Forest volume dynamic exploration system and method based on unmanned aerial vehicle multispectral fusion

Abstract

The invention specifically relates to a forest stock dynamic exploration system and method based on unmanned aerial vehicle multispectral fusion, and relates to the technical field of forest resource monitoring. A data preprocessing module; a feature extraction module; a model construction and estimation module; and a result output and visualization module. According to the method, a multi-dimensional data fusion and dynamic calibration mechanism changes the limitation of a traditional fixed soil regulation coefficient; when the soil regulation coefficient is obtained, four core factors of soil type, terrain illumination, soil humidity and texture partition distance are comprehensively considered; interference of different soil backgrounds, terrain illumination and humidity changes on vegetation index calculation is effectively eliminated, the enhanced vegetation index can reflect the coverage and health condition of forest vegetation more truly, a high-precision data basis is provided for subsequent forest stock estimation based on the vegetation index, and compared with a traditional method, the method has the advantage that the method is more accurate. And the data accuracy and reliability are improved.

Description

Technical Field

[0001] The present invention relates to the technical field of forest resource monitoring, and in particular to a forest volume dynamic exploration system and method based on unmanned aerial vehicle (UAV) multispectral fusion. Background Art

[0002] In the field of forest resource monitoring, accurate estimation of forest stock is of great significance to forestry management, ecological protection and carbon sink accounting.

[0003] Currently, using remote sensing technology to calculate vegetation indices and then estimate forest volume is a common method. Among them, the Enhanced Vegetation Index (EVI) is widely used because it can effectively reflect vegetation coverage and growth conditions. However, existing technologies have many shortcomings in the calculation process of the Enhanced Vegetation Index: First, the soil adjustment coefficient in traditional enhanced vegetation index calculations often uses fixed empirical values, which does not fully consider the spatial heterogeneity of factors such as soil type, topography, and humidity; For example, in actual forest environments, different soil textures such as sand and clay are mixed and distributed, and their spectral reflectance characteristics are significantly different. The fixed soil adjustment coefficient cannot accurately eliminate soil background interference, resulting in deviations in the calculation of the enhanced vegetation index and affecting the accuracy of forest stock estimation.

[0004] Second, the impact of topographic factors on the calculation of the Enhanced Vegetation Index is often overlooked. Differences in illumination conditions across slopes and aspects can alter the spectral reflectance of soil and vegetation. Existing methods lack a mechanism to effectively correct for topographically induced illumination changes and incorporate soil conditioning coefficient adjustments. This makes it difficult for the Enhanced Vegetation Index to accurately reflect vegetation in complex terrain areas such as mountains and hills. Third, soil moisture, a key factor affecting soil spectral reflectance, is not fully quantified in the current EVI calculations. Soil moisture changes can significantly alter soil reflectance, especially in humid or arid regions. If these spectral changes caused by humidity differences are not reflected in the soil conditioning coefficient, the EVI's ability to monitor vegetation will be severely weakened. Therefore, a technical solution is needed that can comprehensively consider multiple factors and dynamically adjust the soil adjustment coefficient to improve the calculation accuracy of the enhanced vegetation index and the estimation level of forest stock. Summary of the Invention

[0005] The purpose of the present invention is to solve the above problems and propose a dynamic exploration system and method for forest volume based on multi-spectral fusion of unmanned aerial vehicles.

[0006] In order to achieve the above object, the present invention adopts the following technical solutions: The forest volume dynamic exploration system based on UAV multispectral fusion includes: Data acquisition module: obtains raw data related to forest volume and vegetation index to provide a basis for subsequent analysis; Data preprocessing module: cleans, calibrates and standardizes raw data to eliminate noise and errors and improve data quality; Feature extraction module: Extracts key features related to forest volume from preprocessed data to provide input variables for model construction. In the process of obtaining the enhanced vegetation index among key features, a comprehensive analysis of soil conditioning-related parameters is performed to obtain the soil corresponding coefficient to match the corresponding soil conditioning parameters. Model building and estimation module: Use feature data to build a regression model to achieve quantitative estimation of forest volume and verify model accuracy; Result output and visualization module: presents the model estimation results in an intuitive form and outputs quantitative reports to assist decision-making applications.

