Automatic biomass sampling method, medium and equipment based on UAV remote sensing calculation

Through drone remote sensing calculation technology, rough remote sensing data are automatically processed and classified, and flight paths are generated, which solves the real-time and accuracy problems of existing sampling methods, and realizes efficient and high-precision biomass sampling, which can reflect vegetation characteristics in real time.

CN118155101BActive Publication Date: 2024-11-26ZHEJIANG TAILE GEOGRAPHIC INFORMATION TECH CO LTD +1
View PDF 9 Cites 0 Cited by

Patent Information

Application Number
CN202410325689.9
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-03-21
Publication Date
2024-11-26
Estimated Expiration
2044-03-21

AI Technical Summary

Technical Problem

The existing biomass sampling methods rely on past data and lack real-time and accuracy, especially in large-area remote sensing estimation, which cannot effectively cover vegetation characteristics, resulting in low sampling accuracy.

Method used

The automatic sampling method based on drone remote sensing calculation is adopted to calculate the distribution of land objects and generate flight paths through the processing and automatic classification of coarse remote sensing data, and achieve high-precision and efficient biomass sampling. Specific steps include multi-spectral data preprocessing, normalized vegetation index calculation, automatic classification of land objects, construction of classification models, generation of drone flight paths, and data collection.

Benefits of technology

High-efficiency and high-precision biomass sampling is achieved, which can reflect the growth characteristics and distribution of vegetation in real time, and improve the accuracy of biomass estimation.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118155101B_ABST
    Figure CN118155101B_ABST
Patent Text Reader

Abstract

The present invention provides an automatic biomass sampling method, medium and equipment based on remote sensing calculation of unmanned aerial vehicles. An automatic biomass sampling method based on remote sensing calculation of unmanned aerial vehicles comprises: based on the rough remote sensing data of the ground object, data processing is performed on the rough remote sensing data, and the ground object types in the remote sensing data are automatically classified; the attribute distribution α of each ground object is calculated; the automatic sampling method is calculated based on the attribute distribution α; the flight path of the sampling unmanned aerial vehicle is generated according to the sample data; after the flight path is set, the unmanned aerial vehicle is used to fly according to the set flight path and shoot and collect data to complete the automatic sampling. This method combines wide-area rough remote sensing and short-range high-precision remote sensing to achieve high-efficiency and high-precision sampling. The present invention can not only efficiently perform sampling and evaluation on vegetation biomass, but also ensure that the sampling method can reflect the real-time growth characteristics and distribution of vegetation, and achieve high-precision biomass estimation.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of biomass sampling technology, and more specifically, to an automatic biomass sampling method, medium and equipment based on unmanned aerial vehicle remote sensing calculation. Background Art

[0002] The estimation of vegetation biomass is a prerequisite for carbon sink assessment. The estimation of biomass is mainly divided into two types: direct measurement based on field sampling surveys and indirect measurement based on remote sensing technology and model extrapolation. Among them, the direct measurement method relies on the credibility of previous years' data and experience, and has disadvantages such as long span and large manpower requirements. The indirect measurement method uses satellites or drones to collect data, which can make the data completely objective and save manpower. However, remote sensing of large areas cannot make accurate estimates due to the long distance. The use of drones for high-precision remote sensing data can improve accuracy, but sampling is also required due to efficiency issues.

[0003] Existing sampling methods mainly rely on past data, typical sampling or random sampling. Existing typical sampling methods rely on past experience and are not real-time; random sampling cannot cover the overall characteristics of vegetation well, thus affecting sampling accuracy. Summary of the invention

[0004] The purpose of the present invention is to provide a method, medium and equipment for automatic sampling of biomass based on remote sensing calculation by unmanned aerial vehicle.

[0005] The present invention aims to solve the problems existing in the prior art.

[0006] Compared with the prior art, the technical solution of the present invention and its beneficial effects are as follows:

[0007] The first aspect disclosed by the present invention provides a method for automatic sampling of biomass based on remote sensing calculation using an unmanned aerial vehicle, comprising: based on rough remote sensing data of the ground objects, performing data processing on the rough remote sensing data, and automatically classifying the types of ground objects in the remote sensing data; calculating the attribute distribution α of each ground object; performing automatic sampling method calculation based on the attribute distribution α; generating a flight path for a sampling unmanned aerial vehicle according to the sample data; after the flight path is set, using the unmanned aerial vehicle to fly according to the set flight path and take pictures to collect data, thereby completing automatic sampling.

