Unmanned aerial vehicle ecological monitoring aerial survey coverage measurement method based on geographic grids

Abstract

The invention discloses an unmanned aerial vehicle ecological monitoring aerial survey coverage measurement method based on geographic grids, and relates to the technical field of aerial survey coverage measurement, and the method comprises the steps: obtaining a boundary of a to-be-monitored region, generating a grid unique number, binding a topographic relief parameter, a canopy height parameter and an observation stability parameter, collecting an incident angle set of each grid, and carrying out the collection of the incident angle set; calculating a view angle change amplitude, calculating a shielding risk parameter and comparing the shielding risk parameter with a shielding risk threshold value when the view angle change amplitude is not met, obtaining coverage judgment, constructing a local three-dimensional voxel model for grids with invalid coverage, reproducing a historical observation attitude to execute sight line penetration analysis, calculating a three-dimensional visibility ratio and comparing the three-dimensional visibility ratio with a three-dimensional visibility threshold value to correct a coverage state; and calculating the coverage improvement amount of the grids which are still invalid after correction according to the change of the visibility ratio, obtaining a revisit income ratio in combination with the number of revisit times, sorting and grading according to the income, clustering according to the spatial proximity relation to generate a revisit route, re-calculating the key criterion after revisit, and outputting a grid coverage result set according to the convergence criterion.

Description

Technical Field

[0001] This invention relates to the field of aerial survey coverage measurement technology, and more specifically, to a method for aerial survey coverage measurement of UAV ecological monitoring based on geographic grids. Background Technology

[0002] UAV aerial surveys are commonly used for multi-phase image acquisition and coverage verification in ecological monitoring scenarios. Coverage measurement results typically serve as the basis for subsequent sampling deployment, key area verification, and comparison of coverage consistency across multiple phases. Existing coverage assessments mostly focus on two-dimensional planar units, emphasizing statistical methods such as flight path overlap, image quantity, or projected coverage. This makes it difficult to uniformly incorporate factors such as terrain undulations within the grid, canopy height differences, and historical imaging stability into a single criterion. When the differences in the set of observation incident angles are small or when affected by terrain or vegetation obstruction, two-dimensional coverage determination is prone to blind spot omissions or overly conservative approaches.

[0003] To mitigate the impact of occlusion, some solutions employ 3D modeling or field-of-view analysis. However, directly constructing a global 3D model within a large monitoring area and performing ray penetration calculations on multiple sorties and attitude data results in a large data volume and computational overhead. Furthermore, if 3D analysis does not distinguish between problematic and non-problematic grids, it is difficult to balance processing efficiency and verification accuracy.

[0004] Furthermore, revisits to areas with insufficient coverage often rely on manual experience or are performed a fixed number of times. There is a lack of quantitative indicators for coverage improvement brought about by revisits, priority allocation rules, and convergence control mechanisms, which can easily lead to uneven allocation of revisit resources or duplicate aerial surveys. For the above reasons, there is an urgent need for an aerial survey coverage measurement method that uses geographic grids as a unified spatial unit and links two-dimensional judgment, local three-dimensional correction, and revisit benefit-driven iteration.

[0005] To address the above problems, this invention proposes a solution. Summary of the Invention

[0006] To overcome the aforementioned deficiencies of the prior art, embodiments of the present invention provide a method for measuring the coverage of UAV ecological monitoring aerial surveys based on geographic grids, in order to solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the present invention provides the following technical solution: A method for measuring the coverage of UAV ecological monitoring aerial surveys based on geographic grids, comprising the following steps; Step S1: Obtain the boundary of the area to be monitored and unify the coordinates. Divide the area into two-dimensional grids according to the grid side length and generate grid numbers. Bind the elevation difference parameter, canopy height parameter, and observation stability parameter to each grid. Step S2: Collect the set of incident angles covering each grid, calculate the change in viewing angle as the difference between the maximum and minimum values ​​of the set of incident angles, and compare it with the viewing angle diversity threshold. If the threshold is met, the two-dimensional coverage is deemed effective. If not, the occlusion risk parameter is calculated using the elevation difference parameter, canopy height parameter, and observation stability parameter, and compared with the occlusion risk threshold to determine whether the two-dimensional coverage is effective or invalid. Step S3: Construct a local three-dimensional voxel model for the two-dimensional invalid grid, voxelize the terrain and canopy three-dimensional data of the target grid and its neighborhood and mark the occupancy status; reproduce the historical observation posture to perform line-of-sight penetration analysis, summarize the visible surface voxels and calculate the three-dimensional visibility ratio, and compare the three-dimensional visibility ratio with the three-dimensional visibility threshold to correct the coverage status. Step S4: For grid cells that are still not covered after correction, calculate the coverage improvement based on the change in 3D visibility ratio, and calculate the revisit benefit ratio based on the number of revisits; sort and classify according to the revisit benefit ratio and generate revisit routes by clustering according to spatial proximity; recalculate the change in viewing angle and 3D visibility ratio after revisiting, and output the grid coverage result set according to the convergence criterion.