[0007] Preferably, the data acquisition module specifically includes: The appropriate drone is selected based on the actual needs of the exploration mission, and the drone flight parameters are adjusted according to the actual terrain and vegetation conditions; The drone is equipped with a multispectral camera to obtain spectral information of forest vegetation through multiple spectral channels; at the same time, a high-precision positioning system is used to record the drone's position and posture in real time, and a large-capacity storage device is used to save the collected data.

[0008] Preferably, the data preprocessing module specifically includes: Optimize the original collected data, eliminate the influence of sensors and atmospheric factors through radiation correction, and restore the true spectral characteristics of the ground objects; The geometric correction algorithm is used in combination with ground control points to correct the geometric deformation of the image and assign accurate geographic coordinates; the feature point matching method is used for image registration to align multiple images in space.

[0009] Preferably, the feature extraction module includes: Mining key information from pre-processed data, extracting single-band reflectance to reflect vegetation spectral characteristics, calculating multiple vegetation indices including the Normalized Difference Vegetation Index and the Enhanced Vegetation Index to enhance vegetation information, and using the gray-level co-occurrence matrix method to extract texture features to describe forest structure; Through multi-dimensional feature extraction, the original image data is converted into feature parameters closely related to forest stock volume, providing effective data support for forest stock volume estimation.

[0010] Preferably, the process of obtaining the enhanced vegetation index includes: The Enhanced Vegetation Index (EVI) is an important indicator used in remote sensing to quantify vegetation coverage, health status, and biomass. Its calculation formula is: ; Where G represents the gain coefficient, which is usually set to 2.5; C1 and C2 are 6 and 7.5 respectively, which are used to adjust the weight of the red light band in the calculation; B is the reflectivity of blue light band; NIR is the reflectance in the near-infrared band; R is the reflectivity of red light band; L is the soil adjustment coefficient; Among them, the reflectance data of the blue light B, red light R and near-infrared NIR bands of the multispectral image are obtained through the multispectral camera of the drone, and are pre-processed with radiation correction and geometric correction to eliminate sensor errors, atmospheric effects and geometric deformation; then the processed reflectance data are substituted into the formula to obtain the enhanced vegetation index.

[0011] Preferably, the process of obtaining the soil adjustment coefficient includes: Based on the soil type, the soil adjustment coefficient range of the forest is obtained by referring to the known empirical range of soil adjustment coefficients, and the median value is extracted from the range as the soil adjustment reference coefficient; Using a terrain radiation correction model, all slopes in the forest terrain are obtained, and the maximum slope is extracted as the reference slope. Preset monitoring points are set up at various altitudes on the reference slope, and the light values ​​at each monitoring point are obtained. Arrange the illumination values ​​in descending order according to their numerical values, extract the maximum illumination value and the minimum illumination value, and calculate the difference between the maximum illumination value and the minimum illumination value to obtain the illumination difference; A standard value of the light difference is preset, the light difference is divided by the standard value of the light difference, and the obtained value is multiplied by the soil adjustment reference coefficient to obtain the light adjustment coefficient; Using microwave remote sensing data, soil moisture at each preset location in the forest is obtained. The moisture values ​​at each location are sorted in descending order, and the three largest moisture values ​​and their corresponding forest locations are extracted. The forest location centers corresponding to the three largest moisture values ​​are obtained, and the three obtained location centers are connected in sequence with straight lines to form a complete triangle. The area of ​​the triangle is calculated and divided by the area of ​​the forest to obtain the percentage. The humidity adjustment coefficient is obtained by multiplying the soil adjustment reference coefficient and the proportion.

