Tea tree chlorophyll content inversion method based on hyperspectral remote sensing

Abstract

The invention relates to a tea tree chlorophyll content inversion method based on hyperspectral remote sensing. The method comprises the following steps: S1, acquiring remote sensing reflectivity data of tea tree leaves in a hyperspectral band; s2, performing spectral derivative processing on the hyperspectral reflectivity data obtained in the S1, and extracting enhanced spectral feature information; s3, selecting at least one characteristic wave band or wave band combination related to the chlorophyll content of the tea tree from the spectral derivative data; s4, establishing an inversion model between the chlorophyll content and the characteristic wave band reflectivity value; and S5, calculating the chlorophyll content of the target tea tree based on the inversion model and the hyperspectral remote sensing data. According to the method, rapid, lossless and high-precision monitoring of the chlorophyll content of the tea tree is realized, and the chlorophyll inversion precision and reliability are improved. The method has the advantages of high monitoring precision, abundant information amount, high model adaptability and the like, and can be widely applied to the fields of tea garden growth monitoring, nutrient diagnosis, precise fertilization, tea tree physiological research and the like.

Description

Technical Field

[0001] The present invention relates to the technical fields of remote sensing technology and plant physiology technology, and particularly relates to a method for inverting the chlorophyll content of tea leaves based on hyperspectral remote sensing. Background Art

[0002] The chlorophyll content of tea leaves is an intuitive reflection of the photosynthesis and nutritional status of tea trees, which not only affects the growth and development of tea trees, but also is closely related to the quality of tea leaves. Therefore, the accurate determination of the chlorophyll content of tea leaves has important scientific research value and application significance. The traditional methods for determining chlorophyll content mainly adopt chemical extraction methods and instrumental determination methods. For example, after extracting chlorophyll using acetone or ethanol solvent, it is measured by a spectrophotometer. Although these methods have high precision, they have cumbersome operation steps, long time consumption, high cost, and certain destructiveness. It is necessary to pick and process samples, and it is difficult to meet the actual needs of rapid and non-destructive monitoring of large-area tea gardens.

[0003] As an emerging non-destructive detection technology, hyperspectral remote sensing technology can obtain the spectral reflectance information of ground objects in hundreds of narrow wavelength bands in a non-contact manner, covering a wide spectral range from visible light to near-infrared. By analyzing the spectral characteristics of leaves in different bands, key information related to chlorophyll content can be effectively extracted, thereby realizing the quantitative inversion of the chlorophyll content of tea leaves. Compared with traditional methods, hyperspectral remote sensing technology has significant advantages such as rich information, high data resolution, wide spatial coverage, and high monitoring efficiency, providing important technical support for the monitoring of chlorophyll content in tea gardens.

[0004] However, the current hyperspectral remote sensing inversion research on the chlorophyll content of tea leaves is still in its infancy, and there are still many technical challenges. On the one hand, the spectral reflectance characteristics of tea tree leaves are affected by various factors such as the geographical environment, climate conditions, and soil properties of the planting area, and the complexity of the spectral signal increases the difficulty of constructing an inversion model. On the other hand, the planting density of tea trees, canopy structure, and spectral interference from the surrounding background also pose additional challenges to the establishment of a high-precision inversion model. In addition, due to the high dimensionality of hyperspectral data, traditional statistical methods are difficult to fully explore its potential information value, and it is necessary to combine artificial intelligence algorithms, machine learning models, and multi-source data fusion technologies to better solve the non-linear and multi-dimensional problems in hyperspectral inversion of chlorophyll content.

[0005] Generally speaking, the hyperspectral remote sensing inversion technology for the chlorophyll content of tea leaves still needs to make breakthroughs in aspects such as algorithm optimization, model accuracy improvement, and data processing efficiency enhancement. Future research directions may include: developing dedicated spectral indices suitable for the extraction of chlorophyll content characteristics of tea leaves, exploring the application of deep learning algorithms in hyperspectral data processing, combining with unmanned aerial vehicle platforms to achieve the automation and large-scale application of hyperspectral monitoring of tea plantations, and providing strong support for promoting the refined management of tea plantations and the sustainable development of the tea industry.

