Ramie nitrogen content estimation method based on hyperspectral technology and electronic equipment

Abstract

According to the ramie nitrogen content estimation method based on the hyperspectral technology and the electronic equipment provided by the invention, the nitrogen content of ramie in different growth periods is estimated through two scales of leaves and canopies, and the multi-scale hyperspectral characteristic difference of ramie in different growth periods is analyzed by combining spectral data of different spatial resolutions; the method can determine the hyperspectral information of different spatial scales to estimate the difference of the ramie nitrogen content, and improves the estimation capability of the spatial distribution and dynamic change trend of the ramie nitrogen nutrition state. According to the method, characteristic wavelength screening is performed by adopting Pearson correlation analysis, effective dimension reduction can be performed on spectral data after derivative transformation, noise and redundant information are removed, and the model performance is improved. By removing the wavelengths which are not related to the target variable or are slightly related to the target variable, the model is more stable, computing resources can be saved, and time is saved.

Description

Technical Field

[0001] The invention belongs to the technical field of plant nitrogen content estimation, and in particular relates to a ramie leaf nitrogen content estimation method based on hyperspectral technology, electronic equipment and a medium. Background Art

[0002] Nitrogen is the most important nutrient element for ramie growth and yield quality formation. Rapid and accurate assessment of ramie nitrogen nutritional status is crucial for optimizing nitrogen fertilizer management and improving resource utilization efficiency. Traditional nitrogen diagnostic methods mainly rely on field sampling and laboratory measurements, which are time-consuming and labor-intensive, and difficult to achieve real-time and accurate monitoring.

[0003] In recent years, hyperspectral remote sensing technology has achieved a lot of results in the estimation of crop chlorophyll, leaf area index, nitrogen, phosphorus, potassium and other physiological and ecological parameters. This technology analyzes the differences in crop leaf spectra, builds a nitrogen nutrition estimation model for crop leaf spectra, and predicts the nitrogen content of crops. Currently, nitrogen nutrition diagnosis based on hyperspectral technology has been used in crops such as rice, wheat, corn and cotton, but there has been no report on the quantitative monitoring of nitrogen nutrition in ramie using hyperspectral technology.

[0004] Patent application CN118094082A discloses a method for estimating nitrogen content in macadamia leaves based on spectral sensitive variable optimization. By analyzing the correlation between nitrogen content in macadamia leaves and spectral data, a model for estimating nitrogen content in macadamia leaves is established based on nitrogen sensitive spectral variables. It only analyzes the spectral data of macadamia leaves during the first growth period, and the accuracy of predicting nitrogen content is low. Summary of the invention

[0005] The purpose of the present invention is to address the deficiencies of the existing technology and provide a method and electronic equipment for estimating the nitrogen content of ramie based on hyperspectral technology, so as to improve the monitoring capability of the spatial distribution and dynamic change trend of the nitrogen nutritional status of ramie.

[0006] In order to achieve the above object, the technical solution adopted by the present invention is:

[0007] A method for estimating nitrogen content in ramie based on hyperspectral technology comprises the following steps:

[0008] S1: The canopy spectral data and leaf spectral data of ramie at different growth stages were measured respectively, and the spatial resolution of the canopy spectral data and the leaf spectral data were different;

[0009] S2: Perform derivative transformation on canopy spectral data and leaf spectral data;

[0010] S3: Pearson correlation analysis was used to analyze the correlation between the canopy spectral data processed by S2 and the nitrogen content, and the canopy characteristic wavelength was selected; Pearson correlation analysis was used to analyze the correlation between the leaf spectral data processed by S2 and the nitrogen content, and the leaf characteristic wavelength was selected;

[0011] S4: using the leaf characteristic wavelength and the canopy characteristic wavelength as inputs of a random forest network model, training the random forest network model, and obtaining a nitrogen content estimation model, wherein the random forest network model comprises an input layer, a plurality of hidden layers connected in parallel, and an output layer, wherein the input layer is respectively connected to the plurality of hidden layers, and the plurality of hidden layers are respectively connected to the output layer;

[0012] The canopy characteristic wavelength and leaf characteristic wavelength of ramie at different growth stages are input into the nitrogen content estimation model to predict the nitrogen content of ramie.

