Method for identifying eggs of tenebrio molitor by using polarized light imaging technology

Abstract

The invention discloses a tenebrio molitor egg identification method using a polarized light imaging technology, and the method comprises the following specific steps: 1, collecting an image of a target tea tree leaf through a polarized light imaging system, and calculating a Mueller matrix image of the target tea tree leaf; step 2, calculating a covariance matrix of the Mueller matrix through the Mueller matrix image, and obtaining a characteristic value of the Mueller matrix through the covariance matrix; 3, calculating a polarization purity index and an overall polarization purity parameter of the target tea tree leaf according to the characteristic value; 4, acquiring a polarization fusion image; 5, calculating a gray level co-occurrence matrix of the polarization fusion image, and calculating contrast, energy, inverse difference moment and entropy; step 6, training the support vector machine to obtain a trained support vector machine; and step 7, inputting the contrast, the energy, the inverse difference moment and the entropy into the trained support vector machine to realize the detection of the tenebrio molitor eggs. The method is high in recognition accuracy and high in working efficiency.

Description

Technical Field

[0001] The invention belongs to the technical field of insect egg identification methods, and particularly relates to a method for identifying tea net bug eggs using polarized light imaging technology. Background Art

[0002] Polarization information carries multidimensional vector information and can be used as a carrier to convey richer information. During the propagation of light, it interacts with and scatters particles in the environment, changing the distribution of light and the information it carries.

[0003] The tea net bug, a typical pest in tea-growing areas, has rapidly spread across tea plantations across China in recent years, severely impacting both yield and quality. The bug primarily infests mature leaves, but in severe cases also attacks tender leaves. Its nymphs and adults suck the sap from tea leaves, causing dense white spots on the leaves. In severe cases, entire leaves may turn brown and fall off.

[0004] The tea net bug reproduces two to three generations annually, with eggs entering a dormant period in winter. Detection of overwintering eggs provides data support for predicting the peak occurrence of the first generation of nymphs. Accurate egg monitoring can predict the peak hatching period, providing a scientific basis for targeted pesticide application, significantly reducing pesticide use. Tea net bug eggs are laid in the mesophyll of tea leaves, with an average length of 0.45 mm and a width of 0.24 mm. The operculum is oval or pear-shaped, measuring 0.14 mm long and 0.09 mm wide, and is located on the dorsal surface of the leaf, making it difficult to directly observe. Monitoring tea net bug eggs presents numerous technical challenges: the eggs are tiny and are often laid in the tissue near the main vein on the dorsal surface of the leaf; the eggs are covered with a black secretion, similar in color to the dorsal surface of the leaf. Traditional identification relies primarily on manual experience and microscopic identification. This method has significant drawbacks: manual identification accuracy is significantly affected by the user's experience and is inefficient; microscopic observation requires time-consuming sample processing and results in low identification efficiency. Summary of the Invention The present invention aims to provide a method for identifying tea net bug eggs using polarized light imaging technology, so as to solve the problems of poor recognition accuracy and low working efficiency in existing methods.

