A quantitative inversion method for photosynthetic pigments in Catalpa ovata leaves based on hyperspectral reflectance

Abstract

The present application discloses a method for quantitative inversion of photosynthetic pigments in catalpa ovata leaves based on hyperspectral reflectivity, comprising: using a hyperspectral instrument to collect spectral reflectivity information corresponding to a plurality of catalpa ovata leaves respectively, wherein the spectral reflectivity information is used to indicate the reflectivity of light of different wavelengths of a spectrum on the catalpa ovata leaves; measuring photosynthetic pigment content information of a plurality of catalpa ovata leaves; constructing a sample set according to the spectral reflectivity information and photosynthetic pigment content information respectively corresponding to the plurality of catalpa ovata leaves; determining, according to the sample set, independent variables corresponding to the spectral reflectivity information, and determining an inversion model corresponding to the independent variables; and determining the photosynthetic pigment content information of a target catalpa ovata leaf based on the spectral reflectivity information corresponding to the target catalpa ovata leaf using the inversion model.

Description

Technical Field

[0001] The present application relates to the technical field of spectral analysis, and in particular to a method for quantitatively inverting photosynthetic pigments in catalpa ovata leaves based on high spectral reflectance. Background Art

[0002] Catalpa ovata is a tree of the genus Catalpa in the Bignoniaceae family, which is widely distributed in China. It is widely used in production because of its straight trunk and small expansion and contraction. In order to evaluate the physiological state of Catalpa ovata, it is necessary to measure the photosynthetic pigments in the leaves of Catalpa ovata, so as to evaluate the physiological state of Catalpa ovata according to the content of photosynthetic pigments in the leaves of Catalpa ovata.

[0003] However, the common method of measuring leaf photosynthetic pigments requires destroying the leaves and has the following technical problems:

[0004] 1. Complex sampling methods: UAV platforms or imaging platforms need to segment images, which requires high standards for technicians and the platforms are expensive.

[0005] 2. Damage to tree leaves: In previous empirical studies, the main methods for determining photosynthetic pigment content are spectrophotometry and high performance liquid chromatography. As we all know, both of these methods require destructive sampling, that is, the leaves must be damaged after being picked before quantitative detection can be achieved.

[0006] 3. Inaccurate measurement results: Although the SPAD instrument can achieve the in-situ measurement of tree leaves, it can only detect the relative chlorophyll content of the leaves, and there are still deficiencies in the precise quantitative study of the content of photosynthetic pigments.

[0007] Currently, no effective solution has been proposed to the technical problems that the existing methods for measuring photosynthetic pigments in leaves are complex in sampling, damage to tree leaves and inaccurate in results, making them unsuitable for measuring photosynthetic pigments in catalpa ovata leaves. Summary of the invention

[0008] The embodiments of the present disclosure provide a method for quantitative inversion of photosynthetic pigments in Catalpa ovata leaves based on high spectral reflectance, so as to at least solve the technical problems existing in the existing methods for measuring leaf photosynthetic pigments, namely, complex sampling methods, damage to tree leaves and inaccurate results, making them unsuitable for measuring photosynthetic pigments in Catalpa ovata leaves.

[0009] According to an embodiment of the present disclosure, a method for quantitative inversion of photosynthetic pigments in catalpa ovata leaves based on high spectral reflectance is provided, including: using a hyperspectral instrument to collect spectral reflectance information corresponding to a plurality of catalpa ovata leaves respectively, wherein the spectral reflectance information is used to indicate the reflectance of light of different wavelengths of the spectrum on the catalpa ovata leaves; measuring photosynthetic pigment content information of a plurality of catalpa ovata leaves; constructing a sample set according to the spectral reflectance information and photosynthetic pigment content information respectively corresponding to the plurality of catalpa ovata leaves; determining, according to the sample set, independent variables corresponding to the spectral reflectance information, and determining an inversion model corresponding to the independent variables; and determining the photosynthetic pigment content information of a target catalpa ovata leaf based on the spectral reflectance information corresponding to the target catalpa ovata leaf using the inversion model.

[0010] Therefore, according to the embodiment of the present disclosure, there is no need for the drone platform or the imaging platform to segment the image, and only the spectral reflectance information of the leaves of the Chinese catalpa tree needs to be collected by using a hyperspectral instrument to determine the photosynthetic pigment content information of the leaves of the Chinese catalpa tree, thereby simplifying the sampling means and not requiring damage to the leaves of the trees. And according to this embodiment, the photosynthetic pigment content information of the leaves of the Chinese catalpa tree is determined by verifying and selecting the spectral index with the highest accuracy and the corresponding univariate linear regression inversion model, or the photosynthetic pigment content information of the leaves of the Chinese catalpa tree is determined according to the corresponding PLSR inversion model based on the spectral reflectance derivative data with the highest accuracy, thereby achieving accurate measurement results. This solves the technical problem that the existing methods for determining photosynthetic pigments in leaves have complex sampling means, damage to the leaves of the trees, and inaccurate results, and are therefore not suitable for determining the photosynthetic pigments of the leaves of the Chinese catalpa tree. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The drawings described herein are used to provide a further understanding of the present disclosure and constitute a part of the present application. The illustrative embodiments of the present disclosure and their descriptions are used to explain the present disclosure and do not constitute an improper limitation on the present disclosure. In the drawings:

[0012] Figure 1 It is a schematic diagram of a process for implementing a method for quantitatively inverting photosynthetic pigments in catalpa ovata leaves based on hyperspectral reflectance according to an embodiment of the present disclosure;

[0013] Figure 2 is a schematic diagram of spectral reflectance curves corresponding to a plurality of catalpa ovata leaves respectively according to an embodiment of the present disclosure;

[0014] Figure 3 is a schematic diagram of a catalpa tree leaf after the leaf disc is taken out according to an embodiment of the present disclosure;

[0015] Figure 4 is a schematic diagram of related information of spectral indices corresponding to different wavelength combinations and content information of each photosynthetic pigment according to an embodiment of the present disclosure; and

[0016] Figure 5 It is a schematic diagram of a specific process of the method for quantitative inversion of photosynthetic pigments in catalpa ovata leaves based on hyperspectral reflectance according to an embodiment of the present disclosure. DETAILED DESCRIPTION

[0017] In order to enable those skilled in the art to better understand the technical solutions of the present disclosure, the technical solutions in the embodiments of the present disclosure will be clearly and completely described below in conjunction with the drawings in the embodiments of the present disclosure. Obviously, the described embodiments are only embodiments of a part of the present disclosure, not all of the embodiments. Based on the embodiments in the present disclosure, all other embodiments obtained by ordinary technicians in this field without making creative work should fall within the scope of protection of the present disclosure.

[0018] It should be noted that the terms "first", "second", etc. in the specification and claims of the present disclosure and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchangeable where appropriate, so that the embodiments of the present disclosure described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product, or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products, or devices.

[0019] According to this embodiment, a method for quantitative inversion of photosynthetic pigments in catalpa leaves based on hyperspectral reflectance is provided, wherein Figure 1 A schematic diagram of the process of the method is shown. Figure 1 As shown, the method includes:

[0020] S102: using a hyperspectral spectrometer to collect spectral reflectance information corresponding to a plurality of catalpa ovata leaves respectively, wherein the spectral reflectance information is used to indicate the reflectance of light of different wavelengths of the spectrum on the catalpa ovata leaves;

[0021] S104: measuring photosynthetic pigment content information of multiple catalpa leaves;

[0022] S106: constructing a sample set according to the spectral reflectance information and photosynthetic pigment content information respectively corresponding to a plurality of catalpa leaves;

[0023] S108: determining, according to the sample set, an independent variable corresponding to the spectral reflectance information, and determining an inversion model corresponding to the independent variable; and

[0024] S110: using the inversion model, based on the spectral reflectance information corresponding to the target catalpa tree leaf, determining the photosynthetic pigment content information of the target catalpa tree leaf. The method is specifically described as follows.

[0025] S102: Using a hyperspectral spectrometer to collect spectral reflectance information corresponding to a plurality of catalpa ovata leaves, wherein the spectral reflectance information is used to indicate the reflectance of light of different wavelengths of the spectrum on the catalpa ovata leaves.

[0026] Specifically, in this embodiment, the instrument used for hyperspectral data collection can be, for example, Field Spec3 (Analytical Spectra Devices., Inc., Boulder, CO, USA). The instrument is preheated for 45 minutes before use to eliminate steps, and calibration and optimization are performed every 30 minutes. The surface of the leaves is wiped clean with a smooth flannel cloth just after the leaves are picked, and the spectral reflectance is collected at two locations of the leaves. When measuring, a leaf clamp is used with a fiber optic probe to measure, with the fiber optic probe end close to the front of the leaf, and the leaf clamp is clamped, and the data is saved after the spectral reflectance curve is stable.

[0027] Thus, according to the above method, a corresponding spectral reflectance curve (i.e., spectral reflectance information) can be generated for each of the plurality of catalpa leaves, thereby generating spectral reflectance curves corresponding to different catalpa leaves. Specifically, Figure 2 Schematic diagram of spectral reflectance curves corresponding to a plurality of catalpa leaves. Figure 2 In the figure, the spectral reflectance curves corresponding to different catalpa leaves are shown in different colors.

[0028] S104: Determine photosynthetic pigment content information of a plurality of catalpa ovata leaves;

[0029] Specifically, for each of the plurality of catalpa leaves, a leaf puncher is used to take a leaf disk with a diameter of 1.424 cm near the area where the spectral reflectance is measured. Figure 3 The figure shows a schematic diagram of a leaf of Catalpa ovata after the leaf disc is taken out. Then, two leaf discs are cut into pieces and put into a centrifuge tube containing 95% ethanol. The centrifuge tube contains 10 ml of ethanol. The outer wall of the centrifuge tube is shielded from light with aluminum foil and placed in a dark room for extraction for 48 hours until the leaves turn completely white. The upper clear liquid is taken into a cuvette and its absorbance (OD) at 470 nm, 649 nm, and 665 nm is measured using a spectrophotometer. And based on the measured absorbance, the chlorophyll a content (Chla), chlorophyll b content (Chlb), total chlorophyll (Chl), and carotenoid content (Car) of the Catalpa ovata leaf are determined.

