A method for detecting the damage level of tea brown spot based on hyperspectral technology

Abstract

The application provides a kind of tea round brown spot hazard grade detection method based on hyperspectral technology, including obtaining the original hyperspectral data of tea in tea garden, and the hyperspectral technology has high resolution, can quickly obtain the continuous wave band hyperspectral reflectivity information of tea in tea garden.The original hyperspectral data is pretreated to obtain pretreated data, and the pretreatment includes filtering and first-order derivation.The first-order derivation can amplify the detail information and noise in the spectrum, so the original hyperspectral data is filtered first, and then the filtered data is first-order derived, so as to eliminate baseline drift and reduce the interference caused by background area, so that the pretreated data has high resolution.A tea round brown spot detection model is constructed, the pretreated data is input into the tea round brown spot detection model, the hazard grade is judged, and the tea round brown spot hazard grade is obtained, which can effectively detect the tea round brown spot hazard grade and has high recognition accuracy.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of tea tree disease detection, and particularly relates to a tea cercospora leaf spot disease damage grade detection method based on hyperspectral technology. BACKGROUND

[0002] Tea cercospora leaf spot disease is one of common leaf diseases of tea trees, and diseased leaves appear in patches, has local infectivity, and mainly occurs in tea gardens with high humidity and fog. The occurrence probability of tea cercospora leaf spot disease is relatively high in early spring, and the diseased tea leaves are purple-brown, a large number of diseased leaves are shed, the tea made of the diseased leaves has high breakage rate, the soup color is dark, and the taste is bitter and astringent. Tea cercospora leaf spot disease has a great influence on tea quality.

[0003] The traditional identification method of tea cercospora leaf spot disease mainly relies on manual identification according to characteristic symptoms or pathological examination, and secondly, images are used for effective identification and disease estimation of crop diseases, but the above method cannot well distinguish the occurrence period of tea cercospora leaf spot disease.

[0004] Therefore, a detection method for detecting different damage grades of tea cercospora leaf spot disease and having high recognition accuracy is needed. SUMMARY

[0005] To overcome the problems in the related art, the purpose of the present application is to provide a tea cercospora leaf spot disease damage grade detection method based on hyperspectral technology, which combines hyperspectral technology and a convolutional neural network, and introduces a gated recurrent module in the convolutional neural network, so as to effectively detect the damage grade of tea cercospora leaf spot disease and has high recognition accuracy.

[0006] A tea cercospora leaf spot disease damage grade detection method based on hyperspectral technology, comprising:

[0007] Obtaining original hyperspectral data of tea leaves in a tea garden; wherein the tea leaves in the tea garden include tea cercospora leaf spot disease and healthy tea leaves;

[0008] Pretreating the original hyperspectral data to obtain pretreated data; the pretreatment includes filtering and first-order derivation;

[0009] Constructing a tea cercospora leaf spot disease detection model, inputting the pretreated data into the tea cercospora leaf spot disease detection model, judging the damage grade, and obtaining the damage grade of tea cercospora leaf spot disease; the tea cercospora leaf spot disease detection model is based on a convolutional neural network and a gated recurrent module.

[0010] In the preferred technical solution of the present application, the pretreatment of the original hyperspectral data to obtain pretreated data comprises:

[0011] The original hyperspectral data is subjected to standard normal transformation to obtain transformed spectral data;

[0012] The transformed spectral data is filtered to obtain filtered data;

[0013] The filtered data is subjected to first-order derivation to obtain pretreated data.

[0014] In the preferred technical solution of the present application, the standard normal transformation of the original hyperspectral data to obtain transformed spectral data comprises:

[0015] The original hyperspectral data is subjected to standard normal transformation according to the following formula:

[0016]

[0017] wherein, x i is the average value of spectral reflectivity of the i-th sample, x i,k is the spectral reflectivity of the k-th spectral band of the i-th sample; m is the total number of spectral bands of the i-th sample, X i,SNV is the spectral data of the i-th sample after transformation, 1≤i≤n, n is the total number of samples;

[0018] The spectral data of all transformed samples is combined to obtain the transformed spectral data.

[0019] In the preferred technical solution of the present application, the filtering of the transformed spectral data to obtain filtered data comprises:

[0020] The transformed spectral data is subjected to SG filtering according to the following formula:

[0021]

[0022] wherein, a j is the j-th polynomial coefficient, r is the size of the SG window, m1 is the reference coefficient, X i,SG is the i-th spectral data after SG filtering;

[0023] All spectral data after SG filtering is combined to obtain the filtered data.