[0008] As a further improvement, the rough remote sensing data based on the ground objects performs data processing on the rough remote sensing data, and automatically classifies the types of ground objects in the remote sensing data, including: based on the rough remote sensing data of the ground objects, performs data preprocessing on the multispectral data in the rough remote sensing data, calculates the normalized vegetation index according to the rough remote sensing data and the results of the data preprocessing, and automatically classifies the types of ground objects in the remote sensing data according to the rough remote sensing data, the data preprocessing results and the normalized vegetation index.

[0009] As a further improvement, the rough remote sensing data based on the ground object performs data preprocessing on the multispectral data in the rough remote sensing data, and calculates the normalized vegetation index according to the rough remote sensing data and the results of the data preprocessing, including: using an image alignment method to spatially align the obtained RGB image and multispectral data; optically calibrating the camera; performing radiation calibration on the multispectral data; ground object reflectance calibration of the multispectral data; and fusing the ground object images to generate an orthophoto containing the entire shooting.

[0010] As a further improvement, the automatic classification of land object types in remote sensing data based on rough remote sensing data, data preprocessing results and normalized vegetation index includes: defining the number of land object types; constructing a classification model; constructing a training data set including a true value database of all types of land objects that have been labeled, and using the training data set to train the classification model; inputting test data into the classification model for predictive processing, and the classification model outputting land object prediction results; and obtaining the number and area of ​​plots corresponding to each type of land object based on the land object prediction results.

[0011] As a further improvement, the calculation of the attribute distribution α of each type of land feature includes: calculating the attribute distribution of each type of land feature Ci as a whole; and calculating the attribute distribution of each independent plot of each type of land feature Ci.

[0012] As a further improvement, the automatic sampling based on the attribute distribution α includes: formulating sampling conditions and performing automatic sampling processing according to the attribute distribution α of each type of land feature Ci.

[0013] As a further improvement, the sampling conditions are formulated, wherein the sampling conditions include: setting a minimum sampling number N, that is, the sampling area or the number of pixels is greater than the threshold N; the attribute vector α of the sample c1_smple The property of the species as a whole c1 The distance dist in the vector space is less than the preset threshold d, which controls the total distribution range of the sampling plots.

[0014] As a further improvement, the method of completing the calculation of the automatic sampling method and generating the flight path of the sampling UAV based on the sample data includes: completing the calculation of the automatic sampling method, obtaining the geographical area covered by the sample, setting detailed sampling parameters within the geographical area covered by the sample, and generating the flight path of the sampling UAV.

[0015] The second aspect of the present invention provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of an automatic biomass sampling method based on drone remote sensing calculation as described in the first aspect of the present invention.

[0016] The third aspect of the present invention provides an electronic device, which includes at least a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program on the memory, it implements the steps of the automatic biomass sampling method based on unmanned aerial vehicle remote sensing calculation described in the first aspect of the present invention.

[0017] The beneficial effects of the present invention are:

[0018] This method combines wide-area rough remote sensing and short-range high-precision remote sensing to achieve high-efficiency and high-precision sampling. The present invention can not only efficiently sample and evaluate vegetation biomass, but also ensure that the sampling method can reflect the real-time growth characteristics and distribution of vegetation, and achieve high-precision biomass estimation. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 It is a schematic diagram of an automatic biomass sampling method based on UAV remote sensing calculation provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0020] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Therefore, the following detailed description of the embodiments of the present invention provided in the drawings is not intended to limit the scope of the claimed invention, but merely represents the selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of the present invention.

[0021] In the description of the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0022] Reference Figure 1As shown, the first aspect disclosed by the present invention provides an automatic biomass sampling method based on UAV remote sensing measurement, including: based on the rough remote sensing data of the ground object, performing data processing on the rough remote sensing data, and automatically classifying the types of ground objects in the remote sensing data; calculating the attribute distribution α of each ground object; performing automatic sampling method calculation based on the attribute distribution α; generating a flight path of the sampling UAV according to the sample data; after the flight path is set, using the UAV to fly according to the set flight path and take pictures to collect data to complete automatic sampling.