[0030] Among them, the chlorophyll a content (Chla), chlorophyll b content (Chlb), total chlorophyll (Chl) and carotenoid content (Car) can be calculated according to the calculation formula eq1-eq4:

[0031]

[0032]

[0033] Chl (mg cm -2 )=Chla+Chlb (eq3)

[0034]

[0035] In the above formula, OD(w) represents the absorbance at wavelength w, 10 is the volume of the extract, and 3.185 is the area of ​​the two leaf disks.

[0036] Thus, through the above method, for each of the multiple catalpa leaves, the chlorophyll a content (Chla), chlorophyll b content (Chlb), total chlorophyll (Chl) and carotenoid content (Car) are determined. At the same time, on this basis, the ratio of chlorophyll a content (Chla) to chlorophyll b content (Chlb) Chla / Chlb is further determined; and the ratio of total chlorophyll (Chl) to carotenoid content (Car) Chl / Car is determined. Thus, Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car are used as the photosynthetic pigment content information corresponding to the catalpa leaves.

[0037] S106: Constructing a sample set according to the spectral reflectance information and photosynthetic pigment content information respectively corresponding to a plurality of Catalpa ovata leaves.

[0038] Specifically, in this embodiment, for each of the plurality of catalpa leaves, the spectral reflectance curve corresponding to the catalpa leaf and the photosynthetic pigment content information of the catalpa leaf (including chlorophyll a content (Chla), chlorophyll b content (Chlb), total chlorophyll (Chl), carotenoid content (Car), Chla / Chlb and Chl / Car) are combined into one sample. Thus, samples corresponding to the plurality of catalpa leaves can be constructed.

[0039] Then, the Kennard-Stone method is used to divide the multiple samples into a training set and a test set with a ratio of 8:2. The model construction described below uses the training set for correlation coefficient calculation and model training. The test set is used to verify the model accuracy. Among them, the sample set includes, for example, sample S 1 ~S m , the training set includes samples X 1 ~Xn , the test set includes Y 1 ~Y L . Wherein, L = mn.

[0040] Table 1 below summarizes the mean, extreme value, range, and standard deviation of the photosynthetic pigment content in the training set, test set, and overall. The mean values ​​of Chla, Chlb, Chl, Car, Chla / Chlb, and Chl / Car for the overall sample were 49.3347 (mg cm -2 )、19.3405(mg cm -2 )、68.6751(mg cm -2 )、8.4555(mg cm -2 ), 2.5701 and 8.1241. Table 1 shows that the overall mean is between the training set and the test set. In addition, after comparing the maximum, minimum and range of each measurement value of the training set, test set and population, it is found that the maximum, minimum and range of the test set and the population are almost equal, that is, the test set covers the characteristic values ​​of the population sample.

[0041] Table 1. Statistics of photosynthetic pigment content in training and test sets

[0042]

[0043] Among them, MEAN represents mean, MAX represents maximum value, MIN represents minimum value, RAN represents range, and SD represents standard deviation.

[0044] S108: Determine, based on the sample set, an independent variable corresponding to the spectral reflectance information, and determine an inversion model corresponding to the independent variable.

[0045] Specifically, on the one hand, in this example, the spectral index corresponding to the spectral reflectance can be used as an independent variable to construct a univariate linear regression model as an inversion model, thereby realizing convenient inversion model construction. In this embodiment, the spectral index can be, for example, a ratio vegetation index SR or a normalized vegetation index ND. On the other hand, this embodiment can determine the corresponding spectral reflectance derivative data based on the spectral reflectance information of each sample, wherein the derivative data includes the reciprocal logarithm of the reflectance (i.e., the logarithm of the reciprocal of the spectral reflectance), the first-order differential of the reflectance (i.e., the first-order differential of the spectral reflectance curve), the second-order differential of the reflectance (i.e., the second-order differential of the spectral reflectance curve), the first-order differential of the reciprocal logarithm of the reflectance (i.e., the first-order differential of the logarithm of the reciprocal of the spectral reflectance) and the second-order differential of the reciprocal logarithm of the reflectance (i.e., the second-order differential of the logarithm of the reciprocal of the spectral reflectance). Then, the PLSR models are constructed respectively using the spectral reflectance and its derivative data, and the PLSR model with the best regression result is selected as the inversion model. The following are explained in detail:

[0046] (I) Construction of an inversion model based on univariate linear regression (S1081)

[0047] S10811: Construct spectral index.

[0048] Specifically, in this embodiment, in order to construct an inversion model based on univariate linear regression, a spectral index is first constructed. In this embodiment, the spectral index may be, for example, a ratio vegetation index SR (Simple Ratio) or a normalized vegetation index ND (Normalized Difference). The ratio vegetation index SR may be constructed according to the formula eq5 shown below, and the normalized vegetation index ND may be constructed according to the formula eq6 shown below:

[0049]

[0050]

[0051] Wherein, Ra and Rb both represent the spectral reflectance at a wavelength a and a wavelength b in the spectrum.

[0052] Therefore, in this embodiment, the reflectivity of each wavelength in the range of 350 nm to 2500 nm is selected as a quantitative value, and the SR index or ND index is constructed with the reflectivity of each wavelength in the range.