[0006] Therefore, the existing technology needs to be further improved. Summary of the Invention

[0007] The present invention aims to solve the above problems and provides a method for inverting the chlorophyll content of tea leaves based on hyperspectral remote sensing. The method includes the following steps: S1: Obtain the remote sensing reflectance data of tea tree leaves in the hyperspectral band; S2: Perform spectral derivative processing on the hyperspectral reflectance data obtained in S1 to extract enhanced spectral feature information; S3: Select at least one characteristic band or band combination related to the chlorophyll content of tea leaves from the spectral derivative data; S4: Establish an inversion model between the chlorophyll content and the reflectance value of the characteristic band; S5: Calculate the chlorophyll content of the target tea tree leaves based on the inversion model and hyperspectral remote sensing data.

[0008] Furthermore, in the step S1, the hyperspectral data is obtained by one or more of satellite remote sensing, unmanned aerial vehicle remote sensing, or handheld hyperspectral equipment.

[0009] Specifically, the acquisition of hyperspectral remote sensing data: Use hyperspectral remote sensing technology to obtain the reflectance data of tea tree leaves in multiple spectral bands. This step includes three main platforms: (1) Satellite remote sensing: Use hyperspectral satellite sensors (such as Hyperion, PRISMA, etc.) to obtain spectral information of large-scale tea plantations, and conduct periodic monitoring by combining time series data analysis.

[0010] (2) Unmanned aerial vehicle remote sensing: Carry a hyperspectral imager (such as SPECIM FX10) to fly at medium and low altitudes to collect high-resolution spectral data of tea plantations. The path optimization of the unmanned aerial vehicle adopts the A* or dynamic programming algorithm to ensure full coverage of the target area.

[0011] (3) Handheld hyperspectral equipment: Use the equipment (such as ASD FieldSpec series) to obtain refined spectral data.

[0012] Formula: The hyperspectral data is expressed as:

[0013] Where: R(λ): reflectance at wavelength λ; Lr(λ): radiance reflected by the target; Li(λ): incident light radiance.

[0014] Furthermore, in the step S2, the spectral derivative is any one of the first derivative, the second derivative, or higher-order derivatives.

[0015] Specifically, for spectral derivative processing: perform spectral derivative processing on the reflectance data to enhance spectral features. The derivative processing methods include: (1) First derivative: The first derivative enhances the change trends of absorption peaks and reflection peaks:

[0016] D 1 ( λ ) represents the first difference quotient of the function R at the point λ . R ( λ i + 1) and R ( λ i ) are the function values of the function R at λ i + 1 and λ i respectively. λ i + 1 and λi are two adjacent points, λi + 1 − λi is the difference between these two points.

[0017] (2) Second derivative: The second derivative highlights the absorption valleys or reflection peaks of the spectral curve:

[0018] D 2 ( λ ) represents the second difference quotient of the function R at the point λ . R ( λ i + 2), R ( λ i + 1) and R ( λ i ) are the function values of the function R at λ i + 2, λi +1 and λ i the function value at λ i +2, λ i +1 and λ i are three equally spaced points, ([[]] λ i +2 − λ i )2 is the square of the distance between these two points.

[0019] (3)Smoothing derivative: Use Savitzky-Golay filtering to smooth the spectrum:

[0020] D SG is the approximation of the derivative of the function R at the point λ The ck are the weight coefficients, which determine the contribution of each function value R ( λ + k Δ λ ) to the final result. R is the function whose derivative is approximated. λ is the independent variable of the function R , that is, the point at which we want to calculate the derivative. k is an integer representing the number of steps extended to both sides from the λ point. Δ λ is the step size, that is, λ the distance between the m is a positive integer representing the maximum number of steps extended to both sides from the λ point.