[0023] The method of processing the rough remote sensing data based on the ground objects and automatically classifying the types of ground objects in the remote sensing data includes: preprocessing the multispectral data in the rough remote sensing data based on the rough remote sensing data of the ground objects, calculating the normalized vegetation index according to the rough remote sensing data and the results of the data preprocessing, and automatically classifying the types of ground objects in the remote sensing data according to the rough remote sensing data, the data preprocessing results and the normalized vegetation index.

[0024] Rough remote sensing data can be data taken by near-ground drones (ground resolution of 10 cm or above) or satellite remote sensing data. Take the data taken by near-ground multispectral drones as an example. Assume that the multispectral drone is equipped with 1 RGB and 5 multispectral cameras (blue B, green G, red R, red edge RE, near infrared NIR). Fly and shoot at an altitude with a ground resolution of 10 cm to obtain rough remote sensing data of the ground objects (including RGB and 5-band multispectral DN value data).

[0025] The rough remote sensing data based on the ground object, performing data preprocessing on the multispectral data in the rough remote sensing data, and calculating the normalized vegetation index according to the rough remote sensing data and the results of the data preprocessing, include:

[0026] The obtained RGB image and multispectral data are spatially aligned using an image alignment method; if the obtained RGB image and multispectral data are not spatially aligned, the above data are spatially aligned using an image alignment method commonly used in traditional 3D reconstruction algorithms.

[0027] Perform optical calibration on the camera to avoid optical problems such as distortion and chromatic aberration caused by the manufacturing process of the lens or sensor.

[0028] Perform radiometric calibration on multispectral data; the original value of multispectral data is usually called DN value, which has no physical meaning and cannot be directly compared. Because the shooting environment (incident light intensity, solar incidence angle, etc.) is different, and the camera itself will adjust the sensor gain and other parameters based on the environment to improve the signal-to-noise ratio of DN. Therefore, it is necessary to remove the above uncertainties caused by the environment. This process is called radiometric calibration. The value after radiometric calibration is called radiometric calibration value. Radiometric calibration value represents the relative reflectivity of the object to incident light in a certain band. The radiometric calibration values ​​of the same object obtained by the same camera at different time periods can be directly compared.

[0029] Reflectance calibration of multispectral data; In order to process multispectral data acquired by different cameras or different data acquisition systems (satellites, drones, etc.), the radiation calibration value can also be calibrated for reflectance normalization; After the reflectance calibration, the multispectral data will be a value between [0,1], which represents the absolute reflectivity of a certain object to incident light in a certain band.

[0030] The fusion of ground object images generates orthophotos of the entire shooting; generating orthophotos is an important part of surveying and remote sensing. It is usually generated by combining the original shooting data with the point cloud model or surface model generated by 3D reconstruction technology. After completing the above preprocessing, the aligned preprocessed images are input into this step for point cloud generation, surface model generation and orthophoto calculation. You can obtain RGB and multi-spectral band orthophotos at the same time. RGB , Map R , Map G , Map B , Map RE , Map NIR}.

[0031] After the above preprocessing, we can obtain the calibration values ​​of the original images of each multispectral band after preprocessing, and obtain the calibrated orthophotos covering the entire shooting range of each band fused together. Based on the radiation calibration value or reflectance calibration value, we can calculate one or more normalized vegetation indices through the following formulas:

[0032] NDVI = (NIR-R) / (NIR+R)

[0033] NDRE=(NIR-RE) / (NIR+RE)

[0034] GNDVI=(GR) / (G+R)

[0035] The calculation of the above vegetation index can be performed on the small image before fusion or on the full-range orthophoto after fusion.

[0036] Automatic classification of land objects can be achieved using RGB data and one or more of the above-mentioned reflectance calibration data and vegetation index data. Take the calculation using three spatially aligned data of RGB, NIR reflectance and vegetation index NDVI as an example.

[0037] The method for automatically classifying the types of land objects in remote sensing data based on rough remote sensing data, data preprocessing results and normalized vegetation index includes: defining the number of types of land objects; building a classification model; building a training data set including a true value database of all types of land objects that have been labeled, and using the training data set to train the classification model; inputting test data into the classification model for prediction processing, and the classification model outputting land object prediction results; and obtaining the number and area of ​​plots corresponding to each type of land object based on the land object prediction results.