[0053] For example, according to formula eq5, for each sample, the reflectivity corresponding to any pair of wavelengths a and b in the range of 350 nm to 2500 nm can be taken to calculate the SR index SR(a,b), thereby determining a series of SR indexes SR(a,b) corresponding to the sample:

[0054] Similarly, according to formula eq6, for each sample, the reflectivity corresponding to any pair of wavelengths in the range of 350 nm to 2500 nm can be taken to calculate the ND index ND(a, b), thereby determining a series of ND indices corresponding to the sample.

[0055] Referring to the above, for each sample in the sample set, the SR index or ND index is constructed. That is, for each sample in the sample set, the SR index SR(a,b) and the ND index ND(a,b) (where a and b are any wavelengths in the range of 350nm to 2500nm) are calculated. Thus, the spectral index is constructed for the sample set.

[0056] S10812: Determine, based on the spectral index and photosynthetic pigment content information of each sample in the sample set, the correlation information between the spectral index corresponding to different wavelength combinations and each type of photosynthetic pigment content information.

[0057] Specifically, in this embodiment, the spectral index corresponding to different wavelength combinations (also called bivariate factors) is, for example, the SR index or ND index corresponding to different wavelength combinations (also called bivariate factors) in the range of 350 nm to 2500 nm. Different types of photosynthetic pigment content information include chlorophyll a content (Chla), chlorophyll b content (Chlb), total chlorophyll (Chl), carotenoid content (Car), Chla / Chlb and Chl / Car.

[0058] In addition, in this embodiment, the relevant information may be, for example, a Pearson correlation coefficient, wherein formula eq7 shows a calculation formula for the Pearson correlation coefficient;

[0059]

[0060] Where N represents the number of samples, μA and σA represent the mean and variance of variable A, μB and σB represent the mean and variance of variable B, respectively.

[0061] Therefore, in this embodiment, the variables corresponding to each sample include: spectral index SR (a, b) or ND (a, b) corresponding to different wavelength combinations (where a and b are any wavelengths in the range of 350 nm to 2500 nm); and photosynthetic pigment content information, including chlorophyll a content (Chla), chlorophyll b content (Chlb), total chlorophyll (Chl), carotenoid content (Car), Chla / Chlb and Chl / Car.

[0062] Therefore, in this embodiment, for any type of photosynthetic pigment content information, the correlation information between SR(a, b) or ND(a, b) of different wavelength combinations and the photosynthetic pigment content information can be calculated.

[0063] For example, for chlorophyll a content (Chla), the Chla and SR (350, 350) of each sample in the training set can be substituted into eq7 as A and B respectively, so as to calculate the Pearson correlation coefficient between Chla and SR (350, 350). And so on, Figure 4 (a) shows the correlation coefficient between the SR index and Chla corresponding to different wavelength combinations.

[0064] Specifically, Figure 4The horizontal and vertical coordinates of (a) are wavelengths ranging from 350nm to 2500nm, so each point in the coordinate system corresponds to a different wavelength combination (a, b) represented by the horizontal and vertical coordinates. And the color of each point is used for the correlation coefficient between the SR index SR (a, b) corresponding to the wavelength combination and Chla.

[0065] Likewise, Figure 4 (c) Figure 4 (e) Figure 4 (g) Figure 4 (i) and Figure 4 (k) shows the correlation coefficients of the constructed SR index SR(a,b) and Chlb, Chl, Car, Chla / Chlb and Chl / Car, respectively. Figure 4 (c) shows the correlation coefficient between the constructed SR index SR(a,b) and Chlb; Figure 4 (e) shows the correlation coefficient between the constructed SR index SR(a,b) and Chl; Figure 4 (g) shows the correlation coefficient between the constructed SR index SR(a,b) and Car; Figure 4 (i) shows the correlation coefficient between the constructed SR index SR(a,b) and Chla / Chlb; Figure 4 (k) shows the correlation coefficient between the constructed SR index SR(a,b) and Chl / Car, where a and b are arbitrary wavelengths in the range of 350 nm to 2500 nm.

[0066] also, Figure 4 (b) Figure 4 (d) Figure 4 (f) Figure 4 (h) Figure 4 (j) and Figure 4 (l) represent the correlation coefficients between the constructed ND index and Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car, respectively. Figure 4 (b) shows the correlation coefficient between the constructed ND index ND(a,b) and Chla; Figure 4 (d) shows the correlation coefficient between the constructed ND index ND(a,b) and Chlb; Figure 4 (f) shows the correlation coefficient between the constructed ND index ND(a,b) and Chl; Figure 4 (h) shows the correlation coefficient between the constructed ND index ND(a,b) and Car; Figure 4 (j) shows the correlation coefficient between the constructed ND index ND(a,b) and Chla / Chlb; Figure 4(l) shows the correlation coefficient between the constructed ND index ND(a,b) and Chl / Car, where a and b are arbitrary wavelengths in the range of 350 nm to 2500 nm.

[0067] S10813: For different types of photosynthetic pigment content information, the corresponding spectral index is determined as the independent variable according to the extreme value of the absolute value of the correlation coefficient, and a univariate linear regression inversion model is constructed.