[0012] Preferably, the method further includes: Based on images collected by drones, forest soils were divided into regions based on soil texture to obtain various types of soil texture areas. The range of soil adjustment coefficients for each soil texture area was determined by referring to the known empirical range of soil adjustment coefficients. The median value within the range was extracted as the adjustment comparison coefficient for the corresponding soil texture area. The maximum and minimum adjustment comparison coefficients were then extracted, as well as the corresponding regional locations of the maximum and minimum adjustment comparison coefficients. Take the center of the area corresponding to each maximum adjustment comparison coefficient as the origin, and the center of the area corresponding to each minimum adjustment comparison coefficient as the end point, connect the origin and the end point with a straight line to obtain the length value; Arrange all length values ​​in ascending order according to their numerical values ​​and extract the minimum length value; determine the soil adjustment comparison coefficient of the two soil texture areas corresponding to the minimum length value; Take half of the minimum length value and multiply it by the soil adjustment comparison coefficient of the two corresponding soil texture areas, sum them up to get the preliminary coefficient, divide the preliminary coefficient by the minimum length value to get the target interpolation coefficient; Subtract the humidity adjustment coefficient from the soil adjustment reference coefficient, then sum it with the humidity adjustment coefficient and the target interpolation coefficient to obtain the soil corresponding coefficient; K groups of threshold value ranges are preset, and the value range of each group of threshold values ​​corresponds to a soil adjustment coefficient. The soil corresponding coefficient is matched with the value range of the k groups of threshold values ​​to obtain the soil adjustment coefficient corresponding to the soil corresponding coefficient.

[0013] Preferably, the model building and estimation module specifically includes: Taking the feature extraction data as input, stepwise regression was used to screen key variables, and the random forest algorithm was used to construct a forest volume estimation model; Optimize model parameters through training sets, evaluate model accuracy with test sets, evaluate model performance using coefficient of determination and root mean square error indicators, and adjust and optimize the model based on the evaluation results; Finally, the actual data will be input into the model to achieve an accurate estimation of forest stock and provide a quantitative basis for forest resource management.

[0014] Preferably, the result output and visualization module specifically includes: The model estimation results are presented intuitively and detailed data is output in CSV form, including the estimated value of accumulation and related statistical information; ArcGIS software was used to map the area, overlaying the estimated forest volume with the digital elevation model and land use data. A graded color scheme was used to divide forest volume into different levels, each represented by a different color. Using 3D visualization software, forest volume data is combined with terrain data to generate a 3D visualization model; Through rotation and zooming operations, users can observe the spatial distribution of forest stock from different angles and gain a more intuitive understanding of the current status and changing trends of forest resources.

[0015] The dynamic exploration method of forest volume based on UAV multispectral fusion includes: Data collection and preprocessing: Obtain raw data related to forest volume and vegetation index to provide a basis for subsequent analysis; clean, calibrate, and standardize the raw data to eliminate noise and errors and improve data quality; Feature extraction: Extract key features related to forest volume from preprocessed data to provide input variables for model construction; Model construction and estimation: Use characteristic data to build a regression model to achieve quantitative estimation of forest volume and verify the accuracy of the model.

[0016] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention changes the limitations of the traditional fixed soil adjustment coefficient through multi-dimensional data fusion and dynamic calibration mechanism. When obtaining the soil adjustment coefficient, four core factors are comprehensively considered: soil type, terrain illumination, soil moisture, and texture zoning distance. This effectively eliminates the interference of different soil backgrounds, terrain illumination, and humidity changes on the vegetation index calculation, allowing the enhanced vegetation index to more realistically reflect the coverage and health of forest vegetation, providing a high-precision data foundation for subsequent forest stock estimation based on the vegetation index, and improving data accuracy and reliability compared to traditional methods.

[0017] 2. The present invention forms a complete and efficient dynamic exploration system for forest stock by constructing five major modules: data acquisition, preprocessing, feature extraction, model construction and estimation, and result output and visualization. The modules are closely connected and work together, which not only realizes the full process automation from data acquisition to result output, greatly improving exploration efficiency, but also adapts to the forest environment under different terrain, climate and soil conditions through dynamic adjustment strategies, providing strong technical support for the dynamic monitoring and scientific management of forest resources, and has broad application prospects and promotion value. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Further details, features and advantages of the present application are disclosed in the following description of exemplary embodiments in conjunction with the accompanying drawings, in which: Figure 1 Flowchart of the present invention. DETAILED DESCRIPTION

[0019] Several embodiments of the present application will be described in more detail below with reference to the accompanying drawings so that those skilled in the art can implement the present application. The present application can be embodied in many different forms and for many different purposes and should not be limited to the embodiments described herein. These embodiments are provided to make the present application comprehensive and complete and to fully convey the scope of the present application to those skilled in the art. The embodiments do not limit the present application.