[0005] The technical solution adopted by the present invention is a method for identifying tea net bug eggs using polarized light imaging technology, and the specific steps are as follows: Step 1, collecting an image of a target tea leaf through a polarized light imaging system, and calculating a Mueller matrix image of the target tea leaf; Step 2: Calculate the covariance matrix of the Mueller matrix through the Mueller matrix image H , and obtain its eigenvalues ​​from the covariance matrix 、 、 、 ; Step 3: According to the eigenvalue 、 、 、 Calculate the polarization purity index of the target tea leaves 、 、 and the overall polarization purity parameter PI ; Step 4: and PI Perform fusion to obtain a polarization fusion image; Step 5, calculate the gray level co-occurrence matrix of the polarization fusion image and calculate the contrast, energy, inverse moment and entropy; Step 6: training the support vector machine to obtain a trained support vector machine; Step 7: Input the contrast, energy, inverse moment and entropy into the trained support vector machine to detect the eggs of the tea net stink bug. The present invention is also characterized in that: The specific process of step 1 is: Step 1.1, collecting images of target tea leaves using a polarized light imaging system; Step 1.2, record the Mueller matrix of the target tea leaf as , the Mueller matrix of the first polarizer is recorded as , the Mueller matrix of the first quarter wave plate is recorded as , the Mueller matrix of the second quarter-wave plate is recorded as , the Mueller matrix of the second polarizer is recorded as , the Stokes vectors of the incident light and the outgoing light are respectively recorded as and , then: (1) What the CMOS camera receives is the light intensity signal, that is, The first component After expansion: (2) In formula (2), is the Mueller matrix element of the target tea leaf, is a matrix u Array element in; Among them, the matrix u for: (3) According to formula (3), formula (2) can be rewritten into Fourier series form, and we have: (4) According to the Fourier series expansion, the corresponding Fourier coefficients are obtained 、 , a total of 25, solve the 16 array elements of the Mueller matrix from the 25 Fourier coefficients, the expression is: (5) Step 1.3, extract the grayscale values ​​of the pixels at the same position in the 30 target tea leaf images collected in step 1.1 to form a column matrix with 30 columns and one row. I (30 times of receiving the column matrix composed of light intensity), through I Solve for 25 Fourier coefficients, the expression is: (6) In formula (6), D It is a 30×15 matrix; F is a column matrix consisting of 25 Fourier coefficients to be solved; Equation (6) belongs to an overdetermined system of equations, which is usually solved using a pseudo-inverse matrix. The specific expression is as follows: (7) Then get F Substitute the array element in formula (5) to obtain the Mueller matrix of the current pixel point M ; Step 1.4, repeat step 1.3 to obtain the Mueller matrix of pixels at different positions in 30 target tea leaf images M , all Mueller matrices M The values ​​of the array elements at the same position in the image are combined according to the corresponding pixel positions to obtain 16 Mueller matrix array element images, thereby obtaining a Mueller matrix image.

[0006] The specific process of step 1.1 is: The polarized light imaging system includes a light source, a polarizer, an analyzer, and a CMOS camera. The polarizer is connected to a first controller, the analyzer is connected to a second controller, and both the first controller and the second controller are connected to a power supply. The polarizer includes a first polarizer and a first quarter wave plate, and the first polarizer is arranged close to the light source; The polarization analyzer includes a second polarizer and a second quarter-wave plate, and the second polarizer is arranged close to the CMOS camera; During the acquisition process, the target tea leaves are placed between the polarizer and the analyzer. The light source emits parallel non-polarized light with stable light intensity and is incident on the first polarizer. The first quarter-wave plate rotates to modulate the non-polarized light into a specific polarization state and then emits it. After the polarized light is irradiated on the surface of the target tea leaves, the light after interacting with the target tea leaves passes through the second quarter-wave plate and finally passes through the second polarizer and is received by the CMOS camera. The first quarter-wave plate rotates 6° each time, and the second quarter-wave plate rotates 30° in the same direction. 30 rotations are performed to obtain 30 images of the target tea leaves.

[0007] In step 3, the polarization purity index of the target tea leaf is calculated 、 、 The expression is: (8); In formula (8), tr is the trace of the matrix, which is the sum of the elements on the main diagonal of the matrix.

[0008] In step 3, calculate the overall polarization purity parameter PI The expression is: (9).

[0009] The specific process of step 4 is: polarization purity index Mapped to the red channel, the overall polarization purity parameter PI Map to the green channel and set the blue channel to 0 to obtain the polarization fused image.