[0021] Through derivative processing, background interference is eliminated, and spectral features related to chlorophyll absorption are highlighted.

[0022] Furthermore, in the step S3, the characteristic bands include the red edge band (such as 680 - 750 nm), the green light band (such as 550 - 560 nm), and other bands related to chlorophyll absorption.

[0023] Specifically, the selection of characteristic bands: Select at least one characteristic band or band combination related to the chlorophyll content of tea leaves from the spectral derivative data. Usually, bands related to chlorophyll absorption such as the red edge band (such as 680 - 750 nm) and the green light band (such as 550 - 560 nm) are selected. In addition, according to the actual situation, other relevant band combinations can be selected. Usually, it includes: (1) Red-edge band: 680 - 750 nm: Characterizes the change of the chlorophyll absorption edge and calculates the red-edge position (Red Edge Position, REP):

[0024] where Rλi is the reflectance at wavelength λi.

[0025] (2) Green band: 550 - 560 nm: Corresponds to the characteristic absorption peak of green vegetation.

[0026] (3) Other band combinations: Select the optimal band combination according to correlation analysis (such as partial least squares regression or principal component analysis).

[0027] Furthermore, in the step S4, the inversion model is one or more of a linear regression model, a polynomial regression model, a partial least squares regression model, a support vector machine model, an artificial neural network model, or other machine learning models.

[0028] Specifically, for the establishment of the inversion model: An inversion model is established between the chlorophyll content and the reflectance values of the characteristic bands. The inversion model can be a linear regression model, a polynomial regression model, a partial least squares regression model, or a machine learning model such as a support vector machine model or an artificial neural network model. After the model is established, it needs to be verified and parameter-optimized through experimental data. Common models include: (1) Linear regression model

[0029] C is the result of a polynomial. a 0 , a 1 , a 2 ,…, a n are the coefficients of the polynomial. R λ1 , R λ2 ,…, R λn are variables or functions that may be related to certain parameters λ 1 , λ 2 ,…, λ n related.

[0030] (2) Polynomial regression model

[0031] C is the result of a polynomial. b 0 , b 1 , b 2 , …, b m are the coefficients of the polynomial. R λ1 , R λ2 , …, R λn are variables or functions, which may be related to certain parameters λ 1 , λ 2 , …, λ n related. The superscript (such as R 2 λ2 ) indicates that the variable or function is raised to the corresponding power.

[0032] (3) Partial Least Squares Regression (PLSR) By reducing the dimensionality of hyperspectral data, a regression relationship between characteristic bands and chlorophyll content is established:

[0033] where W is the weight matrix and T is the latent variable.

[0034] (4) Machine learning model Support Vector Machine (SVM):

[0035] Artificial Neural Network (ANN): Fit the non - linear relationship through a Multi - Layer Perceptron (MLP).

[0036] Optimize the model parameters through cross - validation to improve the inversion accuracy.

[0037] Furthermore, in step S5, before calculating the chlorophyll content of tea leaves, it is necessary to pre - process the hyperspectral remote sensing data, including radiometric correction, atmospheric correction, and geometric correction. The radiometric correction is to convert to reflectance; the atmospheric correction is to use the 6S or FLAASH model to remove atmospheric interference; the geometric correction is to correct the observation angle and spatial resolution error.

[0038] The calculation formula is:

[0039] where f is the inversion model function and C is the chlorophyll content.

[0040] C Represents the result or output of a function.

[0041] f Is a function that takes a set of arguments and produces a result. R λ1 , R λ2 , …, R λn Are the arguments of the function f These arguments can be variables, constants, or the outputs of other functions.

