Method for generating tobacco yield distribution map based on hyperspectral and leaf area index

Abstract

This application discloses a method for generating a tobacco yield distribution map based on hyperspectral and leaf area index, comprising the following steps: determining a tobacco plant leaf area index estimate and a tobacco yield estimation period based on yield estimation requirements; obtaining hyperspectral remote sensing images of the tobacco field to be estimated, hyperspectral data of sample points, tobacco plant LAI, and tobacco leaf yield; establishing a tobacco plant LAI estimation model using the sample point hyperspectral data and tobacco plant LAI; establishing a tobacco leaf yield estimation model using the sample point tobacco plant LAI and tobacco leaf yield; extracting hyperspectral remote sensing images of the tobacco plant planting area; inputting the extracted hyperspectral remote sensing images of the tobacco plant planting area into the established tobacco plant LAI estimation model to calculate a tobacco field LAI estimation map; and inputting the obtained tobacco field LAI estimation map into the established tobacco leaf yield estimation model to calculate a tobacco leaf yield distribution map. Compared to traditional point source-based survey methods, the tobacco leaf yield distribution map can not only provide more accurate tobacco leaf yield information, but also provide information on the spatial distribution of yield.

Description

Technical Field

[0001] The present application relates to the fields of crop yield estimation and hyperspectral remote sensing technology, and in particular to a method for generating a tobacco yield distribution map based on hyperspectral and leaf area index. Background Art

[0002] Flue-cured tobacco is a major cash crop, especially in low- and mid-latitude regions where rain and heat occur simultaneously. Due to the tobacco monopoly system, tobacco cultivation is subject to strict supervision and management. Tobacco leaf yield is a key indicator of tobacco agriculture.

[0003] Conventional tobacco leaf yield estimates are primarily based on regional surveys. Representative sample plots are selected, tobacco leaves harvested and cured within them, and the yield is calculated. This is then extrapolated to estimate the yield for a larger region. Conventional survey methods are not only inefficient and costly, but also yield data derived from limited point-source field surveys lacks accuracy.

[0004] Remote sensing technology, with its advantages of wide coverage, limited surface constraints, and periodic revisits, has been widely used in crop yield estimation. Remote sensing crop yield estimation primarily relies on satellite remote sensing and grain crop yield estimation. Existing yield estimation models primarily target specialty agricultural products such as soybeans and cotton. The development and application of remote sensing yield estimation models have played an irreplaceable role in food security and grain trade.

[0005] Existing optical remote sensing crop yield estimation primarily relies on multispectral remote sensing data, primarily due to its ease of acquisition and mature processing methods. Reflectance spectroscopy technology can obtain continuous reflectance spectra of ground objects. Compared to multispectral remote sensing technology, reflectance spectroscopy offers higher spectral resolution and a greater number of bands, enabling it to obtain richer spectral information about ground objects and providing stronger detection capabilities.

[0006] With the development of remote sensing technology, hyperspectral remote sensing integrates remote sensing technology and reflectance spectroscopy technology. It can obtain reflectance spectral data of ground objects while acquiring remote sensing images. It has the ability to conduct large-scale and precise detection of ground objects, effectively making up for the lack of multi-spectral remote sensing monitoring capabilities and the lack of representativeness of point source monitoring.

[0007] For example, Chen Hong, Xie Ling, and Chen Linlin. Monitoring the growth status of sorghum based on low-altitude UAV remote sensing [J]. Journal of Chinese Agricultural Mechanization, 2021, 42(04): 170-175. DOI: 10.13733 / j.jcam.issn.2095-5553.2021.04.24. used a multispectral camera carried by a UAV to obtain multispectral remote sensing images of sorghum at the jointing, heading and flowering, and filling and maturity stages, and constructed a regression model between four commonly used vegetation indices and the leaf area index (LAI) and vegetation coverage fraction (FVC) to accurately estimate the leaf area index and vegetation coverage of sorghum crops. However, existing models mainly focus on area estimation of narrow-leaf crops such as sorghum and wheat.