[0010] The specific process of step 5 is: Step 5.1: Compress the grayscale of the polarization fusion image and select any point in the compressed polarization fusion image. O , the coordinates are , and select another point that deviates from this point Q , the coordinates are , a 、 b is an integer, and the point O and point Q As a pixel pair, since point Q Select from the entire compressed polarization fusion image, then a 、 b The values ​​are different, so that several pixel pairs can be formed and the grayscale value of each pixel pair is recorded. , assuming that the maximum grayscale of the compressed polarization fusion image is L ,So and Total combination of L × L, traverse the entire compressed polarization fusion image, and count each The frequency of occurrence of the value, and arranged in order into a L × L The upper left corner of the matrix is ​​the frequency of (1,1), and the lower right corner is ( L , L ), then the rows and columns of the matrix are 1, 2, …, L , the elements in the matrix are each Frequency of occurrence; Among them, the grayscale level is compressed to 8 or 16 levels, then L 8 or 16; Step 5.2, divide each element in the matrix by the sum of all elements to get each Probability of occurrence , the elements in the matrix are represented by the probability corresponding to each element Replacement, the square matrix after replacement is the gray level co-occurrence matrix; Step 5.3, calculate contrast, energy, inverse moment and entropy through gray level co-occurrence matrix, the expression is: (10) (11) (12) (13) In formulas (10) to (13), Represents the gray level co-occurrence matrix i Rank j Elements of the column; Contrast is contrast, Energy is energy, IDM is inverse difference moment, and Entropy is entropy.

[0011] The specific process of step 6 is: In step 6.1, build an SVM classifier based on the MATLAB R2021b platform. The SVM classifier selects the linear kernel function, the loss function adopts the SVM standard loss function, the penalty term type is L2 regularization, and the value of the regularization parameter C is 1. Step 6.2: Use a 5-fold cross-validation strategy to divide the data set into a training set and a test set. The training set is input into the support vector machine in step 6.1 for training to obtain a trained support vector machine. In step 6.3, the test set is input into the trained support vector machine for verification.

[0012] The beneficial effects of the present invention are as follows: the present invention utilizes the tea stink bug egg identification method using polarized light imaging technology to realize non-destructive detection of tea stink bug eggs. Compared with existing tea stink bug egg identification methods, such as manual identification and microscope observation, firstly, non-destructive detection of tea stink bug eggs can be realized without picking the leaves of the tested plant. Only the polarized imaging system needs to be fixed and images need to be collected for analysis. Secondly, the polarized optical properties characterize the changes in the leaf microstructure caused by insect egg infection, which can realize high-precision insect egg detection. Finally, compared with the manual observation method, its accuracy has been greatly improved and the efficiency is higher. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] Figure 1 A schematic structural diagram of the polarization imaging system in the method of the present invention; Figure 2 The Mueller matrix of tea leaves infected by tea lacewing insect eggs in the method of the present invention; Figure 3 is the polarization purity index of the tea leaves infected by the eggs of the tea lacewing bug in the method of the present invention images; Figure 4 is the polarization purity index of the tea leaves infected by the eggs of the tea lacewing bug in the method of the present invention images; Figure 5 is the polarization purity index of the tea leaves infected by the eggs of the tea lacewing bug in the method of the present invention images; Figure 6 is the overall polarization purity parameter of the tea leaves infected by the eggs of the tea net bug in the method of the present invention PI images; Figure 7 Figure 1 is a diagram of the polarization image fusion process and results in the method of the present invention; Figure 8 Schematic diagram of the calculation process of the gray-level co-occurrence matrix in the method of the present invention.

[0014] 1. Light source, 2. Polarizer, 3. Analyzer, 4. CMOS camera, 5. First controller, 6. Second controller, 7. Power supply, 8. Target tea leaf; 2-1. First polarizer, 2-2. First quarter-wave plate; 3-1. Second polarizer, 3-2. Second quarter-wave plate. DETAILED DESCRIPTION

[0015] The present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0016] Example 1 The present invention uses a method for identifying tea net bug eggs using polarized light imaging technology, which is specifically implemented according to the following steps: Step 1, collecting an image of a target tea leaf through a polarized light imaging system, and calculating a Mueller matrix image of the target tea leaf; Step 2: Calculate the covariance matrix of the Mueller matrix through the Mueller matrix image H , and the covariance matrix H Get the covariance matrix H The eigenvalue of 、 、 、 ; Step 3: According to the eigenvalue 、 、 、 Calculate the polarization purity index of the target tea leaves 、 、 and the overall polarization purity parameter PI ; Step 4: and PI Perform fusion to obtain a polarization fusion image; Step 5, calculate the gray level co-occurrence matrix of the polarization fusion image and calculate the contrast, energy, inverse moment and entropy; Step 6: training the support vector machine to obtain a trained support vector machine; Step 7: Input the contrast, energy, inverse moment and entropy into the trained support vector machine to detect the eggs of the tea net stink bug.