[0008] In a preferred embodiment, step S1 includes the following: The parameters to be bound to each grid cell include elevation difference, canopy height, and observation stability parameters, including: The elevation difference parameter is obtained by calculating the difference between the elevation of the highest point and the lowest point within the grid. The canopy height parameter is obtained by calculating the difference between the average canopy top elevation and the average ground elevation within the grid area; The observation stability parameter is obtained by calculating the proportion of the number of aerial survey images covering the grid that meet the preset imaging quality standards to the total number of aerial survey images covering the grid. The preset image quality standards include an assessment of image sharpness and appropriateness of exposure.

[0009] In a preferred embodiment, step S2 includes the following: In step S2, the blind zone risk probability under the current two-dimensional observation conditions is evaluated by combining the three binding parameters ER, CH, and OS. The occlusion risk parameter is calculated as follows: Where OS represents the observation stability. The vegetation scale constant is . The topographic scale constant; The specific value of the occlusion risk threshold is set according to the coverage accuracy requirements of the ecological monitoring task.

[0010] In a preferred embodiment, step S3 includes the following: The 3D visibility ratio is the ratio of the sum of the projected areas of all visible surface voxels within a grid on the horizontal plane to the total projected area of ​​the grid on the horizontal plane. The specific value of the three-dimensional visibility threshold is set according to the requirements of the monitoring task for the integrity of regional coverage.

[0011] In a preferred embodiment, step S4 includes the following: The coverage improvement is obtained by calculating the difference between the 3D visibility ratio after the revisit aerial survey and the initial 3D visibility ratio; the revisit benefit ratio is obtained by calculating the quotient obtained by dividing the coverage improvement by the number of revisits. The ranking based on revisit benefit ratio includes: setting benefit thresholds at least two levels, and classifying rasters that need to be revisited into different revisit priority levels by comparing the revisit benefit ratio of each raster with these benefit thresholds; The convergence criteria include at least one of the following: a revisit benefit convergence threshold for judging whether the marginal benefit of a single revisit is too low, an absolute coverage convergence threshold for judging whether the three-dimensional visibility ratio of the raster meets the target requirements, and a global convergence ratio threshold for judging whether the coverage correction process of the entire monitoring area is completed.

[0012] The technical effects and advantages of the UAV ecological monitoring aerial survey coverage measurement method based on geographic grids of this invention are as follows: This invention uses a two-dimensional grid as a unified spatial unit to generate grid numbers, and binds elevation difference parameters, canopy height parameters, and observation stability parameters to each grid, so that coverage measurements are organized and traced back under a unified index, making it easier to incorporate internal grid differences into the same judgment process; for two-dimensional grids with invalid coverage, a local three-dimensional voxel model is further constructed and the historical observation posture is reproduced to perform line-of-sight penetration analysis, calculate the three-dimensional visibility ratio, and correct the coverage status according to the three-dimensional visibility threshold, so that terrain and canopy occlusion factors are explicitly characterized in coverage judgment. During the revisit phase, the coverage improvement is calculated based on the change in the 3D visibility ratio and the revisit benefit ratio is obtained by combining the number of revisits. The revisit routes are generated by ranking and classifying the benefits and clustering them according to spatial proximity. At the same time, the target incident angles with large differences from the historical incident angle set are selected first. The convergence control threshold and convergence criteria are used to terminate revisits with insufficient marginal benefits or where the coverage correction has converged, thereby reducing invalid repeated aerial surveys and forming a stopable iterative process. The final output includes a coverage result set containing the 3D visibility ratio, revisit count, coverage improvement, revisit benefit ratio, and coverage status markers corresponding to the raster number. It also provides a list of invalid rasters as input for subsequent revisit plans and valid rasters as baseline data for coverage measurement results and multi-period coverage consistency comparison. Attached Figure Description

[0013] Figure 1This is a schematic diagram illustrating the steps of a UAV-based aerial survey coverage measurement method for ecological monitoring based on geographic grids, according to the present invention. Detailed Implementation