[0068] Specifically, refer to Figure 4 As shown in (a), it is assumed that the absolute value of the correlation coefficient between the SR index corresponding to wavelength a1 and wavelength b1 and the chlorophyll a content Chla is an extreme value (that is, the color of the coordinate point (a1, b1) is closer to the color corresponding to 1 or -1). Then SR(a1, b1) is used as the independent variable corresponding to the chlorophyll a content Chla. The values ​​of SR(a1, b1) corresponding to each sample in the training set are calculated, and the values ​​of the chlorophyll a content Chla corresponding to each sample are used as the model construction samples. A univariate linear regression model is constructed by the least squares method to achieve convenient inversion model construction. The univariate linear regression inversion model thus constructed can determine the chlorophyll a content Chla corresponding to the catalpa leaf according to the SR(a1, b1) corresponding to the catalpa leaf.

[0069] Likewise, reference Figure 4 As shown in (c), SR(a2, b2) can be used as the independent variable corresponding to the chlorophyll b content Chlb. The values ​​of SR(a2, b2) corresponding to each sample in the training set and the values ​​of the chlorophyll b content Chlb corresponding to each sample are used as the model construction samples, and a univariate linear regression model is constructed by the least squares method to achieve a convenient univariate linear regression inversion model. Among them, the absolute value of the correlation coefficient between the SR index SR(a2, b2) corresponding to the wavelength a2 and the wavelength b2 and the chlorophyll b content Chlb is an extreme value.

[0070] By analogy, for Figure 4 (e) Figure 4 (g) Figure 4 (i) and Figure 4 (k), the SR index of the wavelength combination (also called bivariate factor) corresponding to the absolute extreme value of the correlation coefficient can be selected as the independent variable, and a univariate linear regression inversion model can be constructed to determine the corresponding Chl, Car, Chla / Chlb and Chl / Car according to the corresponding SR index of the corresponding wavelength combination.

[0071] Similarly, for reference Figure 4 (b) Figure 4 (d) Figure 4 (f) Figure 4 (h) Figure 4 (j) and Figure 4 (l), the ND index corresponding to the absolute value extreme of the correlation coefficient (also known as the bivariate factor) can be selected respectively as the independent variable to construct a unary linear regression inversion model for determining the corresponding Chla, Chlb, Chl, Car, Chla / Chlb, and Chl / Car.

[0072] In this application, the coefficient of determination (R 2 ), root mean square error (RMSE), and ratio of performance to deviation (RPD) are used to evaluate the prediction performance of the model. Among them, when RPD < 1.4, it indicates that the model is unreliable and cannot be used for model analysis and evaluation; when 1.4 < RPD < 2.0, the model is considered relatively reliable; when RPD > 2.0, the model is considered highly reliable and can be used for model quantitative analysis. The calculation formulas for R2, RMSE, and RPD are expressed as eq8, eq9, and eq10 respectively:

[0073]

[0074]

[0075]

[0076] Table 2 below shows the results of the unary linear regression of the bivariate factor with the largest correlation coefficient between SR and ND and the target.

[0077] Table 2: Results of the unary linear regression of the bivariate factor with the largest correlation coefficient between SR and ND and the target

[0078]

[0079] Referring to Table 2, unary linear regression is performed by respectively selecting SR and ND with the largest absolute value of the correlation coefficient. The bivariate factors of the independent variable, the maximum value of the correlation coefficient, and the R2, RMSE, and RPD values of the training set and test set are shown in Table 2. It can be seen from Table 2 that the bivariate factors corresponding to the maximum correlation coefficients of SR and ND with Chla and Chl are the same, both being the reflectances at 753 nm and 733 nm. The RPD values of the unary linear regression models of Chla and Car constructed by SR and ND are both greater than 2.0, indicating that the model has excellent inversion ability. The RPD value of the unary linear regression model of Chlb constructed by SR and ND is close to 2.0. The RPD value of the unary linear regression model of Chl / Car is greater than 1.4, proving that the model is at least reliable.

[0080] (II) Constructing a PLSR inversion model (S1082)

[0081] As mentioned above, in addition to constructing a convenient inversion model based on a univariate regression model, the present application also constructs a PLSR inversion model based on a PLSR model (partial least squares regression model) according to the spectral reflectance derivative data corresponding to the spectral reflectance information.

[0082] S10821: Determine spectral reflectance derivative data corresponding to the spectral reflectance information of each sample.

[0083] Specifically, for Figure 2 The spectral reflectance information of each sample shown is used to calculate the corresponding spectral reflectance derivative data, including: the reciprocal logarithm of the reflectance, the first-order differential of the reflectance, the second-order differential of the reflectance, the first-order differential of the reciprocal logarithm of the reflectance, and the second-order differential of the reciprocal logarithm of the reflectance.

[0084] Thus, the spectral reflectance derivative data of each sample includes: spectral reflectance (R), first-order differential of reflectance (FDR), second-order differential of reflectance (SDR), reciprocal logarithm of reflectance (RLR), first-order differential of reciprocal logarithm of reflectance (FDRL) and second-order differential of reciprocal logarithm of reflectance (SDRL). In addition, in this embodiment, for each sample, the above-mentioned spectral reflectance derivative data has different values ​​relative to different wavelengths of 350nm to 2500nm.