[0020] Unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. It will be further understood that terms such as those defined in commonly used dictionaries should be interpreted as having a meaning consistent with their meaning in the relevant art and / or the context of this specification, and will not be interpreted in an idealized or overly formal sense unless expressly defined as such herein.

[0021] See also Figure 1 As shown, the present invention provides a technical solution: The forest volume dynamic exploration system based on UAV multispectral fusion includes: Data acquisition module: obtains raw data related to forest volume and vegetation index to provide a basis for subsequent analysis; Specifically include: The appropriate drone is selected based on the actual needs of the exploration mission, and the drone flight parameters are adjusted according to the actual terrain and vegetation conditions; The drone is equipped with a multispectral camera to obtain spectral information of forest vegetation through multiple spectral channels. At the same time, a high-precision positioning system is used to record the drone's position and posture in real time, and a large-capacity storage device is used to save the collected data. Data preprocessing module: cleans, calibrates and standardizes raw data to eliminate noise and errors and improve data quality; Specifically include: Optimize the original collected data, eliminate the influence of sensors and atmospheric factors through radiation correction, and restore the true spectral characteristics of the ground objects; During the radiometric correction process, we utilize laboratory measurement data from standard reflectance panels (such as Spectralon diffuse reflectance panels) combined with atmospheric correction algorithms such as FLAASH. We comprehensively consider the scattering and absorption effects of atmospheric molecules and aerosols, as well as the sensor's own response characteristics, to accurately convert the digital quantized values ​​of the original image into the true reflectance of the surface, eliminating spectral distortion caused by environmental factors. Use geometric correction algorithms and ground control points to correct image geometric deformation and assign accurate geographic coordinates; use feature point matching methods for image registration to align multiple images in space; During geometric correction, firstly, ground control points (GCPs) are reasonably laid out in the exploration area, generally 6 to 8 per square kilometer. Their precise coordinates are obtained using a total station or high-precision GPS receiver (the error is controlled within 5 cm). Then, a second-order polynomial transformation model is used to fit the GCP coordinate relationship through the least squares method to correct the geometric deformation of the image, so that the image has accurate geographic coordinates and spatial position relationships. In the image registration process, a SIFT-based feature point matching method is used to first extract feature points from the image. Mismatched points are then removed through bidirectional matching and the RANSAC algorithm. Finally, based on pairs of points with the same name, an affine transformation or perspective transformation model is used to achieve precise spatial alignment of multiple images, ensuring data consistency and comparability, and laying a solid foundation for subsequent feature extraction and model analysis. Feature extraction module: Extracts key features related to forest volume from preprocessed data to provide input variables for model construction. In the process of obtaining the enhanced vegetation index among key features, a comprehensive analysis of soil conditioning-related parameters is performed to obtain the soil corresponding coefficient to match the corresponding soil conditioning parameters. include: Mining key information from pre-processed data, extracting single-band reflectance to reflect vegetation spectral characteristics, calculating multiple vegetation indices including the Normalized Difference Vegetation Index and the Enhanced Vegetation Index to enhance vegetation information, and using the gray-level co-occurrence matrix method to extract texture features to describe forest structure; Through multi-dimensional feature extraction, the original image data is converted into characteristic parameters closely related to forest volume, providing effective data support for forest volume estimation; The process of obtaining the enhanced vegetation index includes: The Enhanced Vegetation Index (EVI) is an important indicator used in remote sensing to quantify vegetation coverage, health status, and biomass. Its calculation formula is: ; Where G represents the gain coefficient, which is usually set to 2.5. It is used to amplify the spectral difference between vegetation and non-vegetation, enhance the weight of vegetation signals in the index calculation, and make vegetation information more prominent. In dense forest vegetation areas, a high gain coefficient can effectively distinguish the differences in different vegetation coverage levels. C1 and C2 are 6 and 7.5 respectively, which are used to adjust the weight of the red light band in the calculation. By suppressing the influence of atmospheric scattering and soil background noise in the red light band, the accuracy and stability of the enhanced vegetation index are improved. For example, in areas with complex atmospheric conditions, reasonable C1 and C2 values ​​can reduce the vegetation index error caused by atmospheric interference; B is the reflectance of the blue light band. Introducing the blue light band reflectance, taking advantage of the blue light's sensitivity to chlorophyll concentration in vegetation, in synergy with the red and near-infrared bands, further enhances the ability to monitor the physiological state of vegetation. For example, when monitoring the nutritional status of forest vegetation, the blue light band can provide unique supplementary information. NIR is the reflectance in the near-infrared