[0024] In the preferred technical solution of the present application, the first-order derivation of the filtered data to obtain pretreated data comprises:

[0025] The filtered data is subjected to first-order derivation according to the following formula:

[0026]

[0027] wherein, is the wavelength λ i+1 reflectance after standard normal variate transformation and SG filtering, is the wavelength λ i-1 reflectance after standard normal variate transformation and SG filtering, ΔX is the wavelength λ i-1 to the wavelength λ i+1 wavelength interval, X'(X i，SG ) is the preprocessed data.

[0028] In the preferred technical scheme of the present application, the original hyperspectral data of tea in the tea garden is obtained, comprising:

[0029] The tea leaves suffering from tea brown spot and the healthy tea leaves are obtained.

[0030] The spectral reflectance data of the tea leaves suffering from tea brown spot and the healthy tea leaves are collected by using a hyperspectral imager to obtain the original hyperspectral data of tea in the tea garden; wherein the spectral band range of the hyperspectral imager is 400-1000nm.

[0031] In the preferred technical scheme of the present application, the tea brown spot detection model is constructed, comprising:

[0032] The CNN-GRU neural network is used as the network architecture of the tea brown spot detection model.

[0033] The 5 one-dimensional convolution layers are used as the feature extraction module of the tea brown spot detection model.

[0034] The input end of the feature extraction module is connected with the input module, and the output end of the feature extraction module is connected with the input end of the global average pooling layer.

[0035] In the preferred technical scheme of the present application, after the output end of the feature extraction module is connected with the input end of the global average pooling layer, further comprising:

[0036] The output end of the global average pooling layer is connected with the input end of the gated recurrent module.

[0037] The output end of the gated recurrent module is connected with the input end of the fully connected layer.

[0038] The output end of the fully connected layer is connected with the input end of the activation layer; and the output end of the activation layer outputs the hazard grade of the tea brown spot.

[0039] In the preferred technical scheme of the present application, the gated recurrent module comprises two GRU layers and a dropout layer, wherein the output end of one GRU layer is connected with the input end of the dropout layer, and the output end of the dropout layer is connected with the input end of the other GRU layer.

[0040] In the preferable technical solution of the present application, the tea cercospora leaf spot damage level includes healthy, mild, moderate and severe, wherein the healthy corresponds to 0 of the number of disease spots, the mild corresponds to 1-30 of the number of disease spots, the moderate corresponds to 31-60 of the number of disease spots, and the severe corresponds to more than 60 of the number of disease spots.

[0041] The present application has the following beneficial effects:

[0042] The tea cercospora leaf spot damage level detection method based on the hyperspectral technology provided by the present application includes obtaining original hyperspectral data of tea leaves in a tea garden; wherein the tea leaves in the tea garden include tea leaves suffering from tea cercospora leaf spot and healthy tea leaves. The hyperspectral technology has high resolution and can quickly obtain continuous waveband hyperspectral reflectivity information of the tea leaves in the tea garden. The original hyperspectral data is preprocessed to obtain preprocessed data, and the preprocessing includes filtering and first-order derivation. The first-order derivation can amplify the detailed information in the spectrum, while also amplifying the noise, so when the spectrum is derived, the original hyperspectral data is filtered first, and then the filtered data is first-order derived, so as to eliminate the baseline drift and reduce the interference from the background area, so that the preprocessed data has high resolution. A tea cercospora leaf spot detection model is constructed, the preprocessed data is input into the tea cercospora leaf spot detection model, the damage level is judged, the tea cercospora leaf spot damage level is obtained, and the tea cercospora leaf spot detection model is based on a convolutional neural network and a gated recurrent module. The convolutional neural network can automatically learn the deep features of the preprocessed data and can be used for detecting the tea cercospora leaf spot damage level of the tea leaves. The gated recurrent module is used to solve the gradient disappearance problem of the convolutional neural network. Combined with the hyperspectral technology, the convolutional neural network and the gated recurrent module, the tea cercospora leaf spot damage level can be effectively detected, and the recognition accuracy is high. BRIEF DESCRIPTION OF DRAWINGS

[0043] Figure 1 is a flowchart of the tea cercospora leaf spot damage level detection method based on the hyperspectral technology of the present application;

[0044] Figure 2 is a flowchart of the preprocessing of the original hyperspectral data of the present application;

[0045] Figure 3 is a flowchart of the construction of the tea cercospora leaf spot detection model of the present application;

[0046] Figure 4 is a composition diagram of the tea cercospora leaf spot detection model of the present application. DETAILED DESCRIPTION

[0047] The preferred embodiments of the present application will be described in more detail with reference to the drawings. Although the preferred embodiments of the present application are shown in the drawings, it is understood that the present application can be implemented in various forms and should not be limited by the embodiments set forth herein. On the contrary, these embodiments are provided so that the present application is more thorough and complete, and the scope of the present application is fully conveyed to those skilled in the art.