[0013] The present invention estimates the nitrogen content of ramie at different growth stages at two scales, leaf and canopy, and analyzes the differences in multi-scale hyperspectral characteristics of ramie at different growth stages in combination with spectral data of different spatial resolutions. The differences in the nitrogen content of ramie estimated by hyperspectral information of different spatial scales can be clarified, thereby improving the ability to estimate the spatial distribution and dynamic change trend of the nitrogen nutritional status of ramie.

[0014] Furthermore, the different growth periods include at least two of the following growth periods: seedling stage, row closing stage, vigorous growth stage, and maturity stage.

[0015] Furthermore, in S3, based on the correlation between the canopy spectral data and the nitrogen content, the corresponding wavelength whose absolute value of the Pearson correlation coefficient is greater than the first threshold and whose p value is less than the second threshold is selected as the canopy characteristic wavelength; based on the correlation between the leaf spectral data and the nitrogen content, the corresponding wavelength whose absolute value of the Pearson correlation coefficient is greater than the first threshold and whose p value is less than the second threshold is selected as the leaf characteristic wavelength.

[0016] The two-tailed test is a statistical hypothesis testing method. The p-value is the probability value of the test statistic, which describes the probability of the observed statistical effect (i.e., the Pearson correlation coefficient) or a more extreme situation when the null hypothesis is true (i.e., there is no linear relationship between the two variables). The p-value represents whether the relationship between the two variables is significant. If the p-value is less than the significance level (the second threshold), the linear relationship between the two variables is considered to be significant. If the p-value is greater than the significance level, the linear relationship between the two variables is considered to be insignificant.

[0017] Using Pearson correlation analysis to screen characteristic wavelengths can effectively reduce the dimension of the spectral data after derivative transformation, remove noise and redundant information, and improve model performance. By removing wavelengths that are irrelevant or weakly correlated with the target variable, the model is more stable and can save computing resources and time.

[0018] Based on the same inventive concept, the present invention also provides an electronic device, including:

[0019] one or more processors;

[0020] A memory having one or more programs stored thereon, which, when executed by the one or more processors, enables the one or more processors to implement the steps of a method for estimating nitrogen content in ramie based on hyperspectral technology.

[0021] Based on the same inventive concept, the present invention also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of a method for estimating nitrogen content in ramie based on hyperspectral technology.

[0022] Compared with the prior art, the present invention has the following beneficial effects:

[0023] The present invention estimates the nitrogen content of ramie at different growth stages at two scales, leaf and canopy, and analyzes the differences in multi-scale hyperspectral characteristics of ramie at different growth stages in combination with spectral data of different spatial resolutions. The differences in the nitrogen content of ramie estimated by hyperspectral information of different spatial scales can be clarified, thereby improving the ability to estimate the spatial distribution and dynamic change trend of the nitrogen nutritional status of ramie. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 This is a flow chart of the method for estimating nitrogen content in ramie based on hyperspectral technology;

[0025] Figure 2 is a schematic diagram of the Hand Held 2 spectrometer;

[0026] Figure 3 This is a schematic diagram of the FieldSpec 3 ground object spectrometer;

[0027] Figure 4 is a schematic diagram of the AA3 flow analyzer;

[0028] Figure 5 is the correlation diagram between canopy spectral data and nitrogen content;

[0029] Figure 6 This is the correlation diagram between leaf spectral data and nitrogen content;

[0030] Figure 7 Schematic diagram of the structure of the LNC estimation model. DETAILED DESCRIPTION

[0031] The present invention will be described in detail below in conjunction with the embodiments. It should be noted that the embodiments and features of the embodiments of the present invention can be combined with each other without conflict. For the convenience of description, if the words "upper", "lower", "left" and "right" appear in the following, they only indicate that the upper, lower, left and right directions are consistent with the drawings themselves, and do not limit the structure.

[0032] Example

[0033] like Figure 1 The present embodiment of a method for estimating nitrogen content in ramie based on hyperspectral technology was carried out at the ramie experimental base of Hunan Agricultural University, which has a subtropical monsoon humid climate with four distinct seasons. The annual average temperature is between 13℃ and 20℃, and the annual average rainfall is 1368.6-1442.4mm. The basic physical and chemical properties of the soil are shown in Table 1. The experiment selected Zhongzhu No. 2, and used cutting seedlings for field micro-plot planting. Each micro-plot area is 4.84m 2 , 16 ramie plants were planted. The experimental design included 3 replications, using a completely random arrangement, with a total of 21 microplots.