[0017] Example 2 Based on Example 1, the specific process of step 1 is as follows: Step 1.1, collecting images of target tea leaves using a polarized light imaging system; Among them, the polarized light imaging system has a structure such as Figure 1 As shown, it includes a light source 1, a polarizer 2, an analyzer 3, and a CMOS camera 4. The polarizer 2 is connected to a first controller 5, the analyzer 3 is connected to a second controller 6, and the first controller 5 and the second controller 6 are both connected to a power supply 7. The light source 1 sequentially passes through the polarizer 2, the analyzer 3, and the CMOS camera 4. The polarizer 2 includes a first polarizer 2-1 and a first quarter-wave plate 2-2. The first polarizer 2-1 is arranged close to the light source 1. The first controller 5 is used to control the rotation of the first quarter-wave plate 2-2. The analyzer 3 includes a second polarizer 3-1 and a second quarter wave plate 3-2. The second polarizer 3-1 is arranged close to the CMOS camera 4. The second controller 6 is used to control the rotation of the second quarter wave plate 3-2. During the acquisition process, the target tea leaf 8 is placed between the polarizer 2 and the analyzer 3. The light source 1 emits parallel non-polarized light with stable light intensity, which is incident on the first polarizer 2-1. The first quarter-wave plate 2-2 rotates and modulates the non-polarized light into a specific polarization state light before emitting. After the polarized light is irradiated on the surface of the target tea leaf, the light after interaction with the target tea leaf passes through the second quarter-wave plate 3-2, and finally passes through the second polarizer 3-1 and is received by the CMOS camera 4. The first quarter-wave plate 2-2 rotates 6° each time, and the second quarter-wave plate 3-2 rotates 30° in the same direction. 30 rotations are performed to obtain 30 target tea leaf images. Step 1.2, record the Mueller matrix of the target tea leaf as , the Mueller matrix of the first polarizer 2-1 is recorded as , the Mueller matrix of the first quarter wave plate 2-2 is recorded as , the Mueller matrix of the second quarter wave plate 3-2 is recorded as , the Mueller matrix of the second polarizer 3-1 is recorded as , the Stokes vectors of the incident light and the outgoing light are respectively recorded as and , then: (1) What the CMOS camera 4 receives is the light intensity signal, i.e. The first component After expansion: (2) In formula (2), is the Mueller matrix element of the target tea leaf, is a matrix u Array element in; Among them, the matrix u for: (3) According to formula (3), formula (2) can be rewritten into Fourier series form, and we have: (4) According to the Fourier series expansion, the corresponding Fourier coefficients are obtained 、 , a total of 25, solve the 16 array elements of the Mueller matrix from the 25 Fourier coefficients, the expression is: (5) Step 1.3, extract the grayscale values ​​of the pixels at the same position in the 30 target tea leaf images collected in step 1.1 to form a column matrix with 30 columns and one row. I (30 times of receiving the column matrix composed of light intensity), through I Solve for 25 Fourier coefficients, the expression is: (6) In formula (6), D It is a 30×15 matrix; F is a column matrix consisting of 25 Fourier coefficients to be solved; Equation (6) belongs to an overdetermined system of equations, which is usually solved using a pseudo-inverse matrix. The specific expression is as follows: (7) Then get F Substitute the array element in formula (5) to obtain the Mueller matrix of the current pixel point M ,like Figure 2 As shown; Step 1.4, repeat step 1.3 to obtain the Mueller matrix of pixels at different positions in 30 target tea leaf images M , all Mueller matrices M The values ​​of the array elements at the same position in the image are combined according to the corresponding pixel positions to obtain 16 Mueller matrix array element images, thereby obtaining a Mueller matrix image.