[0038] Assume that according to the demand, it can be divided into C1 (fir), C2 (birch), C3 (grass), ..., C n-1 (Water area), C n (artificial objects) and other N types.

[0039] Build a model (classification model) that can identify these N types of land features. The classification model can be built based on traditional machine learning methods or deep learning methods.

[0040] Regardless of which method is used, you need to first build a training data set, that is, a true value database of the above N types of land features that have been labeled.

[0041] When the truth database contains enough data (for example, each category has tens of thousands of data), use these truth data to input into a deep learning network such as a U-Net network to train a classification model (M) that can classify objects into the above N species. It should be noted that RGB, NIR and NDVI data are used for recognition at the same time, so the truth data also needs to include the above three types of data that have been geographically aligned.

[0042] After the model M is built, the data that needs to be actually recognized (generally called test data) can be input into M for prediction, and M will output a prediction result for each pixel in the data. Assume that the test data input to the model M is I in (i,j){I RGB (i,j),I NIR (i,j),I NDVI (i, j)}, where i and j represent the coordinates of a pixel in the geo-aligned image. The corresponding prediction result can be expressed by the following formula

[0043] I out (i,j)=C k (k=1,2,…N)

[0044] Next, the number and area of ​​plots corresponding to each type of land feature Ci (i=1, 2, ..., N) can be obtained from the above prediction results.

[0045] To determine the number of plots, the first step is to detect the plots. There are many different ways to detect the plots. Assume that in I out The connected pixels belonging to the same type of ground object can be defined as the same ground block and detected if they exceed a pre-defined threshold. Before calculating the number of connected pixels, we can first out Do some post-processing such as filtering, hole filling or morphological processing to reduce the impact of noise. After detecting the plots, you can get the number of plots belonging to each type of land object and the area of ​​each plot. The plot area can be replaced by the number of pixels or calculated using the following formula:

[0046] Number of pixels × Ground Sampling Distance per pixel (GSD)

[0047] GSD is generally calculated by the height at the time of shooting, the sensor area of ​​the camera, and the focal length. The calculation formula is as follows:

[0048] GSD=(P×H) / F

[0049] Among them, GSD represents the ground sampling distance (in meters); P represents the physical size of the unit pixel (in meters / pixel); H represents the camera height (in meters); F represents the focal length of the camera (in millimeters)

[0050] Calculate the attribute distribution of each feature Ci to represent the features of the feature. Through the above steps, the plots and areas of all features and the pixels belonging to the features have been obtained. Assuming that the attribute distribution of C1 (fir tree) is to be calculated, the following calculation can be performed based on all the pixels belonging to C1.

[0051] The calculation of the attribute distribution α of each ground feature includes:

[0052] Calculate the overall attribute distribution of each type of ground feature Ci; for all pixels belonging to C1, use one or more combinations of vegetation indices calculated based on rough remote sensing data as attributes. Assuming that NDVI and NDRE are used as attributes, first obtain one or more statistical values ​​of the mean value E, variance S, standard deviation σ of NDVI and NDRE respectively. Then combine these statistical values ​​into the following vector

[0053] α c1=(E1, δ1, σ1, E2, δ2, σ2) where 1 represents NDVI and 2 represents NDRE

[0054] Calculate the attribute distribution of each independent plot of each type of land feature Ci. From the previous steps, all independent plots of C1 have been obtained. For each plot, the above attributes can also be calculated to obtain the attribute distribution α of the plot C1_t (t=1,2,…; represents different plots)

[0055] For a certain feature C1, we sample a sample for detailed detection. First, we need to set the sampling conditions: the sampling area must be larger than a pre-set threshold. This threshold can be input by the user through the UI.

[0056] The automatic sampling based on the attribute distribution α includes: formulating sampling conditions, and performing automatic sampling processing according to the attribute distribution α of each type of ground feature Ci.