[0034] The experiment was conducted from April to November 2022, with the first hemp harvested on August 31 and the second hemp harvested on November 25. The experiment adopted a two-factor design. The main factor was the nitrogen fertilizer supply level, which was set at four levels (N0 to N3), and the secondary factor was the timing of topdressing, which was divided into the closed row period and the vigorous growth period. The treatment with conventional fertilizer application during the closed row period was used as the control group (CK), with a total of seven treatment combinations. 60% of the nitrogen fertilizer was applied as basal fertilizer, and the remaining 40% was used as topdressing. Phosphate fertilizer (P2O5 140kg / hm 2 ) and potash fertilizer (K2O300kg / hm 2 ) is applied as a base fertilizer at one time. The fertilizers used include urea (nitrogen content 46.2%), superphosphate (P2O5 content 12%) and potassium chloride (K2O content 60%).

[0035] Table 1 Basic physical and chemical properties of soil

[0036]

[0037] 1. The spectral data of ramie canopy were collected by Hand Held 2 handheld spectrometer (ASD, USA). The canopy spectral reflectance data were collected at four key growth stages: seedling stage, row closing stage, vigorous growth stage, and maturity stage. Three points were randomly selected in each plot to collect canopy spectral data, and the average value was used as the canopy spectral measurement value of each plot. During the measurement, the handheld spectrometer was 60 cm above the ramie canopy, and each plot was calibrated with a standard white plate before measurement. Figure 2 , the main parameters are shown in Table 2.

[0038] Table 2 Main parameters of Hand Held 2 spectrometer

[0039]

[0040] Second, the FieldSpec 3 spectrometer and the matching handheld leaf clip spectrometer produced by ASD Company of the United States were used to measure the spectrum of ramie leaves in each growth period. The instrument has a built-in quartz halogen lamp and a stable light source. Whiteboard calibration was performed before the leaf spectrum was measured. Four pockets were randomly selected in each plot (the pockets were selected and then determined), and two plants were selected in each pocket to measure the spectrum data of the top three leaves in turn. There were a total of 24 spectral data in each plot, and the average was finally taken as the final spectrum of the leaves in the plot. FieldSpec 3 spectrometer is as follows Figure 3 , the main parameters are shown in Table 3.

[0041] Table 3 Main parameters of FieldSpec 3 spectrometer

[0042]

[0043] 3. Put the top 3 leaves and canopy of ramie from each growth period sample point into marked envelopes, bring them back to the laboratory, put them in an oven at 104℃ for 30 minutes, and then maintain at 70-80℃ until dried. After drying, weigh them, and then grind the leaf samples, grind them into pieces and digest them until the solution becomes clear and transparent. After the digestion solution has cooled, use the AA3 continuous flow analyzer to determine the total nitrogen content in the digestion solution, and then calculate the total nitrogen content of the ramie leaves. AA3 continuous flow analyzer Figure 4 .

[0044] 4. Preprocessing of canopy spectrum and leaf spectrum data. View Spec Pro 6.2 software was used to output the original and first-order differential data of the collected ramie hyperspectral data. Spectral derivative, also known as spectral differential, is a method of performing derivative operations of different orders on spectral curves. Through derivative operations of different orders, the influence of background factors such as the earth's surface atmosphere, soil type, soil moisture, etc. on the monitoring of crop growth parameters can be eliminated to varying degrees, revealing the intrinsic characteristics of the vegetation spectrum, thereby maximizing the target spectral signal.

[0045] The first-order derivative can eliminate linear background noise in the spectral variation region. The calculation formula is as follows:

[0046]

[0047] Where i is the band number; D = {D1, D2, ..., D i-1 ,D i , D i+1 ,…,D N} is the spectral vector of leaves and canopy at different growth stages, N is the number of bands; λ={λ1,λ2,…,λ i-1 ,λ i ,λ i+1 ,…,λ N} is a wavelength set.

[0048] 5. Spectral parameter selection. The red edge band (680-760nm) is between the red light and near-infrared bands, and contains chlorophyll absorption and canopy scattering information, forming a hypotenuse that approaches a straight line. When performing analysis, the red edge position, red edge amplitude, and red edge area are generally used to describe the "red edge" parameters. These parameters can be used to monitor crop nitrogen content (LNC), chlorophyll status, LAI (leaf area index) and other parameters. Similar to the red edge parameters are the blue edge parameters and the yellow edge parameters, collectively referred to as the three edge parameters. This embodiment selects 9 spectral parameters (see Table 4) based on existing research (Cao Xiaolan. Research on hyperspectral characteristics of ramie leaves and establishment and application of qualitative and quantitative models [D]. Hunan Agricultural University, 2018.) to construct a ramie LNC estimation model.