[0085] S1082: Using different types of spectral reflectance derived data as independent variables, construct corresponding PLSR inversion models for determining the photosynthetic pigment content information of Catalpa ovata leaves.

[0086] Specifically, in this embodiment, PLSR inversion models are constructed for spectral reflectance (R), first-order differential of reflectance (FDR), second-order differential of reflectance (SDR), reciprocal logarithm of reflectance (RLR), first-order differential of reciprocal logarithm of reflectance (FDRL) and second-order differential of reciprocal logarithm of reflectance (SDRL), respectively.

[0087] For example, for spectral reflectance (R), the spectral reflectance (R) values ​​of different wavelengths (e.g., 350nm-2500nm) are used as multiple independent variables, and the photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car are used as dependent variables. A PLSR inversion model 1 is constructed, and the PLSR inversion model 1 is used to determine the photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car of the catalpa ovata leaf according to the spectral reflectance of the catalpa ovata leaf (e.g., 350nm-2500nm).

[0088] For the first order differential of reflectance (FDR), the first order differential of reflectance (FDR) values ​​at different wavelengths (e.g., 350nm to 2500nm) are used as multiple independent variables, and the photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car are used as dependent variables. A PLSR inversion model 2 is constructed, which is used to determine the photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car of the catalpa ovata leaf according to the first order differential of reflectance of the catalpa ovata leaf (e.g., 350nm to 2500nm).

[0089] By analogy, a PLSR inversion model 3 can be constructed, which is used to determine the photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car of the catalpa leaves according to the second-order differential of the reflectivity of the catalpa leaves (e.g., 350nm to 2500nm). And, a PLSR inversion model 4 is constructed, which is used to determine the photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car of the catalpa leaves according to the reciprocal logarithm of the reflectivity of the catalpa leaves (e.g., 350nm to 2500nm). And, construct a PLSR inversion model 5, which is used to determine the photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car of the catalpa leaves according to the reciprocal logarithm first-order differential of the reflectivity of the catalpa leaves (e.g., 350nm-2500nm). And, construct a PLSR inversion model 6, which is used to determine the photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car of the catalpa leaves according to the reciprocal logarithm second-order differential of the reflectivity of the catalpa leaves (e.g., 350nm-2500nm).

[0090] S1083: Use the test set to verify each PLSR inversion model and determine the optimal PLSR model.

[0091] Specifically, in the present embodiment, in order to determine the number of principal component factors during the model construction process, random samples in the training set are selected and screened as the validation set. Considering that the validation set samples should be random and the model capability needs to be verified. Therefore, this study verified each principal component score 50 times, randomly selected 20% of the training set as the validation set for each verification, and recorded the average RMSE value. The program was run according to the experience of the first 30 main components. When the RMSE of the validation set is the smallest, the model determined by the number of principal component factors is the best PLSR model. The PLSR model constructed with the above-mentioned number of principal component factors was tested using the test set samples. The corresponding species of the PLSR model with the best regression result (highest accuracy) is selected as the input variable to achieve accurate quantitative inversion model construction of photosynthetic pigment content.

[0092] In this study, PLS regression was performed on the photosynthetic pigment content, and hyperspectral reflectance, the reciprocal logarithm of reflectance, and their first-order differentials and second-order differentials were selected. However, it was found during the regression process that errors would occur when the second-order differential was scaled. Therefore, Table 3 only shows the results of PLSR using hyperspectral reflectance, the reciprocal logarithm of reflectance, and their first-order differentials. As can be seen from Table 3, when reflectance is used as the independent variable, the RPD of the regression results of the Chla and Chl training and test sets is greater than 2.0. When FDR, RLR and FDRL are used as independent variables, the RPD of the regression results of the Chla, Chlb and Chl training and test sets is greater than 2.0. The RPD of the regression of the Car training and test sets is also greater than 1.4. In addition, Chla / Chlb in each regression equation shows that the RPD of the training set is much larger than that of the test set. Even if the validation set is used to correct the PLSR, overfitting still occurs. It is worth mentioning that the RPD values ​​of Chl / Car in the PLSR model with R, RLR and FDLR as independent variables are greater than 1.4 for both the training set and the test set.

[0093] Table 3. PLSR results of photosynthetic pigments for each factor:

[0094]

[0095] Therefore, in summary, the regression results of PLSR show that the best results are obtained when Chla content and Chlb content are taken as input variables using FDRL (i.e., PLSR inversion model 5), the best results are obtained when Chl content and Car content are taken as input variables using R (i.e., PLSR inversion model 1), and the best results are obtained when Chla / Chlb content and Chl / Car content are taken as input variables using RLR (i.e., PLSR inversion model 4).

[0096] S110: Determine the photosynthetic pigment content information of the target Catalpa ovata leaf based on the spectral reflectance information corresponding to the target Catalpa ovata leaf by using the inversion model.

[0097] Specifically, after determining the optimal univariate linear regression inversion model and PLSR inversion model for inverting Catalpa ovata leaves, the model can be used to determine the photosynthetic pigment content information of the Catalpa ovata leaves.