[0008] Tobacco is a cash crop primarily harvested for its leaves, cultivated over a large area. Yield estimation is crucial for evaluating tobacco varieties and cultivation techniques, and is crucial for practical production. Monitoring the leaf area index (LAI) is a key indicator of crop canopy structure. Hyperspectral remote sensing technology is used to invert the LAI, a parameter commonly used in remote sensing models to estimate yield and soil evaporation and transpiration. However, existing models for estimating tobacco leaf yield using hyperspectral data fail to accurately reflect yield across different areas of a tobacco field, resulting in unintuitive results. Summary of the Invention

[0009] To address the issues of insufficient multispectral remote sensing monitoring capabilities and underrepresentation of point-source monitoring, this application provides a method for generating tobacco yield distribution maps based on hyperspectral and leaf area index. This method not only obtains more accurate tobacco yield data but also captures the spatial distribution of tobacco yield. Combining the characteristics of remote sensing technology, multi-temporal and multi-scale hyperspectral remote sensing imagery can be used to rapidly generate a variety of tobacco yield distribution maps, providing more effective support for tobacco planting and harvesting.

[0010] The present application provides a method for generating a tobacco yield distribution map based on hyperspectral and leaf area index, comprising the following steps:

[0011] S1: Determine the estimated tobacco yield period and sample tobacco plants based on the estimated yield demand;

[0012] S2: Obtain hyperspectral remote sensing images of the tobacco fields to be estimated, and obtain hyperspectral data, leaf area index, and tobacco yield data of sample tobacco plants;

[0013] S3: Using the hyperspectral data and LAI of sampled tobacco plants, a tobacco plant LAI estimation model was established;

[0014] S4: Using the LAI and tobacco leaf yield data of the sample tobacco plants, a tobacco leaf yield prediction model was established using the curve fitting method;

[0015] S5: extracting a hyperspectral remote sensing image of the tobacco planting area from the hyperspectral remote sensing image of the tobacco field to be estimated;

[0016] S6: Input the extracted hyperspectral remote sensing image of the tobacco planting area into the tobacco plant LAI estimation model to estimate the tobacco plant LAI of the tobacco field to be estimated;

[0017] S7: Input the LAI estimation result of the tobacco field to be estimated obtained in step S6 into the tobacco yield estimation model to obtain a tobacco yield distribution map.

[0018] Preferably, step S2 includes the following steps:

[0019] Step S21: obtaining a hyperspectral remote sensing image of the entire tobacco-producing area to be estimated;

[0020] Step S22: Acquire hyperspectral data of the sample point: the hyperspectral data is a hyperspectral remote sensing image or a canopy spectral curve acquired by a portable ground object spectrometer;

[0021] If the hyperspectral data of the sample point is data from a portable ground feature spectrometer, the canopy spectral data measured by the ground feature spectrometer must be intercepted and resampled according to the parameter settings of the hyperspectral camera used in the tobacco field to be estimated, so that the spectral range, number of bands, wavelength, etc. of the ground feature spectrometer data are consistent with the relevant parameters of the hyperspectral camera;

[0022] Step S23: obtaining the LAI of the sample smoking plant: the LAI is obtained by measuring the agronomic traits of the sample smoking plant and then calculating, or directly measuring the LAI of the sample smoking plant using an LAI measuring instrument;

[0023] Step S24: according to the maturity progress of tobacco leaves, tobacco leaves of sample tobacco plants are picked and delivered to the tobacco flue-curing station for curing. The tobacco leaf yield data is the tobacco leaf yield obtained from the tobacco flue-curing station.

[0024] Preferably, step S3 includes the following steps:

[0025] Step S31: using regression analysis, machine learning or deep learning algorithms, and utilizing the hyperspectral data and LAI of sample tobacco plants, a tobacco plant LAI estimation model based on canopy reflectance spectra is constructed;

[0026] Step S32: When modeling, a band selection algorithm is used to select bands for the hyperspectral data, and key spectral bands are selected to construct the LAI estimation model.