[0048] Embodiment 1

[0049] As Figure 1 shown, the present embodiment provides a tea brown spot disease damage grade detection method based on hyperspectral technology, comprising:

[0050] S1: obtaining the original hyperspectral data of tea leaves in a tea garden; wherein the tea leaves in the tea garden include tea leaves suffering from tea brown spot disease and healthy tea leaves.

[0051] S2: preprocessing the original hyperspectral data to obtain preprocessed data; the preprocessing includes filtering and first-order derivation.

[0052] S3: constructing a tea brown spot disease detection model, inputting the preprocessed data into the tea brown spot disease detection model, judging the damage grade, and obtaining the tea brown spot disease damage grade; the tea brown spot disease detection model is based on convolutional neural network and gated recurrent module.

[0053] Step S1 includes the following steps S11-S12:

[0054] S11: obtaining the tea leaves suffering from tea brown spot disease and the healthy tea leaves.

[0055] S12: using a hyperspectral imager to collect spectral reflectance data of the tea leaves suffering from tea brown spot disease and the healthy tea leaves, and obtaining the original hyperspectral data of the tea leaves in the tea garden; wherein the spectral band range of the hyperspectral imager is 400-1000nm.

[0056] In the present embodiment, HyperSIS-VNIR-QE hyperspectral imager is used to collect spectral reflectance, and hyperspectral image is drawn based on spectral reflectance. The hyperspectral image is converted into a gray-scale image, the gray-scale image is corrected by black and white board, and the image corrected by black and white board is used as the original hyperspectral data. The original hyperspectral data includes the ROI region corresponding to the leaf in a single tea leaf sample.

[0057] The average spectral reflectance in the ROI region is calculated, and the original hyperspectral data is preprocessed according to the average spectral reflectance to obtain preprocessed data. The preprocessing includes filtering and first-order derivation. The filtering can eliminate the noise in the ROI region, the first-order derivation can eliminate the baseline drift and reduce the influence of background interference, and the resolution of the ROI region is improved.

[0058] The pre-processed data after step S2 is divided into a training set, a validation set and a test set in a ratio of 3:1:1. The training set, the validation set and the test set are sequentially taken as input variables, and the tea brown spot disease damage level is taken as an output variable. In this embodiment, the pre-processed data contains the ROI region corresponding to the leaf of 1188 tea samples.

[0059] The tea brown spot disease detection model is based on a neural network architecture, and a gating cycle module is introduced in the neural network architecture. The training set is input into the tea brown spot disease detection model, and the tea brown spot disease detection model is trained. The validation set is input into the trained tea brown spot disease detection model, and the parameters of the trained tea brown spot disease detection model are adjusted. The test set is input into the tea brown spot disease detection model with adjusted parameters, and the tea brown spot disease damage level corresponding to the test set is obtained. Based on the tea brown spot disease damage level corresponding to the test set and the true tea brown spot disease damage level, the accuracy, precision, recall and f1 score are calculated, and the tea brown spot disease detection model is evaluated according to the accuracy, precision, recall and f1 score.

[0060] The tea brown spot disease damage level detection method based on the hyperspectral technology provided in this embodiment includes obtaining original hyperspectral data of tea leaves in a tea garden; wherein the tea leaves in the tea garden include tea leaves suffering from tea brown spot disease and healthy tea leaves. The hyperspectral technology has high resolution and can quickly obtain continuous waveband hyperspectral reflectivity information of the tea leaves in the tea garden. The original hyperspectral data is pre-processed to obtain pre-processed data, and the pre-processing includes filtering and first-order derivation. The first-order derivation can amplify the detailed information in the spectrum, but at the same time, it can also amplify the noise. Therefore, when the spectrum is derived, the original hyperspectral data is first filtered, and then the filtered data is first-order derived, so as to eliminate baseline drift, reduce interference from the background area, and make the pre-processed data have high resolution. A tea brown spot disease detection model is constructed, the pre-processed data is input into the tea brown spot disease detection model, the damage level is judged, and the tea brown spot disease damage level is obtained. The tea brown spot disease detection model is based on a convolutional neural network and a gating cycle module. The convolutional neural network can automatically learn the deep features of the pre-processed data, and can be used for detecting the tea brown spot disease damage level of the tea leaves. The gating cycle module is used to solve the problem of gradient disappearance existing in the convolutional neural network. Combined with the hyperspectral technology, the convolutional neural network and the gating cycle module, the tea brown spot disease damage level can be effectively detected, and the recognition accuracy is high.