[0014] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0015] For examples, please refer to Figure 1 As shown, this invention discloses a method for measuring aerial coverage of UAV ecological monitoring based on geographic grids, including the following steps: Step S1: Obtain the boundary of the area to be monitored and unify the coordinates. Divide the area into two-dimensional grids according to the grid side length and generate grid numbers. Bind the elevation difference parameter, canopy height parameter, and observation stability parameter to each grid. Step S2: Collect the set of incident angles covering each grid, calculate the change in viewing angle as the difference between the maximum and minimum values ​​of the set of incident angles, and compare it with the viewing angle diversity threshold. If the threshold is met, the two-dimensional coverage is deemed effective. If not, the occlusion risk parameter is calculated using the elevation difference parameter, canopy height parameter, and observation stability parameter, and compared with the occlusion risk threshold to determine whether the two-dimensional coverage is effective or invalid. Step S3: Construct a local three-dimensional voxel model for the two-dimensional invalid grid, voxelize the terrain and canopy three-dimensional data of the target grid and its neighborhood and mark the occupancy status; reproduce the historical observation posture to perform line-of-sight penetration analysis, summarize the visible surface voxels and calculate the three-dimensional visibility ratio, and compare the three-dimensional visibility ratio with the three-dimensional visibility threshold to correct the coverage status. Step S4: For grid cells that are still not covered after correction, calculate the coverage improvement based on the change in 3D visibility ratio, and calculate the revisit benefit ratio based on the number of revisits; sort and classify according to the revisit benefit ratio and generate revisit routes by clustering according to spatial proximity; recalculate the change in viewing angle and 3D visibility ratio after revisiting, and output the grid coverage result set according to the convergence criterion.

[0016] In step S1, the boundary of the area to be monitored is obtained and the coordinates are unified. A two-dimensional grid is divided according to the grid side length and grid numbers are generated. Elevation difference parameters, canopy height parameters, and observation stability parameters are bound to each grid. Specific details include: The boundary polygon of the area to be monitored is obtained and converted to a unified geographic coordinate system. The area is divided into a series of two-dimensional grid cells with a predetermined grid side length. A unique grid number GI is generated for each grid cell. The GI is converted from its index in the row and column matrix or the coordinates of the center point and is used to uniquely identify the geographic location of the grid cell. Specifically, a combination of row index and column index encoding or latitude and longitude encoding based on the WGS84 coordinate system can be used to ensure the spatial uniqueness of each grid cell. Grid division is not only to simplify data management, but more importantly, to make the subsequent coverage assessment based on a unified spatial cell. For each grid GI, the basic environmental information is synchronously bound to include the elevation difference parameter ER, which describes the degree of terrain undulation within the grid; the canopy height parameter CH, which describes the average height of vegetation canopy or other vertical obstacles within the grid; and the observation stability parameter OS, which describes the stability of the image quality of the grid in historical aerial surveys. The ER value is calculated by subtracting the lowest elevation from the highest elevation within the grid; the CH value is obtained by the difference between the digital surface model and the digital terrain model; and the OS value is calculated by the ratio of the number of historical valid images to the total number of images. This invention adds this information to the grid numbering to allow for differentiated judgments for different grids during subsequent criterion design.