[0027] Preferably, the step of “extracting a hyperspectral remote sensing image of a tobacco planting area” in S5 includes the following steps:

[0028] Step S51: performing radiation correction preprocessing and geometric correction preprocessing on the hyperspectral remote sensing image of the tobacco field to be estimated according to data processing requirements to obtain a reflectance remote sensing image with accurate geometric position;

[0029] Step S52: extracting tobacco planting areas from the tobacco field hyperspectral remote sensing image using a remote sensing image classification method, or constructing spectral parameters to extract tobacco planting areas;

[0030] Step S53: cropping the hyperspectral remote sensing image of the tobacco field to be estimated in accordance with the extracted tobacco planting area to obtain a hyperspectral remote sensing image of the tobacco planting area.

[0031] Preferably, step S6 also includes a step of processing the obtained tobacco plant LAI estimation map, specifically: according to the definition of LAI, processing the estimated tobacco plant LAI results, eliminating LAI values less than or equal to 0 in the LAI estimation map, and obtaining a processed tobacco plant LAI estimation map.

[0032] Preferably, the spectral range of the hyperspectral remote sensing image used is the visible near-infrared spectrum.

[0033] Preferably, step S7 further includes the following steps:

[0034] Step S71: performing yield statistics on the tobacco yield distribution map to obtain tobacco yield information;

[0035] Step S72: performing zoning statistics on the obtained tobacco yield distribution map according to water and fertilizer conditions.

[0036] Preferably, when calculating production statistics, the statistical parameter is at least one of the mean, median, sum, and variance.

[0037] Another aspect of the present application further provides a device for generating a tobacco yield distribution map based on a hyperspectral spectrum and a leaf area index, comprising:

[0038] The sample determination module is used to determine the estimated tobacco yield period and sample tobacco plants based on the estimated yield demand;

[0039] The data acquisition module is used to obtain hyperspectral remote sensing images of the tobacco fields to be estimated, and obtain hyperspectral data, leaf area index, and tobacco yield data of sample tobacco plants;

[0040] The first modeling module is used to establish a tobacco plant LAI estimation model using the hyperspectral data and LAI of the sample tobacco plants;

[0041] The second modeling module is used to establish a tobacco leaf yield estimation model using the LAI and tobacco leaf yield data of the sample tobacco plants and a curve fitting method;

[0042] An extraction module is used to extract a hyperspectral remote sensing image of the tobacco planting area from the hyperspectral remote sensing image of the tobacco field to be estimated;

[0043] The first estimation module is used to input the extracted hyperspectral remote sensing image of the tobacco planting area into the tobacco plant LAI estimation model to estimate the tobacco plant LAI of the tobacco field to be estimated;

[0044] The second estimation module is used to input the LAI estimation results of the tobacco fields to be estimated into the tobacco yield estimation model to estimate the tobacco yield distribution map.

[0045] The beneficial effects of this application include:

[0046] 1) The tobacco yield distribution map generation method based on hyperspectral and leaf area index provided in this application is aimed at the needs of tobacco yield estimation and combines the advantages of hyperspectral remote sensing images. A tobacco yield distribution map generation method based on hyperspectral and leaf area index is proposed. It can use multi-phase and multi-scale hyperspectral remote sensing images to quickly generate a variety of tobacco yield distribution maps, providing more accurate and richer tobacco yield information.

[0047] 2) The method for generating tobacco yield distribution maps based on hyperspectral and leaf area index provided in this application can graphically display the estimated tobacco field yield results. Data collection does not require manual collection, reducing labor intensity. At the same time, the graphical display of the estimation results can improve the estimation accuracy and facilitate the review of tobacco yields in different regions.