band; R is the reflectivity of red light band; L is the soil adjustment coefficient, which reduces the impact of soil background on vegetation index; The multispectral camera of the drone acquires the reflectance data of the blue light B, red light R and near-infrared (NIR) bands of the multispectral image, and then undergoes radiation correction and geometric correction preprocessing to eliminate sensor errors, atmospheric effects and geometric deformation. The processed reflectance data is then substituted into the formula to obtain the enhanced vegetation index. The process of obtaining the soil adjustment coefficient includes: Based on the soil type, the soil adjustment coefficient range of the forest is obtained by referring to the known empirical range of soil adjustment coefficients, and the median value is extracted from the range as the soil adjustment reference coefficient; Using a terrain radiation correction model, all slopes in the forest terrain are obtained, and the maximum slope is extracted as the reference slope. Preset monitoring points are set up at various altitudes on the reference slope, and the light values ​​at each monitoring point are obtained. Arrange the illumination values ​​in descending order according to their numerical values, extract the maximum illumination value and the minimum illumination value, and calculate the difference between the maximum illumination value and the minimum illumination value to obtain the illumination difference; A standard value of the light difference is preset, the light difference is divided by the standard value of the light difference, and the obtained value is multiplied by the soil adjustment reference coefficient to obtain the light adjustment coefficient; Using microwave remote sensing data, soil moisture at each preset location in the forest is obtained. The moisture values ​​at each location are sorted in descending order, and the three largest moisture values ​​and their corresponding forest locations are extracted. The forest location centers corresponding to the three largest moisture values ​​are obtained, and the three obtained location centers are connected in sequence with straight lines to form a complete triangle. The area of ​​the triangle is calculated and divided by the area of ​​the forest to obtain the percentage. The humidity adjustment coefficient is obtained by multiplying the soil adjustment reference coefficient and the proportion; Based on images collected by drones, forest soils were divided into regions based on soil texture to obtain various types of soil texture areas. The range of soil adjustment coefficients for each soil texture area was determined by referring to the known empirical range of soil adjustment coefficients. The median value within the range was extracted as the adjustment comparison coefficient for the corresponding soil texture area. The maximum and minimum adjustment comparison coefficients were then extracted, as well as the corresponding regional locations of the maximum and minimum adjustment comparison coefficients. Take the center of the area corresponding to each maximum adjustment comparison coefficient as the origin, and the center of the area corresponding to each minimum adjustment comparison coefficient as the end point, connect the origin and the end point with a straight line to obtain the length value; Arrange all length values ​​in ascending order according to their numerical values ​​and extract the minimum length value; determine the soil adjustment comparison coefficient of the two soil texture areas corresponding to the minimum length value; Take half of the minimum length value and multiply it by the soil adjustment comparison coefficient of the two corresponding soil texture areas, sum them up to get the preliminary coefficient, divide the preliminary coefficient by the minimum length value to get the target interpolation coefficient; Subtract the humidity adjustment coefficient from the soil adjustment reference coefficient, then sum it with the humidity adjustment coefficient and the target interpolation coefficient to obtain the soil corresponding coefficient; Preset k groups of threshold value ranges, each threshold value range corresponds to a soil adjustment coefficient, match the soil corresponding coefficient with the k groups of threshold value ranges to obtain the soil adjustment coefficient corresponding to the soil corresponding coefficient; The soil adjustment coefficient acquisition process uses multi-dimensional data fusion and dynamic calibration, organically combining factors such as soil type empirical benchmarks, terrain illumination influences, soil moisture spatial distribution, and texture zoning distance effects. This allows for rapid establishment of benchmark values ​​using prior knowledge, while also enabling real-time response to environmental factors through terrain radiation correction and moisture geometric quantification. Spatial interpolation can also be used to address the heterogeneity of mixed soils. Ultimately, threshold matching is used to ensure that the results are consistent with actual application scenarios, significantly improving the adaptability of the soil adjustment coefficient to complex forest environments and providing strong support for the accurate calculation of vegetation indices and improving the accuracy of forest volume estimation. Model building and estimation module: Use feature data to build a regression model to achieve quantitative estimation of forest volume and verify model accuracy; Specifically include: Taking the feature extraction data as input, stepwise regression was used to screen key variables, and the random forest algorithm was used to construct a forest volume estimation model; It includes: taking forest volume as the dependent variable, and single-band reflectance, vegetation index, texture characteristics, etc. as independent variables, and gradually introducing and eliminating variables to select variables that have a significant impact on the dependent variable and do not have multicollinearity.