[0017] Elevation difference (ER) reflects the severity of terrain undulation within a grid and is a core parameter for assessing terrain occlusion risk. Its calculation is based on digital elevation model (DEM) data, extracting elevation data from all ground points within the current GI grid area and selecting the highest elevation value. and the lowest point elevation value The difference between the two is ER, and the calculation formula is as follows: Both are based on the same elevation datum. The value of ER ranges from [0, +∞). The larger the value of ER, the more significant the terrain undulation within the grid, and the higher the probability that the line of sight will be blocked by terrain protrusions during UAV aerial surveying. When ER=0, the terrain within the grid is absolutely flat. Canopy height (CH) reflects the average height of vertical obstacles such as vegetation within the grid and is a core parameter for assessing vegetation shading risk. Its calculation is based on the difference between airborne lidar point cloud data or digital surface models and the DEM, extracting the average canopy top elevation within the current GI grid area. and average ground elevation The difference between the two is CH, calculated using the following formula: If there is no tall vegetation and no building covering the grid, then CH≤0.5m, which can be approximated as 0. The value of CH ranges from [0,+∞). The larger the value of CH, the higher the vertical obstacle in the grid, and the higher the probability that the UAV's aerial survey line of sight will be blocked. Observation stability (OS) reflects the stability and reliability of the raster's imaging quality across multiple UAV observations. It is a core parameter for evaluating the validity of image data and is defined as the proportion of images that meet the preset imaging quality requirements out of the total number of images covering the current GI raster. The calculation formula is as follows: , n is the total number of aerial images covering the current grid, and k is the number of images that meet the imaging quality requirements. The value of OS ranges from [0,1]. The closer OS is to 1, the more stable the observation conditions of the grid and the more reliable the image data. The closer OS is to 0, the worse the observation conditions and the lower the credibility of the image data. OS only evaluates the stability of image quality and does not involve the diversity of observation perspectives. It should be noted that images meeting imaging quality requirements can be judged based on image sharpness and appropriate exposure. Image sharpness is the primary indicator, used to eliminate the impact of blurring and defocusing on the accuracy of coverage. For each image to be judged, the response value of its Laplacian operator is calculated, and the variance of its grayscale change is taken as the blur index BI. A preset blur threshold BI = 0.2 is used. When BI ≤ BIT, the image is judged to be sharp; otherwise, it is considered blurry and unacceptable. To avoid information loss due to underexposure or overexposure, the deviation between the average image brightness value and the ideal exposure level is used as a criterion. An exposure deviation parameter is constructed, and the center grayscale value in the image histogram is statistically analyzed. Compare it with the desired gray level Compare and convert to exposure deviation EB; set the exposure deviation threshold EBT=±0.5EV. If |EB|≤EBT is satisfied, the image is considered to have appropriate exposure and can be used for subsequent OS evaluation; otherwise, it is an overexposed or underexposed sample. If both BI≤BIT and |EB|≤EBT are satisfied, the image is judged to meet the imaging quality requirements. Output a raster attribute association dataset containing a unique raster ID (GI), terrain relief parameter (ER), vegetation shading parameter (CH), observation stability parameter (OS), and image quality judgment result. This dataset provides accurate spatial units and attribute data support for the differentiated design of subsequent two-dimensional coverage validity criteria, ensuring that coverage assessment can be tailored to the differences in terrain, vegetation, and observation conditions of each raster, thus laying a data foundation for improving the accuracy of coverage measurement.