[0048] 3) The tobacco yield distribution map generation method based on hyperspectral and leaf area index provided in this application can not only obtain more accurate tobacco yields, but also obtain spatial distribution maps of yields compared to traditional point source-based survey methods. Combining the advantages of remote sensing technology, the tobacco yield distribution map generation method based on hyperspectral remote sensing imagery and leaf area index can obtain multi-phase and multi-scale tobacco yield distribution maps, providing more effective support for tobacco planting and harvesting. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic diagram of the process of generating a tobacco yield distribution map based on hyperspectral and leaf area index provided in this application;

[0050] Figure 2 This is the tobacco plant LAI map estimated by the present invention using UAV hyperspectral remote sensing images;

[0051] Figure 3 This is the tobacco yield distribution map estimated by the present invention using UAV hyperspectral remote sensing images. Figure 4 A schematic diagram of the structure provided for this application;

[0052] Figure 4Schematic diagram of the module of the tobacco yield distribution map generation device based on hyperspectral and leaf area index provided in this application. DETAILED DESCRIPTION

[0053] To make the objectives, technical solutions, and advantages of the embodiments of the present invention more clear, the technical solutions of the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings herein can be arranged and designed in various different configurations.

[0054] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention as claimed, but rather merely represents selected embodiments of the present invention. All other embodiments derived by persons of ordinary skill in the art based on the embodiments of the present invention without creative effort are also within the scope of protection of the present invention.

[0055] Technical features that are not used to solve the technical problems of this application are all set or installed according to common methods in the prior art and will not be repeated here.

[0056] See also Figure 1 The method for generating a tobacco yield distribution map based on hyperspectral and leaf area index provided in this application comprises the following steps:

[0057] S1: Determine the estimated tobacco yield period and sample tobacco plants based on the estimated yield demand;

[0058] S2: Obtain hyperspectral remote sensing images of the tobacco fields to be estimated, and obtain hyperspectral data, leaf area index, and tobacco yield data of sample tobacco plants;

[0059] S3: Using the hyperspectral data and LAI of sampled tobacco plants, a tobacco plant LAI estimation model was established;

[0060] S4: Using the LAI and tobacco leaf yield data of the sample tobacco plants, a tobacco leaf yield prediction model was established using the curve fitting method;

[0061] S5: extracting a hyperspectral remote sensing image of the tobacco planting area from the hyperspectral remote sensing image of the tobacco field to be estimated;

[0062] S6: Input the extracted hyperspectral remote sensing image of the tobacco planting area into the tobacco plant LAI estimation model to estimate the tobacco plant LAI of the tobacco field to be estimated;

[0063] S7: Input the LAI estimation result of the tobacco field to be estimated obtained in step S6 into the tobacco yield estimation model to obtain a tobacco yield distribution map.

[0064] Preferably, step S2 includes the following steps:

[0065] Step S21: obtaining a hyperspectral remote sensing image of the entire tobacco-producing area to be estimated;

[0066] Step S22: Acquire hyperspectral data of the sample point: the hyperspectral data is a hyperspectral remote sensing image or a canopy spectral curve acquired by a portable ground object spectrometer;

[0067] If the hyperspectral data of the sample point is data from a portable ground feature spectrometer, the canopy spectral data measured by the ground feature spectrometer must be intercepted and resampled according to the parameter settings of the hyperspectral camera used in the tobacco field to be estimated, so that the spectral range, number of bands, wavelength, etc. of the ground feature spectrometer data are consistent with the relevant parameters of the hyperspectral camera;

[0068] Step S23: obtaining the LAI of the sample smoking plant: the LAI is obtained by measuring the agronomic traits of the sample smoking plant and then calculating, or directly measuring the LAI of the sample smoking plant using an LAI measuring instrument;

[0069] Step S24: according to the maturity progress of tobacco leaves, tobacco leaves of sample tobacco plants are picked and delivered to the tobacco flue-curing station for curing. The tobacco leaf yield data is the tobacco leaf yield obtained from the tobacco flue-curing station.

[0070] Preferably, step S3 includes the following steps:

[0071] Step S31: using regression analysis, machine learning or deep learning algorithms, and utilizing the hyperspectral data and LAI of sample tobacco plants, a tobacco plant LAI estimation model based on canopy reflectance spectra is constructed;

[0072] Step S32: When modeling, a band selection algorithm is used to select bands for the hyperspectral data, and key spectral bands are selected to construct the LAI estimation model.