[0057] The sampling conditions are formulated, wherein the sampling conditions include:

[0058] Set the minimum sampling number N, that is, the sampling area or number of pixels is greater than the threshold N;

[0059] The attribute vector α of the sample c1_smple The property of the species as a whole c1 The distance dist in the vector space is less than the pre-set threshold d; dist(α c1_smple ,α c1 ) <d

[0060] If there is a plot in C1 that meets the above two conditions, you can directly select the plot as a sample. If there are such independent plots, you can choose a plot with a smaller distance dist or a plot with a smaller number of samples.

[0061] If no independent plot meets the above two conditions, it is necessary to select sampling across plots. When sampling across plots, some additional conditions can be added on the basis of the above two conditions to restrict the sampling range:

[0062] Control the total distribution range of the sampling plots. The purpose of this condition is to reduce the flight distance and time during subsequent detailed sampling and improve efficiency.

[0063] Based on the histogram distribution of the overall attribute of C1 (assuming that the overall histogram is divided into 10 intervals, and the pixels belonging to each interval are marked on the orthophoto) and the minimum number of samples N, calculate the minimum number of samples Ni (i = 1, 2, ..., 10) required in each interval. Then you can start sampling the required number of samples from the first interval. After the first interval is sampled, start sampling from the nearest pixel belonging to the second interval, and then complete sampling in the 10th interval. The above sampling order can be customized by the user.

[0064] In addition, to improve efficiency, you can also sample only from the interval closest to the mean.

[0065] The method of completing the calculation of the automatic sampling method and generating the flight path of the sampling UAV according to the sample data includes: completing the calculation of the automatic sampling method, obtaining the geographical area covered by the sample, setting detailed sampling parameters in the geographical area covered by the sample, and generating the flight path of the sampling UAV.

[0066] After the automatic sampling method is calculated, it is necessary to generate the UAV flight route for detailed sampling on the above samples.

[0067] After the sampling method is calculated, the geographical area covered by the sample is obtained. In this geographical interval, the flight route is generated according to the detailed sampling parameters set by the user (GSD, overlap rate (intra-row, inter-row), flight direction, etc.). The generation method of the flight route is a relatively mature technology and will not be repeated here.

[0068] Here, different flight parameters are generally set according to the processing requirements after obtaining detailed sampling data. For example, if you need to obtain 3D information of plants in each plot through 3D reconstruction to obtain the number and height of plants, you often need to set the GSD to less than 1 cm. The lower the GSD value, the more detailed the data obtained, and the longer the operation time. Users can set appropriate parameters according to their needs.

[0069] After the above flight path is set, the drone can be used to shoot and collect detailed data according to the set path to complete the collection of detailed samples. The collected detailed data can still be 1 RGB and 5 multi-spectral data, or only RGB data can be used. Then use these data for three-dimensional reconstruction to obtain the 3D spatial information of the sample taken. Three-dimensional reconstruction can be achieved using existing commercial software or self-developed algorithms. After completing the 3D reconstruction, the number of plants in the sample of each land feature, the height of each plant, the breast diameter width and other indicators (biological attributes) can be obtained in the 3D information. Then, based on these attributes, the biomass (carbon fixation) of the land feature C1 sample can be calculated according to the currently commonly used biomass calculation method. C1_sampleThis calculation method is public and will not be elaborated on here.

[0070] Finally, the biomass of the above samples can be C1_sample Calculate the biomass of the entire C1 feature in the area C1 According to the above sampling rules, it can be ensured that the obtained samples are similar to the overall attribute distribution of the feature, so this calculation can be simply completed using the following formula

[0071] Bio C1 =Bio C1_sample *(Num C1 / Num C1_sample )

[0072] Among them, Num C1 is the area or number of pixels occupied by the C1 sample, Num C1_sample is the area or number of pixels occupied by C1 as a whole

[0073] The second aspect of the present invention provides a storage medium storing a computer program, which, when executed by a processor, implements the steps of an automatic biomass sampling method based on drone remote sensing calculation as described in the first aspect of the present invention.

[0074] The third aspect of the present invention provides an electronic device, which includes at least a memory and a processor, wherein the memory stores a computer program, and when the processor executes the computer program on the memory, it implements the steps of the automatic biomass sampling method based on unmanned aerial vehicle remote sensing calculation described in the first aspect of the present invention.