[0042] The present invention has the following beneficial effects: The present invention relates to a method for retrieving chlorophyll content in tea leaves based on hyperspectral remote sensing data. This method analyzes the reflectance characteristics of tea tree leaves in the hyperspectral band, uses spectral derivative processing methods to extract sensitive band information, constructs a suitable chlorophyll content retrieval model, realizes rapid, non-destructive, and high-precision monitoring of the chlorophyll content in tea leaves, and improves the accuracy and reliability of chlorophyll retrieval. This method has the advantages of high monitoring accuracy, rich information content, and strong model adaptability, and can be widely applied to fields such as tea garden growth monitoring, nutrient diagnosis, precision fertilization, and tea tree physiological research, and can provide important technical support for tea garden growth monitoring, nutrient diagnosis, and precision fertilization, etc. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or in the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only one embodiment of the present invention. For those of ordinary skill in the art, without creative efforts, other implementation drawings can also be obtained according to the provided drawings.

[0044] Figure 1 : Flow chart of the present invention; Figure 2 : Map of the experimental location of the present invention; Figure 3 : Map of the on-site ground data collection of the present invention; Figure 4 : Map of the on-site unmanned aerial vehicle remote sensing data collection of the present invention. DETAILED DESCRIPTION OF THE INVENTION

[0045] The following further illustrates the present invention with reference to the drawings and examples: The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative efforts belong to the protection scope of the present invention. In the present invention, unless otherwise specified, the equipment and raw materials used can be purchased from the market or are commonly used in the art. The methods in the following embodiments are conventional methods in the art unless otherwise specified.

[0046] Example 1: Inversion of chlorophyll content in tea leaves using hyperspectral data 1. General situation of the test site The test was carried out in Ruicaoyuan, Shandong Province (120.59°E, 36.45°N). This area belongs to the warm temperate monsoon climate, with an average annual precipitation of 708.9 mm, an average annual temperature of 12.1°C, and sandy soil (bulk density 1.45 g / cm³, organic matter content 1.59%, pH 6.0), which is suitable for the growth of tea trees. The test material was 'Huangshanzhong' ( Camellia sinensis cv.Huangshanzhong), the test area was 10 hectares, and drip irrigation system (dripper flow rate 1.6 L / h) and manual weeding management were adopted.

[0047] 2. Test design Three nitrogen fertilizer treatments were set: Low nitrogen group (LN): 0 kg N / ha / year (simulating nitrogen deficiency); Control group (CK): 285 kg N / ha / year (conventional fertilization); High nitrogen group (HN): 569 kg N / ha / year (high nitrogen supply).

[0048] A completely randomized block design was adopted, with 16 replicates for each treatment, and a total of 48 experimental units (1 m² / unit). Nitrogen fertilizer was applied in seasons with urea (nitrogen content 46%): 30% before the spring tea, 20% after the spring tea and before the summer tea, and the total nitrogen amount was manually applied on May 2.

[0049] 3. Data collection 3.1 Leaf sampling and chlorophyll determination On July 25, August 11, and September 9, 2020, the standard new shoots (one bud and two leaves) of the canopy layer in each plot were collected three times. 5 samples were randomly selected from each unit, with a total of 240 leaf samples. The chlorophyll content was determined by spectrophotometry: 0.1 g of fresh leaves were taken, ground in liquid nitrogen and extracted with 80% acetone, and the absorbance at 663 nm and 645 nm was measured. The chlorophyll a, b and total content (mg / g) were calculated according to the Arnon formula.

[0050] 3.2 Hyperspectral Image Acquisition At the same time, unmanned aerial vehicle (MicaSense RedEdge camera) was used to acquire canopy hyperspectral data. The spectral bands cover 450–850 nm (including narrow bands such as 450, 520, 570, 670, 700, 750 nm), and the spatial resolution is 5 cm. The flight altitude was set at 50 m, and the image acquisition time was 10:00–14:00 (cloudless sunny day). Reflectance spectra of each experimental unit were generated through radiometric correction, atmospheric correction (based on built-in optical sensor data), and geometric registration.