[0073] Preferably, the step of “extracting a hyperspectral remote sensing image of a tobacco planting area” in S5 includes the following steps:

[0074] Step S51: performing radiation correction preprocessing and geometric correction preprocessing on the hyperspectral remote sensing image of the tobacco field to be estimated according to data processing requirements to obtain a reflectance remote sensing image with accurate geometric position;

[0075] Step S52: extracting tobacco planting areas from the tobacco field hyperspectral remote sensing image using a remote sensing image classification method, or constructing spectral parameters to extract tobacco planting areas;

[0076] Step S53: cropping the hyperspectral remote sensing image of the tobacco field to be estimated in accordance with the extracted tobacco planting area to obtain a hyperspectral remote sensing image of the tobacco planting area.

[0077] Preferably, step S6 also includes a step of processing the obtained tobacco plant LAI estimation map, specifically: according to the definition of LAI, processing the estimated tobacco plant LAI results, eliminating LAI values less than or equal to 0 in the LAI estimation map, and obtaining a processed tobacco plant LAI estimation map.

[0078] Preferably, the spectral range of the hyperspectral remote sensing image used is the visible near-infrared spectrum.

[0079] In one embodiment, the weight of the tobacco leaves in the experimental sample is dry weight.

[0080] In a specific embodiment, the tobacco plants planted in the tobacco field to be evaluated are K326 tobacco plants.

[0081] Preferably, step S7 further includes the following steps:

[0082] Step S71: performing yield statistics on the tobacco yield distribution map to obtain tobacco yield information;

[0083] Step S72: performing zoning statistics on the obtained tobacco yield distribution map according to water and fertilizer conditions.

[0084] Preferably, when calculating production statistics, the statistical parameter is at least one of the mean, median, sum, and variance.

[0085] See also Figure 4 Another aspect of the present application further provides a device for generating a tobacco yield distribution map based on hyperspectral and leaf area index, comprising:

[0086] The sample determination module is used to determine the estimated tobacco yield period and sample tobacco plants based on the estimated yield demand;

[0087] The data acquisition module is used to obtain hyperspectral remote sensing images of the tobacco fields to be estimated, and obtain hyperspectral data, leaf area index, and tobacco yield data of sample tobacco plants;

[0088] The first modeling module is used to establish a tobacco plant LAI estimation model using the hyperspectral data and LAI of the sample tobacco plants;

[0089] The second modeling module is used to establish a tobacco leaf yield estimation model using the LAI and tobacco leaf yield data of the sample tobacco plants and a curve fitting method;

[0090] An extraction module is used to extract a hyperspectral remote sensing image of the tobacco planting area from the hyperspectral remote sensing image of the tobacco field to be estimated;

[0091] The first estimation module is used to input the extracted hyperspectral remote sensing image of the tobacco planting area into the tobacco plant LAI estimation model to estimate the tobacco plant LAI estimation map of the tobacco field to be estimated;

[0092] The second estimation module is used to input the LAI estimation results of the tobacco fields to be estimated into the tobacco yield estimation model to estimate the tobacco yield distribution map.

[0093] Example

[0094] The specific steps of the method for generating a tobacco yield distribution map based on hyperspectral remote sensing images and leaf area index of the present invention are as follows:

[0095] S1: Determine the estimated tobacco yield period and experimental samples based on the estimated yield demand;

[0096] The main steps include:

[0097] In this embodiment, tobacco leaf yield estimation is carried out in Chengjiang City, Yunnan Province, taking K326 tobacco plant as an example.

[0098] According to the estimated demand for tobacco leaf yield, the mature stage of the middle leaves of the tobacco plant is selected as the period for estimating tobacco leaf yield;

[0099] By understanding the water and fertilizer conditions of the tobacco fields to be estimated and combining them with the growth of tobacco leaves, the fields were divided into four levels and 12 plots according to the amount of fertilizer applied. Three representative tobacco plants were selected from each plot as experimental samples.

[0100] S2: Obtain hyperspectral remote sensing images of the tobacco fields to be evaluated, and obtain hyperspectral data, leaf area index (LAI), and tobacco yield of experimental samples;

[0101] The main steps include:

[0102] Use the hyperspectral camera carried by the UAV to obtain the hyperspectral remote sensing image of the entire tobacco-producing area to be assessed;

[0103] The canopy reflectance spectra of the experimental tobacco samples selected in S1 were measured using an ASD Field Spec 3 ground feature spectrometer;

[0104] Simultaneously with the canopy spectral measurements of the tobacco samples, agronomic traits of the tobacco plants were measured to obtain the number of leaves per plant and the length and width of each leaf. The LAI of the experimental sample tobacco plants was calculated according to the LAI calculation method used in tobacco agriculture.