[0018] Example 3 Based on Example 2, in step 3, the polarization purity index of the target tea leaf is calculated. 、 、 and the overall polarization purity parameter PI The expression is: (8) (9) In formula (8), tr is the trace of the matrix, which is the sum of the main diagonal elements of the matrix; Polarization purity index Images such as Figure 3 As shown, the polarization purity index Images such as Figure 4 As shown, polarization purity index Images such as Figure 5 As shown, the overall polarization purity parameter PI Images such as Figure 6 shown.

[0019] Example 4 Based on Example 3, the specific process of step 4 is as follows: like Figure 7 As shown, the polarization purity index Mapped to the red channel, this parameter describes the degree of depolarization of the sample and the overall polarization purity parameter PI Mapped to the green channel, this parameter can effectively characterize the purity of the scattering system, and the blue channel is set to 0 to obtain a polarization fusion image.

[0020] Example 5 Based on Example 4, the specific process of step 5 is as follows: Step 5.1: Compress the grayscale of the polarization fusion image and select any point in the compressed polarization fusion image. O , the coordinates are , and select another point that deviates from this point Q , the coordinates are , a 、 b is an integer, and the point O and point Q As a pixel pair, since point Q Select from the entire compressed polarization fusion image, then a 、 b The values ​​are different, so that several pixel pairs can be formed and the grayscale value of each pixel pair is recorded. , assuming that the maximum grayscale of the compressed polarization fusion image is L ,So and Total combination of L × L , traverse the entire compressed polarization fusion image, and count each The frequency of occurrence of the value, and arranged in order into a L × L The upper left corner of the matrix is ​​the frequency of (1,1), and the lower right corner is ( L , L ), then the rows and columns of the matrix are 1, 2, …, L , the elements in the matrix are each Frequency of occurrence; Among them, the grayscale level is compressed to 8 or 16 levels, then L 8 or 16; Step 5.2, divide each element in the matrix by the sum of all elements to get each Probability of occurrence , the elements in the matrix are represented by the probability corresponding to each element Replacement, the square matrix after replacement is the gray level co-occurrence matrix; like Figure 8 As shown, if the coordinates of a pixel point are determined to be , when the parameter a 、 b When the value is 0 or 1, we can get 、 、 and 4 pixels, which correspond to Pixels at 0°, 90°, 45° and 135° with a spacing of 1. These pixels are They constitute pixel pairs in different directions respectively; Take the 0° direction as an example to calculate the gray level co-occurrence matrix. The maximum gray level of the pixel in this image is 6. The gray level co-occurrence matrix constructed is a 6×6 square matrix. Appears once, pixel pairs It appears twice, so the element value of the first row and first column in the gray-level co-occurrence matrix is ​​1, and the element value of the first row and second column is 2. According to the same rule, the values ​​of the remaining matrix elements can be determined; Step 5.3, calculate contrast, energy, inverse moment and entropy through gray level co-occurrence matrix, the expression is: (10) (11) (12) (13) In formulas (10) to (13), Represents the gray level co-occurrence matrix i Rank j Elements of the column; Contrast is contrast, Energy is energy, IDM is inverse difference moment, and Entropy is entropy.

[0021] Example 6 Based on Example 5, the specific process of step 6 is as follows: In step 6.1, build an SVM (support vector machine) classifier based on the MATLAB R2021b platform. The SVM classifier uses the linear kernel function, the standard SVM loss function (hinge loss function), the penalty term type is L2 regularization, and the regularization parameter C is 1. In step 6.2, the dataset was divided into five mutually exclusive subsets using a 5-fold cross-validation strategy. In each subset, the ratio of healthy leaves to infected leaves was 1:1. One subset was used as the test set, and the remaining four subsets were combined as the training set. The training set was input into the support vector machine in step 6.1 for training to obtain a trained support vector machine. Step 6.3: Input the test set into the trained support vector machine for verification; Accuracy, precision, and recall are selected as performance indicators for evaluating support vector machines. The calculation formula is as follows: (14) (15) (16); In formulas (14) to 16, TP is a true example (the number of samples that are actually positive and predicted by SVM to be positive), TN is a true negative example (the number of samples that are actually negative and predicted by SVM as negative), FP The number of false positives that are actually negative and that SVM mistakenly predicts as positive (also known as "false positives"), FN False negative examples (the number of samples that are actually positive but are mistakenly predicted as negative by SVM (also known as "missing reports")); The polarization image acquisition method can reflect the polarization imaging detection accuracy at a single pixel. Currently, commonly used methods based on light intensity polarization detection only detect the average light intensity of the target, and methods based on spectral polarization detection obtain the total spectral characteristic curve of the target. Although these methods can show the overall trend of the target to a certain extent, they cannot detect the changes in the microscopic structure of the target. The Mueller matrix image and polarization-related parameter image obtained by the method of the present invention can clearly show the area and range of changes in the microscopic structure of the target, and can obtain more information about the target.