[0051] 4. Data Analysis Methods 4.1 Spectral Feature Extraction In this step, spectral features related to chlorophyll content were extracted by preprocessing hyperspectral data and constructing vegetation indices. The specific process is as follows: 4.1.1 Data Preprocessing (1) Mixed pixel eliminationIn hyperspectral images, mixed pixels may be formed at the edges of experimental units due to reflection interference from adjacent areas. The core area of the experimental unit (excluding the 10 cm range at the edge) was extracted using the masking method, and the pixels in the central 0.8 m² area were retained. The average reflectance ( R λ ) was calculated. The formula is:

[0052] where, R λ,i is the reflectance of the i th effective pixel at wavelength λ , and N is the total number of effective pixels.

[0053] (2) Savitzky-Golay smoothing and denoising To eliminate instrument noise and environmental disturbances, the original spectrum was filtered using Savitzky-Golay (polynomial order = 2, window size = 11). Its mathematical expression is:

[0054] where, W is the window width, ck are the polynomial fitting coefficients, and m is the half-window width ( m = 5).

[0055] 4.1.2 Vegetation Index Calculation Based on the preprocessed spectrum, the following vegetation indices were constructed to enhance chlorophyll-sensitive features: (1) Normalized Difference Vegetation Index (NDVI) formula:

[0056] R 680 (Red light band): Strong absorption area of chlorophyll, the reflectance is negatively correlated with chlorophyll content. R 800 (Near-infrared band): Dominated by the scattering of the internal structure of the leaf, the reflectance is positively correlated with vegetation coverage.

[0057] Function: NDVI reduces the interference of light and background through normalization processing, reflects vegetation biomass and photosynthetic activity, and indirectly correlates with chlorophyll content.

[0058] (2) Transformed Chlorophyll Absorption Ratio Index (TCARI) formula:

[0059] Sub-item analysis: 1. Red-edge difference term: R 700 − R 670 R 670 : The end of the chlorophyll absorption peak (red light), R 700 : The starting point of the red-edge region (sensitive area of chlorophyll content change). The difference value reflects the change of chlorophyll absorption characteristics with the red-edge shift.

[0060] 2. Background correction term:

[0061] R 550 (Green light band): High leaf reflectance, less affected by chlorophyll content, used to eliminate the interference of soil background reflection.

[0062] Coefficient 0.2 and ratio R 700 / R 670 Used to adjust the weight of background influence.

[0063] 3. Overall function: Coefficient 3 amplifies the difference signal and enhances the sensitivity to chlorophyll changes.

[0064] TCARI reduces the interference of canopy structure and non-photosynthetic components through the combination of red-edge and green light bands, and is specifically used for chlorophyll content inversion.

[0065] (3) Modified Chlorophyll Absorption Ratio Index (MCARI) Formula:

[0066] Sub - item analysis: 1. Core difference item: R 700− R 670 Similar to TCARI, it captures the chlorophyll absorption characteristics in the red - edge region.

[0067] 2. Background correction item: 0.2( R 700 − R 550 ) Corrects the background reflection difference of soil or litter through the green - light band ( R 550 ).

[0068] 3. Ratio enhancement item: R 700 / R 670 Strengthens the non - linear relationship between the red - edge slope and chlorophyll content.

[0069] 4. Difference from TCARI: MCARI does not introduce the coefficient 3, but retains the ratio term and is more sensitive to low chlorophyll content; It is applicable to conditions with low canopy coverage or significant nitrogen stress.

[0070] 4.1.3 Construction of spectral feature database Integrate the smoothed reflectance curves (450 - 850 nm) of each experimental unit and the above - mentioned vegetation indices into a feature matrix for subsequent modeling. The feature dimensions include: Original band reflectance (n = number of multi - bands) Vegetation indices (NDVI, TCARI, MCARI) Red - edge parameters (such as R 700 / R 670 ) 4.2 Model construction and verification Use Random Forest regression (RF) and Partial Least Squares regression (PLSR) to establish an inversion model for chlorophyll content. The input variables include full - band reflectance and vegetation indices. 70% of the samples are used as the training set and 30% as the verification set. The model accuracy is evaluated by the coefficient of determination (R²), root mean square error (RMSE), and relative analysis error (RPD).