[0105] According to the maturity of tobacco leaves, the tobacco plant samples determined in S1 are picked and sent to the tobacco station for drying, and the tobacco station provides the tobacco leaf yield. In this embodiment, the tobacco leaf yield obtained is the tobacco leaf yield per mu.

[0106] S3: Using the hyperspectral data and LAI data of the sample points, a tobacco plant LAI estimation model was established;

[0107] The main steps include:

[0108] The ASD Field Spec 3 spectrometer has a spectral range of 350-2500nm. During field spectral measurements, atmospheric water vapor has strong absorption in the water vapor absorption band. Incorporating the spectral characteristics of vegetation, the 350-1000nm band was intercepted from the 350-2500nm range for tobacco plant LAI modeling.

[0109] The spectra of the sample points here are obtained by a portable ground object spectrometer. The data obtained by the ground object spectrometer needs to be processed according to the parameters of the hyperspectral camera carried by the UAV;

[0110] The spectral range of the UAV hyperspectral data is 400-1000nm, with a total of 176 bands. Therefore, the ASD spectrometer is cropped and resampled to obtain a reflectance spectrum that is consistent with the spectral range, number of bands, and wavelength parameters of the UAV hyperspectral data.

[0111] When modeling, the genetic algorithm is first used to select the bands of the canopy reflectance spectrum;

[0112] Partial least squares regression (PLSR) was then used to establish a tobacco plant LAI estimation model based on canopy reflectance spectra.

[0113] S4: Using the LAI data of the sample points and the tobacco yield data, a tobacco yield estimation model was established;

[0114] Using the tobacco plant LAI and tobacco leaf yield data obtained in S2, a tobacco leaf yield prediction model based on LAI was established using the curve fitting method.

[0115] In this example, considering the model complexity and yield estimation accuracy, a linear function is selected to establish a tobacco yield estimation model based on LAI. The specific model is as follows:

[0116] Yield=41.32*LAI+82.81

[0117] Among them, Yield represents the yield per mu of tobacco field, in kilograms;

[0118] LAI represents the leaf area index of tobacco plants and is dimensionless.

[0119] S5: Extracting a hyperspectral remote sensing image of the tobacco planting area from the hyperspectral remote sensing image of the tobacco field to be estimated;

[0120] The main steps include:

[0121] Perform radiometric and geometric corrections on the acquired UAV hyperspectral remote sensing images to obtain reflectance remote sensing images with accurate geometric positions;

[0122] The Normalized Difference Vegetation Index (NDVI) is calculated using reflectance remote sensing images. The calculation formula is as follows:

[0123] NDVI=(R NIR -R RED ) / (R NIR +R RED )

[0124] Where NDVI is the normalized difference vegetation index;

[0125] R NIR is the reflectivity in the near-infrared band;

[0126] R RED is the reflectivity of the red light band;

[0127] According to the NDVI calculation results, a threshold of 0.4 was set, and areas with NDVI values greater than 0.4 were extracted as tobacco planting areas;

[0128] The hyperspectral remote sensing image is cropped according to the extracted tobacco planting area to obtain a hyperspectral remote sensing image of the tobacco planting area.

[0129] S6: The tobacco plant LAI estimation model is used to estimate tobacco plant LAI using the extracted hyperspectral remote sensing image of the tobacco planting area;

[0130] The main steps include:

[0131] The tobacco plant LAI estimation model based on canopy reflectance spectrum established in S3 was applied to the tobacco plant area hyperspectral remote sensing image extracted by UAV in S5 to obtain the tobacco plant LAI estimation map;

[0132] According to the definition of LAI, the LAI values less than or equal to 0 in the tobacco plant LAI estimation map are removed to obtain the final tobacco plant LAI estimation map, as shown in the figure below: Figure 2 shown.