[0018] In step S2, the set of incident angles covering each grid is collected, and the change in viewing angle is calculated as the difference between the maximum and minimum values ​​of the incident angle set. This is then compared with a viewing angle diversity threshold. If the threshold is met, the two-dimensional coverage is deemed valid. If not, occlusion risk parameters are calculated using elevation difference parameters, canopy height parameters, and observation stability parameters, and compared with an occlusion risk threshold to determine whether the two-dimensional coverage is valid or invalid. Specific details include: The core of the two-dimensional coverage effectiveness determination is to judge whether the grid has obtained a sufficiently diverse observation perspective based on the imaging line-of-sight incident feature (IA) of the UAV image combined with a preset threshold, thereby initially screening out grids with the risk of coverage blind spots. Here, IA represents the incident angle and direction of the line of sight when the UAV camera images the grid. The angle between the camera's line of sight and the vertical direction in IA is the core evaluation index. The value range of θ is [0°, 90°), where θ=0° indicates vertical top-down shooting, and the larger θ is, the larger the tilt angle is. The angle variation magnitude (IV) reflects the dispersion of the incident angles of all images covering the same grid, and is used to assess angle diversity. It collects the set of incident angles corresponding to all aerial survey images covering the current GI grid. Let n be the total number of images. The difference between the maximum and minimum values ​​in this set is called the IV, and the calculation formula is as follows: Where θ is the incident angle when the current grid is observed for the j-th time. The maximum value of the incident angle. The minimum incident angle is represented by IV, which ranges from [0° to 90°]. The larger the IV, the greater the difference in the angle from which the UAV observes the grid. The greater the diversity of viewpoints, the more effectively the UAV can avoid blind spots under a single viewpoint. The smaller the IV, the more consistent all observation viewpoints tend to be, and the more likely it is to form uncovered areas due to occlusion. A view diversity threshold τ is set to determine whether the raster has sufficient view diversity. The value of τ is determined based on the vegetation type and terrain complexity of the monitoring area, and the specific value range can be set as follows: In simple environments, such as flat terrain or low vegetation, τ can be set to a recommended value of 15°, meaning that when IV ≥ 15°, the visual diversity is sufficient. In moderately complex environments, such as gentle slopes or medium to tall vegetation, τ can be set to a recommended value of 25°, meaning that when IV ≥ 25°, the visual diversity is sufficient. In complex environments, such as steep slopes or tall vegetation, τ can be set to a recommended value of 35°, meaning that when IV ≥ 35°, the visual diversity is sufficient. For the current GI grid, if IV≥τ, it indicates that the grid has obtained a sufficiently diverse range of observation perspectives. The occlusion risk under a single perspective can be avoided through the complementarity of multiple perspectives. It is preliminarily determined that its two-dimensional coverage is effective and there is no need to proceed to the subsequent occlusion risk assessment. If IV<τ, it indicates that the diversity of perspectives is insufficient and there is a potential risk of blind spots due to occlusion. An occlusion risk parameter OR is introduced, and the blind zone risk probability under the current two-dimensional observation conditions is evaluated by combining three bound parameters: ER, CH, and OS. The OR value ranges from [0,1]. The larger the value, the higher the probability of unobserved blind zones in the raster; the smaller the value, the lower the blind zone risk. The formula is expressed as: ; where, OS is the observation stability, representing the reliability of the image data. The lower the OS, the worse the image quality. Even if the environmental occlusion risk is low, blind spot misjudgment may occur due to image problems, thus increasing the OR value; is the vegetation scale constant, used to map CH to the vegetation occlusion risk factor in the range of [0, 1], preset according to the vegetation type. Among them, it can be set as: herbaceous vegetation area = 0.5m, shrub area = 2m, forest area = 5m; is the terrain scale constant, used to map ER to the terrain occlusion risk factor in the range of [0, 1], preset according to the terrain type. Among them, it can be set as: flat area = 1m, gentle slope area = 5m, steep slope area = 10m; is the vegetation occlusion risk factor OR. The larger CH is, the closer OR is to 1, and the higher the vegetation occlusion risk; when CH = 0, OR = 0, there is no vegetation occlusion risk; is the terrain occlusion risk factor OR. The larger ER is, the closer OR is to 1, and the higher the terrain occlusion risk; when ER = 0, OR = 0, there is no terrain occlusion risk; Set the occlusion risk threshold T, used to determine the two-dimensional coverage effectiveness of the grid with insufficient view diversity. T is preset according to the monitoring accuracy requirements. The recommended value range is 0.4 - 0.6. In high-precision monitoring scenarios, such as endangered species habitat monitoring, T can be set to 0.4. In conventional monitoring scenarios, T can be set to 0.5. In low-precision monitoring scenarios, T can be set to 0.6; If OR ≥ T, it indicates that there is a significant blind spot risk in this grid, and the two-dimensional coverage is determined to be invalid; if OR < T, it indicates that although the view is single, due to flat terrain, short vegetation or reliable image quality, the blind spot risk is low, and the two-dimensional coverage is determined to be effective and does not need to enter three-dimensional correction; All GI grids can be divided into two categories through two-level criteria: when IV ≥ τ, or when IV < τ and OR < T, that is, the two-dimensional coverage effective grid; when IV < τ and OR ≥ T, the two-dimensional coverage invalid grid. Only the two-dimensional coverage invalid grid is subjected to subsequent three-dimensional visibility analysis.

[0019] In step S3, a local three-dimensional voxel model is constructed for the two-dimensional coverage invalid grid, and the terrain and canopy three-dimensional data of the target grid and its neighborhood range are voxelized and the occupancy status is marked; the historical observation posture is reproduced to perform line-of-sight penetration analysis, the visible surface voxels are summarized and the three-dimensional visibility ratio is calculated, and the three-dimensional visibility ratio is compared with the three-dimensional visibility threshold to correct the coverage status. The specific content includes: For two-dimensional ineffective rasters, a local three-dimensional voxel model is constructed to calculate the three-dimensional visibility ratio through fine line-of-sight penetration analysis, which accurately corrects the coverage effectiveness and eliminates the conservatism or misjudgment of two-dimensional evaluation. The local three-dimensional voxel model only covers the target raster and its adjacent rasters. Compared with the global three-dimensional model, it has the advantages of small data volume and high computational efficiency, and is suitable for real-time processing in the field. Using 3D data bound to GI raster, a local 3D model is constructed using the voxelization method. The specific steps are as follows: Centered on the target GI grid, expand to 1-2 adjacent grids to determine the modeling range. For example, if the target grid is 5m×5m, the modeling range is 15m×15m to ensure coverage of all potential obstacles. The modeling range is discretized into cubic voxels in the x, y, and z directions with a fixed resolution. The voxel resolution is set according to the monitoring accuracy. It can be 0.5m×0.5m×0.5m for high accuracy or 1m×1m×1m for conventional accuracy. Based on LiDAR point cloud or DSM / DEM data, determine whether each voxel is occupied by physical objects, such as terrain, vegetation, and buildings: if the number of physical object point clouds contained in the voxel is ≥3, or the voxel center elevation is ≥DSM elevation -0.1m, then mark it as occupied; otherwise mark it as idle.