[0098] Specifically, for the target catalpa tree leaf to be detected, a hyperspectral instrument is used to collect the spectral reflectance information corresponding to the target catalpa tree leaf. The spectral index (i.e., SR index or ND index) corresponding to the optimal univariate linear regression inversion model is determined based on the collected spectral reflectance information, and then substituted into the corresponding univariate linear regression inversion model to determine the corresponding photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car of the target catalpa tree leaf.

[0099] Alternatively, the spectral reflectance derivative data corresponding to the optimal PLSR inversion model is determined based on the collected spectral reflectance information, and substituted into the corresponding PLSR inversion model to determine the corresponding photosynthetic pigment content information Chla, Chlb, Chl, Car, Chla / Chlb and Chl / Car of the target catalpa tree leaves.

[0100] at last, Figure 5 The specific flow chart of the method described in this embodiment is shown. Figure 5 As shown, the method described in this embodiment includes:

[0101] First, a plurality of catalpa ovata leaves are selected and collected (S202).

[0102] Then, a hyperspectral spectrometer is used to collect corresponding spectral reflectance information for each catalpa tree leaf (S204).

[0103] Then, the photosynthetic pigment content of each catalpa elata leaf is measured, including the chlorophyll a content (Chla), chlorophyll b content (Chlb), total chlorophyll (Chl) and carotenoid content (Car) of the catalpa elata leaf (S206).

[0104] Then, according to the photosynthetic pigment content of each catalpa leaf, the corresponding photosynthetic pigment content ratio is determined, including: the ratio of chlorophyll a content (Chla) to chlorophyll b content (Chlb) Chla / Chlb; and the ratio of total chlorophyll (Chl) to carotenoid content (Car) Chl / Car, thereby constructing a sample set (S208).

[0105] Then, for each catalpa leaf sample, a reflectance index is constructed, such as an SR index and an ND index ( S210 ).

[0106] Then, the correlation coefficients between the reflectance index and the photosynthetic pigment content (ratio) corresponding to different bivariate factors were calculated, and a correlation coefficient matrix was constructed based on the correlation coefficients. Figure 4 Each graph indicates information of a corresponding correlation coefficient matrix (S212).

[0107] The reflectance index and the photosynthetic pigment content (ratio) of the bivariate factor corresponding to the maximum value in the absolute value of the correlation coefficient are determined (S214).

[0108] The reflectance index corresponding to the selected bivariate factor and the numerical value corresponding to the selected photosynthetic pigment content (ratio) in each sample are used to construct a model, and a simple variable regression model (i.e., a univariate linear regression inversion model) is constructed (S216).

[0109] The corresponding reflectance derivative data are calculated according to the spectral reflectance information of each sample, including: first-order differential of reflectance (FDR), second-order differential of reflectance (SDR), reciprocal logarithm of reflectance (RLR), first-order differential of reciprocal logarithm of reflectance (FDRL) and second-order differential of reciprocal logarithm of reflectance (SDRL) (S218).

[0110] A PLSR model (S220) is constructed with spectral reflectance (R), first differential of reflectance (FDR), second differential of reflectance (SDR), reciprocal logarithm of reflectance (RLR), first differential of reciprocal logarithm of reflectance (FDRL) and second differential of reciprocal logarithm of reflectance (SDRL) as independent variables.

[0111] The PLSR inversion model with the best inversion effect is determined, and the suboptimal PLSR inversion model is discarded (S222, S224), thereby determining a high-precision PLSR regression inversion model (S226).

[0112] For the target catalpa tree leaf to be detected, spectral reflectance information is collected, and the photosynthetic pigment content (ratio) of the target catalpa tree leaf is quantitatively inverted (S228) using the simple variable regression model constructed in S216 or the high-precision regression model determined in S226, thereby determining the catalpa tree photosynthetic pigment content of the target catalpa tree leaf (S230).

[0113] Therefore, according to the embodiment of the present disclosure, there is no need for the drone platform or the imaging platform to segment the image, and only the spectral reflectance information of the leaves of the Chinese catalpa tree needs to be collected by using a hyperspectral instrument to determine the photosynthetic pigment content information of the leaves of the Chinese catalpa tree, thereby simplifying the sampling means and not requiring damage to the leaves of the trees. And according to this embodiment, the photosynthetic pigment content information of the leaves of the Chinese catalpa tree is determined by verifying and selecting the spectral index with the highest accuracy and the corresponding univariate linear regression inversion model, or the photosynthetic pigment content information of the leaves of the Chinese catalpa tree is determined according to the corresponding PLSR inversion model based on the spectral reflectance derivative data with the highest accuracy, thereby achieving accurate measurement results. This solves the technical problem that the existing methods for determining photosynthetic pigments in leaves have complex sampling means, damage to the leaves of the trees, and inaccurate results, and are therefore not suitable for determining the photosynthetic pigments of the leaves of the Chinese catalpa tree.

[0114] The serial numbers of the above embodiments of the present invention are only for description and do not represent the advantages or disadvantages of the embodiments.

[0115] In the above embodiments of the present invention, the description of each embodiment has its own emphasis. For parts that are not described in detail in a certain embodiment, reference can be made to the relevant descriptions of other embodiments.