[0022] Optimize model parameters through training sets, evaluate model accuracy with test sets, evaluate model performance using coefficient of determination and root mean square error indicators, and adjust and optimize the model based on the evaluation results; Ultimately, actual data will be input into the model to achieve accurate estimation of forest volume and provide a quantitative basis for forest resource management; Specifically, the feature-extracted image data is fed into a trained random forest model. Each decision tree predicts the input data, generating an estimated forest volume. The predictions from all decision trees are then averaged to produce the final forest volume estimate. For each pixel or plot, a corresponding volume estimate is generated. Result output and visualization module: presents the model estimation results in an intuitive form and outputs quantitative reports to assist decision-making applications; Specifically include: The model estimation results are presented intuitively and detailed data is output in CSV form, including the estimated value of accumulation and related statistical information; ArcGIS software was used to map the area, overlaying the estimated forest volume with the digital elevation model and land use data. A graded color scheme was used to divide forest volume into different levels, each represented by a different color. At the same time, map elements such as legends, scales, and compasses are added to produce intuitive and beautiful thematic maps of forest volume spatial distribution. Diverse output methods facilitate users to quickly understand and analyze forest volume data and assist in decision-making. Using 3D visualization software, forest volume data is combined with terrain data to generate a 3D visualization model; Through rotation and zooming operations, users can observe the spatial distribution of forest stock from different angles and gain a more intuitive understanding of the current status and changing trends of forest resources.

[0023] The dynamic exploration method of forest volume based on UAV multispectral fusion includes: Data collection and preprocessing: Obtain raw data related to forest volume and vegetation index to provide a basis for subsequent analysis; clean, calibrate, and standardize the raw data to eliminate noise and errors and improve data quality; Feature extraction: Extract key features related to forest volume from preprocessed data to provide input variables for model construction; Model construction and estimation: Use characteristic data to build a regression model to achieve quantitative estimation of forest volume and verify the accuracy of the model.

[0024] The above formulas are obtained by collecting a large amount of data and performing software simulation, and a formula close to the actual value is selected. The influencing weight factors and specific coefficient values ​​in the formula are set by technical personnel in this field according to actual conditions, and can be adjusted and modified later.

