch and roll angles of the UAV are monitored in real time using an inertial measurement unit. When the pitch or roll angle exceeds a third preset threshold, the attitude stabilization algorithm is activated to compensate for the attitude deviation caused by wind speed. The adjusted flight speed and attitude stability parameters are fed back to the flight control system to control the image acquisition stability of the multispectral sensor in dynamic environments.

6. The forest area monitoring method based on UAV inspection according to claim 1, characterized in that, The step of comparing the vegetation texture features with the baseline texture features of the corresponding monitoring area in the historical inspection data set to generate vegetation state anomaly indicators includes: The baseline texture features of the same monitoring area over a historical time period are extracted from the historical inspection data set. The baseline texture features include the vegetation canopy grayscale distribution histogram and texture direction gradient statistics. The gray-level distribution histograms of the current vegetation texture features and the baseline texture features are normalized, and the Bach distance is calculated as the first difference index. The cosine similarity is calculated as the second difference index after the texture direction gradient statistics are vectorized. After normalizing the first and second difference indices to the same dimension intervals, the weighted sum is calculated to generate the canopy density decrease rate. The leaf area in the current vegetation cover image is identified by image segmentation algorithm. The color space distribution features of the leaf area are extracted and compared with the baseline color features to determine the proportion of discolored leaf area. If the decrease in canopy density or the proportion of leaf discoloration areas exceeds the fourth preset threshold, the corresponding monitoring area is marked as an area with abnormal vegetation status.

7. The forest area monitoring method based on UAV inspection according to claim 1, characterized in that, The adjustment of the drone inspection frequency for the corresponding monitoring area based on the comprehensive risk score includes: An initial inspection frequency is assigned to each monitoring area, and the initial inspection frequency is related to the geographical area and vegetation type of the monitoring area. Establish a mapping table between the comprehensive risk score and the inspection frequency, wherein the mapping table is used to represent the inspection frequency that increases exponentially with the comprehensive risk score; When the comprehensive risk score of the monitored area exceeds the fifth preset threshold for multiple consecutive inspection cycles, the emergency inspection mode is triggered and the inspection frequency of the monitored area is adjusted according to the mapping table. The adjusted inspection frequency will be synchronized to the drone's task scheduling system, and monitoring area nodes with comprehensive risk scores ranked above the preset percentile will be prioritized in the task queue.

8. A forest area monitoring system based on unmanned aerial vehicle (UAV) inspection, characterized in that, The system includes a processor and a memory, the memory and the processor being connected. The memory is used to store programs, instructions or code, and the processor is used to execute the programs, instructions or code in the memory to implement the forest area monitoring method based on UAV inspection as described in any one of claims 1-7.

Images