[0020] The final VM model, in its relational form, intuitively depicts the distribution of obstacles on the ground and in the vertical space within the target area, providing a three-dimensional spatial carrier for subsequent line-of-sight penetration analysis. Based on the constructed local VM model, the camera viewpoints and attitudes of the UAV's historical observations are reproduced. Line-of-sight penetration analysis is performed on each image covering the target grid to determine whether the line of sight is obstructed by obstacles. The specific steps are as follows: Starting from the camera viewpoint, a ray is generated along the camera's optical axis, corresponding to the imaging position of the target grid in the image. The direction vector of the ray is calculated from the camera's attitude parameters. Traverse each voxel along the ray direction and check if the ray intersects with an occupied voxel: if it intersects, the line of sight is blocked and the ray endpoint is an invisible area; if it does not intersect, the ray reaches the grid surface and the surface area is considered a visible area. Repeat the above steps for all surface voxels within the target grid, i.e., traverse in the x and y directions, and record the state of each surface voxel being covered by at least one line of sight ray or not covered by any ray. For example, if the target grid is a dense forest area, the rays of a vertically viewed image will be obscured by tree canopy voxels, and the ground voxels below the tree canopy will be marked as invisible; while the rays of an obliquely viewed image may bypass the edge of the tree canopy and reach some ground voxels, which will be marked as visible. Three-dimensional visibility ratio (TVR) is used to quantify the actual effective coverage ratio of a target raster under existing observation conditions and is a core indicator for coverage effectiveness correction. It is defined as the ratio of the visible surface area within a raster to the total surface area of ​​the raster, and the calculation formula is as follows: ;in, The visible surface area within the grid is the sum of the areas of all surface voxels marked as visible through line-of-sight analysis. TVR is the total surface area of ​​the grid, that is, the projected area of ​​the grid on the horizontal plane. The value range of TVR is [0,1]. TVR=1 means that the grid is completely covered, that is, there are no blind spots. TVR=0 means that the grid is completely uncovered, that is, there are no blind spots. A three-dimensional visibility threshold θ is set to correct the determination result of the two-dimensional coverage effectiveness. The three-dimensional visibility threshold θ is set according to the coverage requirements of the monitoring task, and can be set as follows: three-dimensional visibility threshold θ = 0.95 for high-precision monitoring, three-dimensional visibility threshold θ = 0.9 for conventional monitoring, and three-dimensional visibility threshold θ = 0.85 for low-precision monitoring. For a grid with invalid 2D coverage, if TVR ≥ θ, it indicates that although the 2D assessment determines it to be invalid, the actual effective coverage ratio meets the requirements and should be corrected to effective coverage; if TVR < θ, it indicates that the actual coverage is insufficient, the invalid coverage determination is maintained, and coverage needs to be supplemented through subsequent revisit aerial surveys.