Claims

1. A method for identifying tea stink bug eggs using polarized light imaging technology, characterized in that: The specific steps are as follows: Step 1, collecting an image of a target tea leaf through a polarized light imaging system, and calculating a Mueller matrix image of the target tea leaf; Step 2: Calculate the covariance matrix of the Mueller matrix through the Mueller matrix image H , and obtain its eigenvalues ​​from the covariance matrix 、 、 、 ; Step 3: According to the eigenvalue 、 、 、 Calculate the polarization purity index of the target tea leaves 、 、 and the overall polarization purity parameter PI ; Step 4: and PI Perform fusion to obtain a polarization fusion image; Step 5, calculate the gray level co-occurrence matrix of the polarization fusion image and calculate the contrast, energy, inverse moment and entropy; Step 6: training the support vector machine to obtain a trained support vector machine; Step 7: Input the contrast, energy, inverse moment and entropy into the trained support vector machine to detect the eggs of the tea net stink bug.

2. The method for identifying tea stink bug eggs using polarized light imaging technology according to claim 1, wherein: The specific process of step 1 is: Step 1.1, collecting images of target tea leaves using a polarized light imaging system; Step 1.2, record the Mueller matrix of the target tea leaf as , the Mueller matrix of the first polarizer (2-1) is recorded as , the Mueller matrix of the first quarter wave plate (2-2) is recorded as , the Mueller matrix of the second quarter wave plate (3-2) is recorded as , the Mueller matrix of the second polarizer (3-1) is recorded as , the Stokes vectors of the incident light and the outgoing light are respectively recorded as and , then: （1） The CMOS camera (4) receives the light intensity signal, i.e. The first component After expansion: （2） In formula (2), is the Mueller matrix element of the target tea leaf, is a matrix u Array element in; Among them, the matrix u for: （3） According to formula (3), formula (2) can be rewritten into Fourier series form, and we have: （4） According to the Fourier series expansion, the corresponding Fourier coefficients are obtained 、 , a total of 25, solve the 16 array elements of the Mueller matrix from the 25 Fourier coefficients, the expression is: （5） Step 1.3, extract the grayscale values ​​of the pixels at the same position in the 30 target tea leaf images collected in step 1.1 to form a column matrix with 30 columns and one row. I ,pass I Solve for 25 Fourier coefficients, the expression is: （6） In formula (6), D It is a 30×15 matrix; F is a column matrix consisting of 25 Fourier coefficients to be solved; Equation (6) belongs to an overdetermined system of equations, which is usually solved using a pseudo-inverse matrix. The specific expression is as follows: （7） Then get F Substitute the array element in formula (5) to obtain the Mueller matrix of the current pixel point M ; Step 1.4, repeat step 1.3 to obtain the Mueller matrix of pixels at different positions in 30 target tea leaf images M , all Mueller matrices M The values ​​of the array elements at the same position in the image are combined according to the corresponding pixel positions to obtain 16 Mueller matrix array element images, thereby obtaining a Mueller matrix image.