[0025] The above description of the embodiments is intended to enable one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention is not limited to the embodiments shown herein but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A dynamic forest volume exploration system based on multi-spectral fusion of UAVs is characterized by: include: Data acquisition module: obtains raw data related to forest volume and vegetation index to provide a basis for subsequent analysis; Data preprocessing module: cleans, calibrates and standardizes raw data to eliminate noise and errors and improve data quality; Feature extraction module: Extracts key features related to forest volume from preprocessed data to provide input variables for model construction. In the process of obtaining the enhanced vegetation index among key features, a comprehensive analysis of soil conditioning-related parameters is performed to obtain the soil corresponding coefficient to match the corresponding soil conditioning parameters. Model building and estimation module: Use feature data to build a regression model to achieve quantitative estimation of forest volume and verify model accuracy; Result output and visualization module: presents the model estimation results in an intuitive form and outputs quantitative reports to assist decision-making applications.

2. The forest volume dynamic exploration system based on multi-spectral fusion of unmanned aerial vehicles according to claim 1 is characterized in that: Data acquisition module, specifically including: The appropriate drone is selected based on the actual needs of the exploration mission, and the drone flight parameters are adjusted according to the actual terrain and vegetation conditions; The drone is equipped with a multispectral camera to obtain spectral information of forest vegetation through multiple spectral channels; at the same time, a high-precision positioning system is used to record the drone's position and posture in real time, and a large-capacity storage device is used to save the collected data.

3. The forest volume dynamic exploration system based on multi-spectral fusion of unmanned aerial vehicles according to claim 2 is characterized in that: Data preprocessing module, specifically including: Optimize the original collected data, eliminate the influence of sensors and atmospheric factors through radiation correction, and restore the true spectral characteristics of the ground objects; The geometric correction algorithm is used in combination with ground control points to correct the geometric deformation of the image and assign accurate geographic coordinates; the feature point matching method is used for image registration to align multiple images in space.

4. The forest volume dynamic exploration system based on UAV multispectral fusion according to claim 3 is characterized in that: The feature extraction module includes: Mining key information from pre-processed data, extracting single-band reflectance to reflect vegetation spectral characteristics, calculating multiple vegetation indices including the Normalized Difference Vegetation Index and the Enhanced Vegetation Index to enhance vegetation information, and using the gray-level co-occurrence matrix method to extract texture features to describe forest structure; Through multi-dimensional feature extraction, the original image data is converted into feature parameters closely related to forest stock volume, providing effective data support for forest stock volume estimation.

5. The forest volume dynamic exploration system based on UAV multispectral fusion according to claim 4 is characterized in that: The process of obtaining the enhanced vegetation index includes: The Enhanced Vegetation Index (EVI) is an important indicator used in remote sensing to quantify vegetation coverage, health status, and biomass. Its calculation formula is: ; Where G represents the gain coefficient, which is usually set to 2.5; C1 and C2 are 6 and 7.5 respectively, which are used to adjust the weight of the red light band in the calculation; B is the reflectivity of blue light band; NIR is the reflectance in the near-infrared band; R is the reflectivity of red light band; L is the soil adjustment coefficient; Among them, the reflectance data of the blue light B, red light R and near-infrared NIR bands of the multispectral image are obtained through the multispectral camera of the drone, and are pre-processed with radiation correction and geometric correction to eliminate sensor errors, atmospheric effects and geometric deformation; then the processed reflectance data are substituted into the formula to obtain the enhanced vegetation index.

6. The forest volume dynamic exploration system based on UAV multispectral fusion according to claim 5 is characterized in that: The process of obtaining the soil adjustment coefficient includes: Based on the soil type, the soil adjustment coefficient range of the forest is obtained by referring to the known empirical range of soil adjustment coefficients, and the median value is extracted from the range as the soil adjustment reference coefficient; Using a terrain radiation correction model, all slopes in the forest terrain are obtained, and the maximum slope is extracted as the reference slope. Preset monitoring points are set up at various altitudes on the reference slope, and the light values ​​at each monitoring point are obtained. Arrange each illumination value in descending order according to its numerical value, extract the maximum illumination value and the minimum illumination value, and calculate the difference between the maximum illumination value and the minimum illumination value to obtain the illumination difference; A standard value of the light difference is preset, the light difference is divided by the standard value of the light difference, and the obtained value is multiplied by the soil adjustment reference coefficient to obtain the light adjustment coefficient; Using microwave remote sensing data, soil moisture at each preset location in the forest is obtained. The moisture values ​​at each location are sorted in descending order, and the three largest moisture values ​​and their corresponding forest locations are extracted. The forest location centers corresponding to the three largest moisture values ​​are obtained, and the three obtained location centers are connected in sequence with straight lines to form a complete triangle. The area of ​​the triangle is calculated and divided by the area of ​​the forest to obtain the percentage. The humidity adjustment coefficient is obtained by multiplying the soil adjustment reference coefficient and the proportion.