[0049] The correlation between ramie LNC and the selected 9 spectral parameters in each growth period was statistically analyzed, and the results are shown in Tables 5 and 6. In the canopy spectrum, the correlation between SDr, SDy, SDb, NDVI, DVI, and RNI and LNC were extremely significant throughout the growth period; the correlation of Dr in the row closing period and the vigorous growth period was extremely significant; the correlation of Dy in the row closing period and the vigorous growth period reached an extremely significant level; and Db was most significant in the mature period. In the leaf spectrum, the correlation between Dr, SDr, SDy, Db, and RNI and LNC were extremely significant throughout the growth period; Dy reached an extremely significant level in the row closing period; SDb reached an extremely significant level in the seedling stage, the vigorous growth period, and the mature period; NDVI and DVI both reached an extremely significant level in the row closing period, the vigorous growth period, and the mature period. It shows that it is feasible to estimate ramie LNC using ramie canopy spectral parameters.

[0050] Table 4 Spectral parameters and their calculation formulas

[0051]

[0052] Table 5 Correlation analysis between spectral parameters of ramie canopy and LNC at different growth stages

[0053]

[0054] Note: * indicates extremely significant correlation at the 5% level, ** indicates extremely significant correlation at the 1% level.

[0055] Table 6 Correlation analysis between spectral parameters and LNC of ramie leaves at different growth stages

[0056]

[0057] Note: * indicates extremely significant correlation at the 5% level, ** indicates extremely significant correlation at the 1% level.

[0058] 6. Feature band selection. Pearson correlation analysis was used to explore the correlation between nitrogen content and the first-order spectrum of the canopy ( Figure 5 ), the correlation between nitrogen content and the first-order spectrum of leaves ( Figure 6 ).according to Figure 5 , Figure 6 The Pearson correlation test was used for the relevant bands in each period, and the extremely significant relevant wavelengths in the correlation coefficient curve band were selected to construct the ramie LNC estimation model.

[0059] The Pearson correlation coefficient formula is as follows:

[0060]

[0061] Among them, r is the Pearson correlation coefficient, X i and Y i They are spectral data and nitrogen content LNC, and are the means of spectral data and nitrogen content LNC, respectively.

[0062] The specific steps are as follows: 1. Prepare two sets of data, one is spectral data, which contains the reflectance of all wavelength points at different growth stages; the other is nitrogen content LNC. 2. Import these two sets of data into SPSS Statistics27 statistical software for correlation analysis. Select the "Correlation" option under the "Analysis" menu, and then select "Bivariate". Add all wavelength points of the spectral data as a set of variables to the "Variable" box, and add the LNC data as another variable. 3. In the "Bivariate Correlation" dialog box, select "Pearson" as the correlation coefficient. For the significance test, select the "Two-tailed" option and check "Mark significant correlation". 4. Run the analysis. SPSS automatically calculates the Pearson correlation coefficient between each wavelength point and LNC, and displays the correlation coefficient matrix. 5. In the matrix, the wavelength points with an absolute value of the correlation coefficient r greater than 0.5 and a two-tailed p value less than the significance level of 0.05 are significantly correlated with LNC.

[0063] The two-tailed test is a statistical hypothesis testing method. The p-value is the probability value of the test statistic, which describes the probability of the observed statistical effect (i.e., the Pearson correlation coefficient) or a more extreme situation when the null hypothesis is true (i.e., there is no linear relationship between the two variables). The p-value represents whether the relationship between the two variables is significant. If the p-value is less than the significance level of 0.05, the linear relationship between the two variables is considered to be significant. If the p-value is greater than the significance level, the linear relationship between the two variables is considered to be insignificant.