[0133] S7: Applying the tobacco leaf yield estimation model to the tobacco plant LAI estimation results to generate a tobacco leaf yield estimation graph;

[0134] The main steps include:

[0135] The tobacco leaf yield prediction model based on LAI established in S4 is applied to the tobacco plant LAI map obtained in S6 to obtain the tobacco leaf yield distribution map, as shown in Figure 3 As shown;

[0136] According to the four levels defined in S1, the tobacco leaf yield distribution map obtained is statistically analyzed in each level of tobacco area, and the average value of each level of tobacco area is used as the unit yield of the tobacco area of that level, that is, the tobacco leaf yield per mu.

[0137] The tobacco leaf yield estimation diagram obtained in this embodiment is as follows Figure 3 shown.

[0138] Although the present invention has been described in detail with reference to the aforementioned embodiments, it is still possible for those skilled in the art to modify the technical solutions described in the aforementioned embodiments, or to make equivalent substitutions for some of the technical features therein. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for generating tobacco yield distribution map based on hyperspectral and leaf area index, characterized in that: The following steps are involved: S1: Determine the estimated tobacco yield period and sample tobacco plants based on the estimated yield demand; S2: Obtain hyperspectral remote sensing images of the tobacco fields to be estimated, and obtain hyperspectral data, leaf area index, and tobacco yield data of sample tobacco plants; S3: Using the hyperspectral data and LAI of sampled tobacco plants, a tobacco plant LAI estimation model was established; S4: Using the LAI and tobacco leaf yield data of the sample tobacco plants, a tobacco leaf yield prediction model was established using the curve fitting method; S5: extracting a hyperspectral remote sensing image of the tobacco planting area from the hyperspectral remote sensing image of the tobacco field to be estimated; S6: Input the extracted hyperspectral remote sensing image of the tobacco planting area into the tobacco plant LAI estimation model to estimate the tobacco plant LAI of the tobacco field to be estimated; S7: Inputting the LAI estimation result of the tobacco field to be estimated obtained in step S6 into the tobacco yield estimation model to estimate the tobacco yield distribution map; The tobacco leaf yield estimation model is: Taking into account the model complexity and yield estimation accuracy, a linear function is selected here to establish a tobacco yield estimation model based on LAI; the specific model is as follows: Yield=41.32*LAI+82.81 Among them, Yield represents the yield per mu of tobacco field, in kilograms; LAI represents the leaf area index of tobacco plant, dimensionless; The tobacco leaf yield data in step S2 is obtained by picking tobacco leaves from sample tobacco plants according to the maturity progress of the tobacco leaves and handing them over to the tobacco flue-curing station for curing. The tobacco leaf yield data is the tobacco leaf yield obtained from the tobacco flue-curing station.

2. The method for generating tobacco yield distribution map based on hyperspectral and leaf area index according to claim 1, characterized in that: Step S2 includes the following steps: Step S21: obtaining a hyperspectral remote sensing image of the entire tobacco-producing area to be estimated; Step S22: Acquire hyperspectral data of the sample point: the hyperspectral data is a hyperspectral remote sensing image or a canopy spectral curve acquired by a portable ground object spectrometer; The hyperspectral data of the sample points are data from a portable ground feature spectrometer. Based on the parameter settings of the hyperspectral camera used in the tobacco field to be estimated, the canopy spectral data measured by the ground feature spectrometer must be intercepted and resampled to ensure that the spectral range, number of bands, and wavelength of the ground feature spectrometer data are consistent with the relevant parameters of the hyperspectral camera. Step S23: Obtaining the LAI of the sample smoking plant: The LAI is obtained by measuring the agronomic traits of the sample smoking plant and then calculating, or by directly measuring the LAI of the sample smoking plant using an LAI measuring instrument.

3. The method for generating tobacco yield distribution map based on hyperspectral and leaf area index according to claim 1, characterized in that: Step S3 includes the following steps: Step S31: using regression analysis, machine learning or deep learning algorithms, and utilizing the hyperspectral data and LAI of sample tobacco plants, a tobacco plant LAI estimation model based on canopy reflectance spectra is constructed; Step S32: When modeling, a band selection algorithm is used to select bands for the hyperspectral data, and key spectral bands are selected to construct the LAI estimation model.