[0116] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0117] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0118] In addition, each functional unit in each embodiment of the present invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit. The above-mentioned integrated unit may be implemented in the form of hardware or in the form of software functional units.

[0119] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, RandomAccess Memory), mobile hard disk, magnetic disk or optical disk and other media that can store program codes.

[0120] The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the scope of protection of the present invention.

Claims

1. A method for quantitative inversion of photosynthetic pigments in catalpa leaves based on hyperspectral reflectance, characterized in that: include: Using a hyperspectral spectrometer to collect spectral reflectance information corresponding to a plurality of catalpa ovata leaves, wherein the spectral reflectance information is used to indicate the reflectance of light of different wavelengths of the spectrum on the catalpa ovata leaves; Measuring the photosynthetic pigment content information of the plurality of catalpa ovata leaves; Constructing a sample set according to the spectral reflectance information and photosynthetic pigment content information respectively corresponding to the plurality of catalpa leaves; Determine, according to the sample set, an independent variable corresponding to the spectral reflectance information, and determine an inversion model corresponding to the independent variable; as well as Using the inversion model, based on the spectral reflectance information corresponding to the target catalpa tree leaves, the photosynthetic pigment content information of the target catalpa tree leaves is determined, and wherein Determining, according to the sample set, an independent variable corresponding to the spectral reflectance information, and determining an operation of an inversion model corresponding to the independent variable, including: Determine a maximum value of a correlation coefficient between a spectral index corresponding to a bivariate factor of each arbitrary wavelength combination within a preset spectral range and at least one type of photosynthetic pigment content information, wherein the spectral index is based on the spectral reflectance information; Taking the spectral index corresponding to the bivariate factor of the wavelength combination corresponding to the maximum value of the correlation coefficient as the independent variable; and A univariate linear regression inversion model for determining the at least one type of photosynthetic pigment content information is constructed by least squares method according to the spectral index corresponding to the bivariate factor, and wherein: According to the sample set, the operation of taking the spectral index corresponding to the bivariate factor as the independent variable includes: taking the ratio vegetation index corresponding to the bivariate factor as the independent variable; or taking the normalized vegetation index corresponding to the bivariate factor as the independent variable.

2. The method according to claim 1, characterized in that The operation of measuring the photosynthetic pigment content information of the plurality of catalpa ovata leaves comprises: Cutting some leaves of the catalpa tree leaves into pieces and putting them into a centrifuge tube containing ethanol for extraction; After the extraction operation, take the upper clear liquid and place it in a cuvette; Determining the absorbance of the upper clear liquid to light of a specified wavelength; and The photosynthetic pigment content information of the catalpa elata leaf is determined according to the absorbance.

3. The method according to claim 2, characterized in that The operation of determining the photosynthetic pigment content information of the catalpa ovata leaf according to the absorbance comprises: Determining the chlorophyll a content, chlorophyll b content, total chlorophyll content and carotenoid content of the catalpa ovata leaf according to the absorbance; and The ratio of the chlorophyll a content to the chlorophyll b content of the catalpa elata leaf, and the ratio of the total chlorophyll content to the carotenoid content of the catalpa elata leaf are determined.

4. The method according to claim 1, characterized in that: The correlation coefficient is the Pearson correlation coefficient.

5. The method according to claim 1, characterized in that The operation of determining the photosynthetic pigment content information of the target catalpa ovata leaf by using the inversion model based on the spectral reflectance information corresponding to the target catalpa ovata leaf comprises: Determining a spectral index corresponding to the bivariate factor based on spectral reflectance information corresponding to the target catalpa tree leaves; and The determined spectral index is substituted into the univariate linear regression inversion model to determine the photosynthetic pigment content information of the target catalpa tree leaves.

6. The method according to claim 1, characterized in that The method further comprises: determining, according to the sample set, an independent variable corresponding to the spectral reflectance information, and determining an operation of an inversion model corresponding to the independent variable. Determining, based on the sample set, spectral reflectance derivative data corresponding to the spectral reflectance information; Determine the spectral reflectance information and each of the spectral reflectance derived data as independent variables, and respectively construct PLSR inversion models for determining the photosynthetic pigment content information; and From the constructed PLSR inversion models, the most optimized PLSR inversion model is determined.

7. The method according to claim 6, characterized in that From the constructed PLSR inversion model, determine the operation of the optimized PLSR inversion model, including: According to the sample set, the determination coefficient R of the constructed PLSR inversion model is determined 2 , root mean square error RMSE and performance to deviation ratio RPD; and According to the determination coefficient R of the constructed PLSR inversion model 2 , root mean square error RMSE and performance to deviation ratio RPD, and determine the optimized PLSR inversion model.

8. The method according to claim 6, characterized in that The operation of determining the photosynthetic pigment content information of the target catalpa ovata leaf by using the inversion model based on the spectral reflectance information corresponding to the target catalpa ovata leaf comprises: Determining independent variables corresponding to the optimized PLSR inversion model from the spectral reflectance information and spectral reflectance derived data of the target catalpa tree leaves; and The determined independent variables are substituted into the optimized PLSR inversion model to determine the photosynthetic pigment content information of the target catalpa ovata leaves.

Images