[0064] At the canopy scale, the sensitive spectral bands of ramie LNC estimation at the seedling stage are 414nm, 459nm, 550nm, at the row closing stage are 913nm, 996nm, 1029nm, 1070nm, at the vigorous growth stage are 329nm, 998nm, 1021nm, 1044nm, and at the mature stage are 413nm, 456nm, 866nm, 908nm. At the leaf scale, the sensitive spectral bands of ramie LNC estimation at the seedling stage are 2061nm, 2073nm, 2077nm, at the row closing stage are 488nm, 981nm, 2084nm, at the vigorous growth stage are 354nm, 399nm, 427nm, 1012nm, and at the mature stage are 1449nm, 1683nm, 2292nm, 2313nm.

[0065] Canopy spectrum ( Figure 5 ) and leaf spectra ( Figure 6 ) have similar response characteristics to ramie LNC, and the spectral reflectance curve has obvious peak-valley characteristics, showing a "green peak" near 460nm and a "red valley" near 684nm. The reflectance rises sharply after 690nm, and maintains a stable high-reflectance platform after entering the near-infrared band, and there are two water absorption valleys near 1440nm and 1930nm. There are obvious differences in the canopy reflectance spectra of ramie at different growth stages, while the differences in the leaf reflectance spectra of ramie at different growth stages are small.

[0066] 7. Constructing a spectral model. In order to accurately estimate the nitrogen content of ramie leaves, the experiment selected the key characteristic wavelengths of each growth period as the model input parameters, and then constructed an LNC estimation model based on the first-order spectral characteristics ( Figure 7). Among them, in the construction of the ramie LNC model based on the canopy spectral band, the wavelengths screened in the seedling stage are 414nm, 459nm, and 550nm, the wavelengths screened in the row closing stage are 913nm, 996nm, 1029nm, and 1070nm, the wavelengths screened in the vigorous growth stage are 329nm, 998nm, 1021nm, and 1044nm, and the wavelengths screened in the maturity stage are 413nm, 456nm, 866nm, and 908nm. In the construction of the ramie LNC model based on the leaf spectral band, the wavelengths screened in the seedling stage are 2061nm, 2073nm, and 2077nm, the wavelengths screened in the row closing stage are 488nm, 981nm, and 2084nm, the wavelengths screened in the vigorous growth stage are 354nm, 399nm, 427nm, and 1012nm, and the wavelengths screened in the maturity stage are 1449nm, 1683nm, 2292nm, and 2313nm. The basic structure of the model is a multi-layer structure including an input layer (characteristic wavelength), a hidden layer (multiple decision trees) composed of a random forest algorithm, and an output layer (leaf nitrogen content prediction value). The model was built with the help of the Matlab 2018a software environment, and the core algorithm used random forests, aiming to improve the accuracy and stability of the prediction of ramie leaf nitrogen content through an integrated learning strategy.

[0067] 8. Model Accuracy Evaluation. The experiment adopted a stratified sampling method. For each key growth period, 80% of the data were randomly selected from the ramie LNC data set as the training set, and the remaining 20% ​​were used as the validation set. When evaluating the performance of the model, the coefficient of determination (R 2 ) and root mean square error (RMSE). 2 The higher the value, the better the model fits the data, that is, the higher the prediction accuracy. The lower the RMSE value, the smaller the deviation between the model's prediction results and the actual observed values, reflecting that the model is more stable. By analyzing the model accuracy results in Tables 7 and 8, the following conclusions can be drawn: The ramie LNC model constructed based on the random forest (RF) algorithm in machine learning performs well overall, especially on the training set. The R 2 The value is more than 0.8, indicating that the model has a strong fitting ability on the training data. Among the canopy spectral band models, the model in the vigorous growth period performs best. 2 The model of leaf spectral bands is the best in the seedling stage. 2 It is 0.935.

[0068] Coefficient of determination (R 2 ) and the root mean square error (RMSE) are calculated as follows:

[0069]

[0070] Among them, y i is the measured value, is the predicted value. is the mean value and n is the number of samples.

[0071] Table 7 Accuracy evaluation of ramie LNC model based on canopy spectral bands

[0072]

[0073] Table 8 Accuracy evaluation of ramie LNC model based on leaf spectral bands

[0074]

[0075] This example uses hyperspectral technology to estimate the nitrogen content of ramie, achieving a rapid and non-destructive assessment of the nitrogen nutritional status of ramie. Compared with traditional nitrogen prediction methods, this technical solution significantly improves detection efficiency, reduces manpower and time costs, and ensures high accuracy. The long-term RF model of ramie vigor constructed based on canopy spectral data has high accuracy, R 2 The RF model constructed based on leaf spectral data at the ramie seedling stage achieved the highest accuracy, R 2 The value of the α-value was 0.935, which further confirmed the reliability and effectiveness of the technology in estimating nitrogen content. By continuously monitoring ramie at different growth stages, the evolution of its nitrogen nutrition status over time was analyzed, and dynamic monitoring in time and space was achieved.