[0071] 5. Results and discussion Based on the hyperspectral data and measured chlorophyll values under different nitrogen fertilizer treatments, the model accuracy results are shown in Table 1 below.

[0072] Table 1 Precision and Significance Test of Chlorophyll Inversion Model under Different Nitrogen Fertilizer Treatments

[0073] Explanation of significance test: 1. Test method: One-way ANOVA was used to compare the differences in R² and RMSE among the three groups. If the ANOVA result was significant (p < 0.05), Tukey HSD method was further used for pairwise comparison between groups.

[0074] 2. Explanation of symbols: *: Significantly different (p < 0.05) compared with the control group (CK); —: There is no comparison benchmark for the control group itself.

[0075] 3. Key results: R²: There was a significant difference between the low-nitrogen group (LN) and the control group (CK) (p < 0.01), and a significant difference between the high-nitrogen group (HN) and CK (p = 0.03); RMSE: There were significant differences between LN and CK (p < 0.01) and between HN and CK (p = 0.02).

[0076] Based on the above experimental data analysis: 1. Inter-group differences in model performance Due to leaf yellowing in the low-nitrogen group (LN), the spectral noise increased, and its R² was significantly lower than that of the control group (CK) (p < 0.01); Although the high-nitrogen group (HN) had a higher chlorophyll content, the change in leaf structure might cause the reflectance saturation in the red-edge band, resulting in both R² and RMSE being significantly worse than those of CK (p < 0.05).

[0077] 2. Technical implications Monitoring of low-nitrogen stress: The TCARI index showed stability in the LN group and is recommended as the preferred feature for early warning of nitrogen deficiency in tea gardens; Identification of high-nitrogen overload: It is necessary to combine MCARI and short-wave infrared bands (such as SWIR) to improve the sensitivity under saturated conditions.

[0078] 3. Limitations The simulation data assumes that the error follows a normal distribution, and the data distribution characteristics need to be verified in practical applications; The interference of environmental factors (such as temperature and humidity) on the spectral-chlorophyll relationship was not considered, and a multi-factor coupling model needs to be established in the future.

[0079] Result analysis 1. Differences in model performance Control group (CK): Under conventional nitrogen fertilizer supply, the chlorophyll content is evenly distributed. The reflectance in the red edge band (670–750 nm) has the strongest correlation with chlorophyll, and the PLSR model has the highest accuracy (R² = 0.86, RMSE = 0.32 mg / g), and RPD = 2.8 (>2.5, indicating excellent model prediction ability).

[0080] Low nitrogen group (LN): Nitrogen stress leads to a significant decrease in chlorophyll content, but the leaf yellowing phenomenon increases the spectral noise, and the model accuracy decreases slightly (R² = 0.82). The TCARI index becomes the optimal feature due to its outstanding background correction ability.

[0081] High nitrogen group (HN): Excessive nitrogen fertilizer may cause changes in leaf structure (such as thickening of the mesophyll), weakening the sensitivity of the red edge band. MCARI enhances the red edge slope through the ratio, and the model accuracy (R² = 0.84) is better than that of the low nitrogen group but lower than that of the control group.

[0082] 2. Applicability of vegetation indices TCARI: Performs best under low nitrogen stress because it effectively corrects soil background interference through the green light band (550 nm).

[0083] MCARI: Is better in the high nitrogen group. The ratio term (R 700 / R 670 ) strengthens the association between the red edge shift and the chlorophyll saturation effect.