[0075] The above embodiments are only used to explain the technical solutions of the present invention rather than to limit them. Those skilled in the art should understand that any modification and equivalent substitution that does not depart from the spirit and scope of the present invention should fall within the protection scope of the claims of the present invention.

Claims

1. A biomass automatic sampling method based on drone remote sensing calculation, characterized in that: include: Based on the rough remote sensing data of the ground objects, the rough remote sensing data is processed and the types of ground objects in the rough remote sensing data are automatically classified; Calculate the attribute distribution α of each feature; Automatic sampling method calculation based on attribute distribution α; Generate a flight path of the sampled UAV based on the sample data; After the flight path is set, the drone is used to fly along the set flight path and take photos to collect data to complete automatic sampling; The calculation of the attribute distribution α of each type of land object includes: calculating the attribute distribution of each type of land object Ci as a whole; calculating the attribute distribution of each independent land parcel of each type of land object Ci; The automatic sampling method calculation based on the attribute distribution α includes: formulating sampling conditions, and performing automatic sampling processing according to the attribute distribution α of each type of ground feature Ci; The sampling conditions include: setting a minimum sampling number N, that is, the sampling area or the number of pixels is greater than the threshold N; the attribute vector α of the sample c1_smple The property α of the species as a whole c1 The distance dist in the vector space is less than the preset threshold d, which controls the total distribution range of the sampling plots.

2. The method for automatic biomass sampling based on drone remote sensing calculation according to claim 1 is characterized in that: The method of processing the rough remote sensing data based on the ground object and automatically classifying the ground object types in the rough remote sensing data includes: Based on the coarse remote sensing data of the ground objects, the multispectral data in the coarse remote sensing data are preprocessed, the normalized vegetation index is calculated according to the coarse remote sensing data and the results of the data preprocessing, and the types of ground objects in the coarse remote sensing data are automatically classified according to the coarse remote sensing data, the data preprocessing results and the normalized vegetation index.

3. The method for automatic biomass sampling based on drone remote sensing calculation according to claim 2 is characterized in that: The rough remote sensing data based on the ground object, performing data preprocessing on the multispectral data in the rough remote sensing data, and calculating the normalized vegetation index according to the rough remote sensing data and the results of the data preprocessing, include: Use image alignment method to perform spatial alignment processing on the obtained RGB image and multispectral data; Perform optical calibration on the camera; Perform radiometric calibration on multispectral data; Calibration and processing of ground reflectance of multispectral data; The ground object image fusion generates an orthophoto containing the entire shooting.

4. The method for automatic biomass sampling based on drone remote sensing calculation according to claim 2 is characterized in that: The method of automatically classifying the types of objects in the rough remote sensing data according to the rough remote sensing data, the data preprocessing results and the normalized difference vegetation index includes: Define the number of types of features; Build a classification model; Constructing a training data set including a true value database of all types of ground objects that have been annotated, and using the training data set to train the classification model; Inputting the test data into the classification model for prediction processing, and the classification model outputs the ground object prediction result; The number and area of ​​land parcels corresponding to each type of land feature are obtained according to the land feature prediction result.

5. The method for automatic biomass sampling based on drone remote sensing calculation according to claim 1 is characterized in that: Complete the automatic sampling method calculation and generate the flight path of the sampling drone based on the sample data, including: The geographic area covered by the sample is obtained, detailed sampling parameters are set within the geographic area covered by the sample, and a flight path of a sampling drone is generated.

6. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the automatic biomass sampling method based on drone remote sensing calculation as described in any one of claims 1 to 5 are implemented.

7. An electronic device, characterized in that: The method comprises at least a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program in the memory, the steps of the automatic biomass sampling method based on drone remote sensing calculation as described in any one of claims 1 to 5 are implemented.

Citation Information

Patent Citations

  • Method for calculating aboveground biomass based on UAV remote sensing images

    CN107833201A

  • Multi-source remote sensing data classification method for extracting classification sample points based on unmanned aerial vehicle

    CN111242224A

  • A multispectral image-based slope ground feature refined classification method

    CN113723291A

  • Invasive plant biomass estimation method based on unmanned aerial vehicle remote sensing

    CN115331124A

  • Vegetation dense area route planning method and system, storage medium and electronic equipment

    CN115344057A