4. The method for generating tobacco yield distribution map based on hyperspectral and leaf area index according to claim 1, characterized in that: The "Extracting Hyperspectral Remote Sensing Images of Tobacco Planting Areas" step in S5 includes the following steps: Step S51: performing radiation correction preprocessing and geometric correction preprocessing on the hyperspectral remote sensing image of the tobacco field to be estimated according to data processing requirements to obtain a reflectance remote sensing image with accurate geometric position; Step S52: extracting tobacco planting areas from the tobacco field hyperspectral remote sensing image using a remote sensing image classification method, or constructing spectral parameters to extract tobacco planting areas; Step S53: cropping the hyperspectral remote sensing image of the tobacco field to be estimated in accordance with the extracted tobacco planting area to obtain a hyperspectral remote sensing image of the tobacco planting area.

5. The method for generating tobacco yield distribution map based on hyperspectral and leaf area index according to claim 1, characterized in that: Step S6 also includes a step of processing the obtained tobacco plant LAI estimation map, specifically: according to the definition of LAI, processing the estimated tobacco plant LAI results, eliminating LAI values less than or equal to 0 in the LAI estimation map, and obtaining a processed tobacco plant LAI estimation map.

6. The method for generating tobacco yield distribution map based on hyperspectral and leaf area index according to claim 1, characterized in that: The spectral range of the hyperspectral remote sensing images used is the visible and near-infrared spectrum.

7. The method for generating tobacco yield distribution map based on hyperspectral and leaf area index according to claim 1, characterized in that: Step S7 further includes the following steps: Step S71: performing yield statistics on the tobacco yield distribution map to obtain tobacco yield information; Step S72: performing zoning statistics on the obtained tobacco yield distribution map according to water and fertilizer conditions.

8. The method for generating tobacco yield distribution map based on hyperspectral and leaf area index according to claim 7, characterized in that: In production statistics, the statistical parameter is at least one of the mean, median, sum, and variance.

9. A device for generating a tobacco yield distribution map based on hyperspectral and leaf area index according to any one of claims 1 to 8, characterized in that: include: The sample determination module is used to determine the estimated tobacco yield period and sample tobacco plants based on the estimated yield demand; The data acquisition module is used to obtain hyperspectral remote sensing images of the tobacco fields to be estimated, and obtain hyperspectral data, leaf area index and tobacco yield data of sample tobacco plants; The first modeling module is used to establish a tobacco plant LAI estimation model using the hyperspectral data and LAI of the sample tobacco plants; The second modeling module is used to establish a tobacco leaf yield estimation model using the LAI and tobacco leaf yield data of the sample tobacco plants and a curve fitting method; An extraction module is used to extract a hyperspectral remote sensing image of the tobacco planting area from the hyperspectral remote sensing image of the tobacco field to be estimated; The first estimation module is used to input the extracted hyperspectral remote sensing image of the tobacco planting area into the tobacco plant LAI estimation model to estimate the tobacco plant LAI of the tobacco field to be estimated; A second estimation module is used to input the LAI of the tobacco plants of the tobacco field to be estimated obtained by the first estimation module into a tobacco leaf yield estimation model to estimate the tobacco leaf yield distribution map; The tobacco leaf yield estimation model is: Taking into account the model complexity and yield estimation accuracy, a linear function is selected here to establish a tobacco yield estimation model based on LAI; the specific model is as follows: Yield=41.32*LAI+82.81 Among them, Yield represents the yield per mu of tobacco field, in kilograms; LAI represents the leaf area index of tobacco plant, dimensionless; The tobacco leaf yield data in the data acquisition module is obtained by picking tobacco leaves from sample tobacco plants according to the maturity progress of the tobacco leaves and handing them over to the tobacco flue-curing station for curing. The tobacco leaf yield data is the tobacco leaf yield obtained from the tobacco flue-curing station.

Images