[0021] In step S4, for grid cells that still have invalid coverage after correction, the coverage improvement is calculated based on the change in 3D visibility ratio, and the revisit benefit ratio is calculated in conjunction with the number of revisits. Revisit routes are generated by sorting and classifying the revisit benefit ratios and clustering them according to spatial proximity. After each revisit, the change in viewing angle and the 3D visibility ratio are recalculated. Based on the convergence criteria, a grid coverage result set is output, which includes: For grids that are still determined to be invalid coverage after 3D correction, a UAV revisit aerial survey plan needs to be developed. In order to improve the efficiency of aerial survey and avoid waste of resources, the revisit benefit ratio (RBR) is introduced to dynamically adjust the revisit priority, and a convergence control mechanism is set up to ensure efficient convergence of the coverage correction process. Coverage Improvement (CI) characterizes the effect of revisit aerial surveys on improving grid coverage. It is defined as the difference between the current TVR and the initial TVR, and the calculation formula is as follows: ;in, The three-dimensional visibility ratio is the TVR obtained from the initial three-dimensional analysis after the initial aerial survey. To measure the three-dimensional visibility ratio after performing the RC revisit aerial survey; This refers to the number of revisits, i.e., the number of aerial surveys added to this grid. The value range of CI is [0, 1-TVR]. The larger the CI, the more significant the coverage improvement effect brought by the revisit aerial survey. The Revisit Benefit Ratio (RBR) characterizes the average coverage improvement efficiency per revisit aerial survey, defined as the ratio of coverage improvement to the number of revisits. The calculation formula is as follows: ; A higher RBR value indicates a better average coverage improvement per revisit and a higher revisit benefit; a lower RBR value indicates a lower marginal benefit from revisiting. For example: If a grid has a TVR of 0.6 (i.e., initial coverage of 60%), after the first revisit the TVR becomes 0.8, then CI = 0.2, RC = 1, and RBR = 0.2; after the second revisit the TVR becomes 0.9, then CI = 0.3, RC = 2, and RBR = 0.15. It can be seen that as the number of revisits increases, RBR gradually decreases, reflecting the law of diminishing marginal returns. Each time a revisit aerial survey plan is formulated, the current revisit benefit ratio (RBR) is calculated for all invalid coverage graticles, and graticles with high RBR are sorted from highest to lowest. Revisits are then prioritized for graticles with high RBR. The specific strategy is as follows: Set revisit priority tier thresholds, including a first-level revenue threshold. With secondary return threshold ,and Greater than Suitable for general scenarios For 0.10, The value is 0.05. Invalid rasters are classified according to the following criteria: when the RBR is not less than 0.05. When the RBR is not less than 1, it is listed as a first-priority grid and revisited first as a high-yield grid; and less than When the RBR is less than 1, it is classified as a second-priority grid and revisited as the second-priority grid in the medium-yield grid. It is classified as a level 3 priority grid and is temporarily deferred for revisiting as a low-yield grid. Based on the priority ranking results, the revisit route is planned using the principle of proximity clustering: the first-priority grid is used as the cluster core, and adjacent high-priority grids are included in the same route task cluster according to the spatial proximity relationship of the grid center point. Under the premise of not exceeding the single-flight aerial survey constraint, some second-priority grids are absorbed, thereby reducing the round-trip flight distance of the UAV. During the revisit process, for each grid cell to be revisited, the target incident angle that differs significantly from the set of historical observation incident angles is prioritized. This ensures that the angle change magnitude IV of the grid cell after supplementary observation satisfies the criterion of the angle diversity threshold τ, i.e., IV is not less than τ after supplementation, so as to specifically avoid blind spots formed by historical occlusion directions. A convergence control threshold is set to determine when to stop revisiting a single grid or the global region, avoiding invalid and repeated aerial surveys. A dual-threshold convergence mechanism is adopted, including a revisit benefit convergence threshold and an absolute coverage convergence threshold, and corresponding coverage improvement tolerance threshold and global convergence ratio threshold are set, as follows: Set a revisit reward convergence threshold The value can be set to a range of 0.01 to 0.03, which is recommended for general scenarios. The value is 0.02 when the RBR of a certain grid is less than 0.02. When the grid is determined to be in a state of insufficient marginal benefit, it is stopped from being included in the subsequent revisit queue. An absolute coverage convergence threshold θ is set to determine whether the 3D visibility ratio (TVR) meets the target requirements. When the TVR of a certain grid is not less than θ and the TVR improvement of two consecutive revisits is not greater than the coverage improvement tolerance threshold ε, the coverage correction process of the grid is determined to be converged, and revisits are stopped. The value of ε can be 0.01. Set a global convergence ratio threshold η. For typical scenarios, η is recommended to be 0.90. When the proportion of graticles within the monitoring area that meet the following conditions is not less than η, the global coverage correction process is considered converged, and the entire revisit aerial survey process is terminated: graticles with valid coverage have a TVR of not less than θ; graticles with converged coverage have an RBR of less than θ. Or, the criteria for stopping visits are met if TVR is not less than θ and the two consecutive increases are not greater than ε. When the global convergence ratio threshold η is met, the revisit aerial survey process is terminated, and the monitoring area grid coverage result set is output: for each grid number GI, the final three-dimensional visibility ratio (TVR), revisit count (RC), coverage improvement (CI), revisit benefit ratio (RBR), and coverage status label are given; among them, the list of invalid coverage grids is used as input for the next round of revisit aerial survey plan and view supplement selection, and the valid coverage grids are used to form coverage measurement results and serve as the benchmark data for subsequent ecological monitoring spatial sampling deployment, key area verification, and multi-period aerial survey coverage consistency comparison.