3. The method for identifying tea stink bug eggs using polarized light imaging technology according to claim 2, wherein: The specific process of step 1.1 is: The polarized light imaging system comprises a light source (1), a polarizer (2), an analyzer (3), and a CMOS camera (4); the polarizer (2) is connected to a first controller (5); the analyzer (3) is connected to a second controller (6); and both the first controller (5) and the second controller (6) are connected to a power supply (7); The polarizer (2) comprises a first polarizer (2-1) and a first quarter-wave plate (2-2), wherein the first polarizer (2-1) is arranged close to the light source (1); The polarizer (3) includes a second polarizer (3-1) and a second quarter-wave plate (3-2), and the second polarizer (3-1) is arranged close to the CMOS camera (4); During the acquisition process, the target tea leaf is placed between the polarizer (2) and the analyzer (3), and the light source (1) emits parallel non-polarized light with stable light intensity, which is incident on the first polarizer (2-1). The first quarter-wave plate (2-2) rotates and modulates the non-polarized light into a specific polarization state light before emitting it. After the polarized light is irradiated on the surface of the target tea leaf, the light that interacts with the target tea leaf passes through the second quarter-wave plate (3-2) and finally passes through the second polarizer (3-1) and is received by the CMOS camera (4). The first quarter-wave plate (2-2) rotates 6° each time, and the second quarter-wave plate (3-2) rotates 30° in the same direction. 30 rotations are performed to obtain 30 images of the target tea leaf.

4. The method for identifying tea stink bug eggs using polarized light imaging technology according to claim 1, wherein: In step 3, the polarization purity index of the target tea leaf is calculated 、 、 The expression is: （8）； In formula (8), tr is the trace of the matrix, which is the sum of the elements on the main diagonal of the matrix.

5. The method for identifying tea stink bug eggs using polarized light imaging technology according to claim 1, wherein: In step 3, calculate the overall polarization purity parameter PI The expression is: （9）。 6. The method for identifying tea stink bug eggs using polarized light imaging technology according to claim 1, wherein: The specific process of step 4 is: polarization purity index Mapped to the red channel, the overall polarization purity parameter PI Map to the green channel and set the blue channel to 0 to obtain the polarization fused image.

7. The method for identifying tea stink bug eggs using polarized light imaging technology according to claim 1, wherein: The specific process of step 5 is: Step 5.1: Compress the grayscale of the polarization fusion image and select any point in the compressed polarization fusion image. O , the coordinates are , and select another point that deviates from this point Q , the coordinates are , a 、 b is an integer, and the point O and point Q As a pixel pair, since point Q Select from the entire compressed polarization fusion image, then a 、 b The values ​​are different, so that several pixel pairs can be formed and the grayscale value of each pixel pair is recorded. , assuming that the maximum grayscale of the compressed polarization fusion image is L ,So and Total combination of L × L , traverse the entire compressed polarization fusion image, and count each The frequency of occurrence of the value, and arranged in order into a L × L The upper left corner of the matrix is ​​the frequency of (1,1), and the lower right corner is ( L , L ), then the rows and columns of the matrix are 1, 2, …, L , the elements in the matrix are each Frequency of occurrence; Among them, the grayscale level is compressed to 8 or 16 levels, then L 8 or 16; Step 5.2, divide each element in the matrix by the sum of all elements to get each Probability of occurrence , the elements in the matrix are represented by the probability corresponding to each element Replacement, the square matrix after replacement is the gray level co-occurrence matrix; Step 5.3, calculate contrast, energy, inverse moment and entropy through gray level co-occurrence matrix, the expression is: （10） （11） （12） （13） In formulas (10) to (13), Represents the gray level co-occurrence matrix i Rank j Elements of the column; Contrast is contrast, Energy is energy, IDM is inverse difference moment, and Entropy is entropy.

8. The method for identifying tea stink bug eggs using polarized light imaging technology according to claim 1, wherein: The specific process of step 6 is: In step 6.1, build an SVM classifier based on the MATLAB R2021b platform. The SVM classifier selects the linear kernel function, the loss function adopts the SVM standard loss function, the penalty term type is L2 regularization, and the value of the regularization parameter C is 1. Step 6.2: Use a 5-fold cross-validation strategy to divide the data set into a training set and a test set. The training set is input into the support vector machine in step 6.1 for training to obtain a trained support vector machine. In step 6.3, the test set is input into the trained support vector machine for verification.

Images