7. The forest volume dynamic exploration system based on multi-spectral fusion of unmanned aerial vehicles according to claim 6 is characterized in that: Also includes: Based on images collected by drones, forest soils were divided into regions based on soil texture to obtain various types of soil texture areas. The range of soil adjustment coefficients for each soil texture area was determined by referring to the known empirical range of soil adjustment coefficients. The median value within the range was extracted as the adjustment comparison coefficient for the corresponding soil texture area. The maximum and minimum adjustment comparison coefficients were then extracted, as well as the corresponding regional locations of the maximum and minimum adjustment comparison coefficients. Take the center of the area corresponding to each maximum adjustment comparison coefficient as the origin, and the center of the area corresponding to each minimum adjustment comparison coefficient as the end point, connect the origin and the end point with a straight line to obtain the length value; Arrange all length values ​​in ascending order according to their numerical values ​​and extract the minimum length value; determine the soil adjustment comparison coefficient of the two soil texture areas corresponding to the minimum length value; Take half of the minimum length value and multiply it by the soil adjustment comparison coefficient of the two corresponding soil texture areas, sum them up to get the preliminary coefficient, divide the preliminary coefficient by the minimum length value to get the target interpolation coefficient; Subtract the humidity adjustment coefficient from the soil adjustment reference coefficient, then sum it with the humidity adjustment coefficient and the target interpolation coefficient to obtain the soil corresponding coefficient; K groups of threshold value ranges are preset, and the value range of each group of threshold values ​​corresponds to a soil adjustment coefficient. The soil corresponding coefficient is matched with the value range of the k groups of threshold values ​​to obtain the soil adjustment coefficient corresponding to the soil corresponding coefficient.

8. The forest volume dynamic exploration system based on multi-spectral fusion of unmanned aerial vehicles according to claim 1 is characterized in that: Model building and estimation module, including: Taking the feature extraction data as input, stepwise regression was used to screen key variables, and the random forest algorithm was used to construct a forest volume estimation model; Optimize model parameters through training sets, evaluate model accuracy with test sets, evaluate model performance using coefficient of determination and root mean square error indicators, and adjust and optimize the model based on the evaluation results; Finally, the actual data will be input into the model to achieve an accurate estimation of forest stock and provide a quantitative basis for forest resource management.

9. The forest volume dynamic exploration system based on UAV multispectral fusion according to claim 7 is characterized in that: Result output and visualization module, specifically including: The model estimation results are presented intuitively and detailed data is output in CSV form, including the estimated value of accumulation and related statistical information; ArcGIS software was used to map the area, overlaying the estimated forest volume with the digital elevation model and land use data. A graded color scheme was used to divide forest volume into different levels, each represented by a different color. Using 3D visualization software, forest volume data is combined with terrain data to generate a 3D visualization model; Through rotation and zooming operations, users can observe the spatial distribution of forest stock from different angles and gain a more intuitive understanding of the current status and changing trends of forest resources.

10. A method for dynamic exploration of forest volume based on multi-spectral fusion of unmanned aerial vehicles, according to any one of claims 1 to 9, wherein the method comprises: include: Data collection and preprocessing: Obtaining raw data related to forest volume and vegetation index to provide a basis for subsequent analysis; Clean, calibrate and standardize raw data to eliminate noise and errors and improve data quality; Feature extraction: Extract key features related to forest volume from preprocessed data to provide input variables for model construction; Model construction and estimation: Use characteristic data to build a regression model to achieve quantitative estimation of forest volume and verify the accuracy of the model.

Images