[0022] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0023] The above embodiments can be implemented, in whole or in part, by software, hardware, firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, in the form of a computer program product.

[0024] Those skilled in the art will recognize that the modules and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and inventive constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0025] In addition, the functional modules in the various embodiments of this application can be integrated into one processing module, or each module can exist physically separately, or two or more modules can be integrated into one module.

[0026] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A geographic grid-based UAV ecological monitoring photogrammetry coverage measurement method, characterized in that, The method comprises the steps of: Step S1, obtaining the boundary of the monitoring area and unifying the coordinates, dividing the two-dimensional grid according to the grid length and generating the grid number, binding the elevation difference parameter, the crown height parameter and the observation stability parameter for each grid; Step S2, collecting the incident angle set covering each grid, calculating the view angle change amplitude as the difference between the maximum and minimum values of the incident angle set, and comparing it with the view angle diversity threshold value, determining the two-dimensional coverage effective when meeting the threshold value, and calculating the occlusion risk parameter according to the elevation difference parameter, the crown height parameter and the observation stability parameter when not meeting the threshold value, and comparing it with the occlusion risk threshold value to determine the two-dimensional coverage effective or invalid; Step S3, constructing a local three-dimensional voxel model for the two-dimensional coverage invalid grid, voxelizing the terrain and crown layer three-dimensional data in the target grid and the adjacent range and marking the occupation state; reappearing the historical observation posture to perform the line-of-sight penetration analysis, summarizing the visible ground voxels and calculating the three-dimensional visibility ratio, comparing the three-dimensional visibility ratio with the three-dimensional visibility threshold value to correct the coverage state; Step S4, for the grid still invalid after correction, calculating the coverage improvement amount according to the three-dimensional visibility ratio change, and calculating the revisit benefit ratio combined with the revisit frequency; sorting and grading according to the revisit benefit ratio, and clustering to generate the revisit route according to the spatial proximity relationship, recalculating the view angle change amplitude and the three-dimensional visibility ratio after the revisit, and outputting the grid coverage result set according to the convergence criterion.

2. The method according to claim 1, wherein, In step S1, the elevation difference parameter, the crown height parameter and the observation stability parameter for each grid are bound, which includes: The elevation difference parameter is obtained by calculating the difference between the highest point elevation and the lowest point elevation in the grid; The crown height parameter is obtained by calculating the difference between the average crown top elevation and the average ground elevation in the grid range; The observation stability parameter is obtained by calculating the proportion of the number of aerial survey images meeting the preset imaging quality standard covering the grid to the total number of aerial survey images covering the grid. 3.The method of claim 2, wherein, The preset imaging quality standard includes the evaluation of image clarity and exposure appropriateness.

4. The method of claim 1, wherein, In step S2, the blind area risk probability of the three binding parameters of ER, CH and OS under the current two-dimensional observation condition is evaluated, and the occlusion risk parameter is calculated by the following method: ; wherein, OS is the observation stability, is the vegetation scale constant, is the terrain scale constant.

5. The method of claim 4, wherein, The specific value of the occlusion risk threshold value is set according to the requirement of the ecological monitoring task on the coverage accuracy.

6. The method of claim 1, wherein, In step S3, the three-dimensional visibility ratio is the ratio of the sum of the projection areas of all visible ground voxels in the grid on the horizontal plane to the total projection area of the grid on the horizontal plane.

7. The method of claim 6, wherein the method further comprises: The specific value of the three-dimensional visibility threshold value is set according to the requirement of the monitoring task on the completeness of the regional coverage.

8. The method of claim 1, wherein, In step S4, the coverage improvement amount is obtained by calculating the difference between the three-dimensional visibility ratio after the revisit aerial survey and the initial three-dimensional visibility ratio; the revisit benefit ratio is obtained by calculating the quotient of the coverage improvement amount divided by the revisit frequency.

9. The method of claim 8, wherein, The sorting and grading according to the revisit benefit ratio includes: setting at least two levels of benefit threshold values, comparing the revisit benefit ratio of each grid with these benefit threshold values, and dividing the revisit grids into different priority levels. 10.The method of claim 1, wherein, The convergence criterion includes at least one of the following: a revisit benefit convergence threshold value for judging whether the marginal benefit of a single revisit is too low, an absolute coverage convergence threshold value for judging whether the three-dimensional visibility ratio of the grid meets the target requirement, and a global convergence proportion threshold value for judging whether the coverage correction process of the entire monitoring area is completed.

Images