[0076] This embodiment estimates the nitrogen content of ramie leaves from two scales, leaf and canopy, and can clarify the difference in estimating the nitrogen content of ramie leaves using hyperspectral information at different spatial scales.

[0077] This embodiment combines hyperspectral data of different spatial resolutions (handheld spectrometer and ground object spectrometer) to analyze the differences in multi-scale hyperspectral characteristics of ramie at different growth stages, so as to improve the monitoring capability of the spatial distribution and dynamic change trend of the nitrogen nutritional status of ramie.

[0078] Another embodiment of the present invention provides an electronic device, including:

[0079] one or more processors;

[0080] A memory having one or more programs stored thereon, which, when executed by one or more processors, enables the one or more processors to implement the steps of a method for estimating nitrogen content in ramie based on hyperspectral technology.

[0081] In some implementations, the memory may be a high-speed random access memory (RAM), and may also include a non-volatile memory, such as at least one disk memory.

[0082] In some other implementations, the processor may be a central processing unit (CPU), a digital signal processor (DSP), or other general-purpose processors of various types, which are not limited herein.

[0083] Another embodiment of the present invention provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of a method for estimating nitrogen content in ramie based on hyperspectral technology.

[0084] The contents explained in the above embodiments should be understood as these embodiments are only used to more clearly illustrate the present invention, and are not used to limit the scope of the present invention. After reading the present invention, various equivalent forms of modifications to the present invention by those skilled in the art all fall within the scope defined by the claims attached to this application.

Claims

1. A method for estimating nitrogen content in ramie based on hyperspectral technology, characterized in that: The following steps are involved: S1: The canopy spectral data and leaf spectral data of ramie at different growth stages were measured respectively, and the spatial resolution of the canopy spectral data and the leaf spectral data were different; S2: Perform derivative transformation on canopy spectral data and leaf spectral data; S3: Pearson correlation analysis was used to analyze the correlation between the canopy spectral data processed by S2 and the nitrogen content, and the canopy characteristic wavelength was selected; Pearson correlation was used to analyze the correlation between the leaf spectral data treated with S2 and the nitrogen content, and the characteristic wavelength of the leaves was selected; S4: Using the leaf characteristic wavelength and the canopy characteristic wavelength as inputs of a random forest network model, training the random forest network model, and obtaining a nitrogen content estimation model, wherein the random forest network model comprises an input layer, multiple parallel hidden layers, and an output layer, wherein the input layer is respectively connected to the multiple hidden layers, and the multiple hidden layers are respectively connected to the output layer.

2. The method for estimating nitrogen content in ramie based on hyperspectral technology according to claim 1, characterized in that: The different growth periods include at least two of the following growth periods: seedling stage, row closing stage, vigorous growth stage, and maturity stage.

3. The method for estimating nitrogen content in ramie based on hyperspectral technology according to any one of claims 1 or 2, characterized in that: The canopy characteristic wavelength and leaf characteristic wavelength of ramie at different growth stages are input into the nitrogen content estimation model to predict the nitrogen content of ramie.

4. The method for estimating nitrogen content in ramie based on hyperspectral technology according to claim 1, characterized in that: In S3, according to the correlation between the canopy spectral data and the nitrogen content, the corresponding wavelength whose absolute value of the Pearson correlation coefficient is greater than the first threshold and whose p value is less than the second threshold is selected as the canopy characteristic wavelength; According to the correlation between leaf spectral data and nitrogen content, the corresponding wavelength whose absolute value of the Pearson correlation coefficient is greater than the first threshold and whose p value is less than the second threshold is selected as the leaf characteristic wavelength.

5. An electronic device, characterized in that: include: one or more processors; A memory having one or more programs stored thereon, which, when the one or more programs are executed by the one or more processors, enables the one or more processors to implement the steps of the method according to any one of claims 1 to 4.

6. A computer-readable storage medium, characterized in that: The computer program is stored therein, and when the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 4 are implemented.

Images