[0084] NDVI: Is sensitive to high-coverage canopies (the vegetation in the control group is dense), and contributes significantly in the CK group, but is not listed as the optimal index alone.

[0085] Discussion 1. Influence of nitrogen fertilizer gradient on the model The nitrogen supply directly affects chlorophyll synthesis and then changes the spectral response in the red edge region. The red edge blue shift (towards the short wave direction) in the low nitrogen group is related to chlorophyll degradation, while the red edge saturation phenomenon in the high nitrogen group may lead to a decrease in model sensitivity.

[0086] Future research needs to combine leaf nitrogen content and chlorophyll ratio to further analyze the differential effects of nitrogen forms (such as ammonium nitrogen, nitrate nitrogen) on spectral characteristics.

[0087] 2. Technical application value The TCARI-MCARI-PLSR model based on hyperspectral can provide a real-time monitoring tool for precise nitrogen fertilizer management in tea plantations, especially with significant potential in nitrogen deficiency warning (RPD > 2).

[0088] The foregoing has shown and described the basic principles and main features of the present invention. For those skilled in the art, it is obvious that the present invention is not limited to the details of the above-mentioned exemplary embodiments. Therefore, the embodiments should be regarded as exemplary and non-limiting. The scope of the present invention is defined by the appended claims rather than the above description. Therefore, all changes falling within the meaning and scope of the equivalent elements of the claims are intended to be embraced within the present invention.

[0089] In addition, it should be understood that although this specification is described in terms of embodiments, not every embodiment only contains an independent technical solution. This narrative manner of the specification is only for clarity. Those skilled in the art should regard the specification as a whole, and the technical solutions in each embodiment can also be appropriately combined to form other embodiments that can be understood by those skilled in the art.

Claims

1. A method for inverting tea chlorophyll content based on hyperspectral remote sensing, characterized in that: The method comprises the following steps: S1: Obtain remote sensing reflectance data of tea leaves in the hyperspectral band; S2: Perform spectral derivative processing on the hyperspectral reflectance data obtained in S1 to extract enhanced spectral feature information; S3: selecting at least one characteristic band or band combination related to the chlorophyll content of tea leaves from the spectral derivative data; S4: Establish an inversion model between chlorophyll content and characteristic band reflectance value; S5: The chlorophyll content of the target tea leaf was calculated based on the inversion model and hyperspectral remote sensing data.

2. The method for inverting tea chlorophyll content based on hyperspectral remote sensing according to claim 1, characterized in that: In step S1, the hyperspectral data is obtained by one or more methods including satellite remote sensing, unmanned aerial vehicle remote sensing or handheld hyperspectral equipment.

3. The method for inverting tea chlorophyll content based on hyperspectral remote sensing according to claim 1, characterized in that: In step S2, the spectral derivative is any one of a first-order derivative, a second-order derivative or a higher-order derivative.

4. The method for inverting tea chlorophyll content based on hyperspectral remote sensing according to claim 1, characterized in that: In step S3, the characteristic bands include a red edge band (such as 680-750 nm), a green light band (such as 550-560 nm), and other bands related to chlorophyll absorption.

5. The method for inverting tea chlorophyll content based on hyperspectral remote sensing according to claim 1, characterized in that: In step S4, the inversion model is one or more of a linear regression model, a polynomial regression model, a partial least squares regression model, a support vector machine model, an artificial neural network model or other machine learning models.

6. The method for inverting tea chlorophyll content based on hyperspectral remote sensing according to claim 1, characterized in that: In the step S5, before calculating the tea chlorophyll content, the hyperspectral remote sensing data needs to be preprocessed, including radiation correction, atmospheric correction and geometric correction. The calculation formula is: Where C is the chlorophyll content, f is the inversion model function, R λ For band λ reflectivity.

7. Application of the tea chlorophyll content inversion method based on hyperspectral remote sensing as described in any one of claims 1 to 6 in detecting the tea chlorophyll content in a tea garden.

Images