[0061] Embodiment 2

[0062] This embodiment provides a tea brown spot disease damage level detection method based on a hyperspectral technology. This embodiment only describes the differences from embodiment 1, such as Figure 2The pre-processing of the original hyperspectral data comprises:

[0063] S21: performing standard normal transformation on the original hyperspectral data to obtain transformed spectral data.

[0064] S22: filtering the transformed spectral data to obtain filtered data.

[0065] S23: performing first-order derivation on the filtered data to obtain pre-processed data.

[0066] The standard normal transformation on the original hyperspectral data to obtain transformed spectral data comprises:

[0067] The original hyperspectral data is subjected to standard normal transformation according to the following formula:

[0068]

[0069] wherein x i is the average value of spectral reflectivity of the i-th sample, x i,k is the spectral reflectivity of the k-th spectral band of the i-th sample; m is the total number of spectral bands of the i-th sample, X i,SNV is the spectral data of the i-th sample after transformation, 1≤i≤n, n is the total number of samples;

[0070] The spectral data of all the transformed samples are combined to obtain the transformed spectral data.

[0071] Each sample is represented as a spectral curve at the spectral level, and the spectral curve is divided into multiple spectral bands during collection, each spectral band corresponding to a spectral reflectivity. The standard normal transformation on the original hyperspectral data makes the transformed spectral data satisfy the characteristics of normal distribution, and the normal distribution has the advantages of concentration, symmetry, stability and independence, and has a high fitting degree for real data.

[0072] The filtering of the transformed spectral data to obtain filtered data comprises:

[0073] The transformed spectral data is subjected to SG filtering according to the following formula:

[0074]

[0075] wherein a j is the j-th polynomial coefficient, r is the size of the SG window, m1 is the reference coefficient, X i,SG is the i-th spectral data after SG filtering;

[0076] Combining all the SG filtered spectral data sets to obtain filtered data.

[0077] The SG filtering has the following advantages: (1) the SG filtering can effectively reduce the noise while keeping the data trend and shape; (2) the SG filter can flexibly adjust the order of the fitting polynomial, so it can adapt to different types of data; (3) the SG filter has faster computing speed and lower storage requirements.

[0078] The first-order derivative of the filtered data is obtained to obtain preprocessed data, including:

[0079] The first-order derivative of the filtered data is obtained according to the following formula:

[0080]

[0081] wherein, is the wavelength λ i+1 The reflectivity after standard normal transformation and SG filtering, is the wavelength λ i-1 The reflectivity after standard normal transformation and SG filtering, ΔX is the wavelength interval from the wavelength λ i-1 to the wavelength λ i+1 X'(X i，SG ) is the preprocessed data.

[0082] The purpose of the first-order derivative is to eliminate the baseline drift and reduce the influence of background interference, and to provide higher resolution. The reflectivity change is the baseline drift caused by the spectral acquisition environment interference, and the first-order derivative can amplify the detailed information in the spectrum, while also amplifying the noise. Therefore, the transformed spectral data is filtered first to remove the noise in the transformed spectral data, and then the filtered data is first-order derived to obtain the preprocessed data.

[0083] The original hyperspectral data is preprocessed to obtain preprocessed data, including standard normal transformation of the original hyperspectral data to obtain transformed spectral data. The transformed spectral data is filtered to obtain filtered data. The first-order derivative of the filtered data is obtained to obtain the preprocessed data. The standard normal transformation makes the transformed spectral data more consistent with the distribution characteristics of the real data, the filtering can remove the noise of the transformed spectral data, and the first-order derivative can eliminate the baseline drift, reduce the influence of background interference, and provide higher resolution.

[0084] Example 3

[0085] As Figure 3As shown, the embodiment provides a tea round brown spot damage grade detection method based on hyperspectral technology, and the embodiment only describes the differences from embodiment 1. The tea round brown spot detection model is constructed, including:

[0086] S31: adopting a CNN-GRU neural network as the network architecture of the tea round brown spot detection model.

[0087] S32: adopting a 5-layer one-dimensional convolutional layer as the feature extraction module of the tea round brown spot detection model.

[0088] S33: connecting the input end of the feature extraction module with the input module and connecting the output end of the feature extraction module with the input end of the global average pooling layer.

[0089] After connecting the output end of the feature extraction module with the input end of the global average pooling layer, the embodiment further includes:

[0090] S34: connecting the output end of the global average pooling layer with the input end of the gated recurrent module.

[0091] S35: connecting the output end of the gated recurrent module with the input end of the fully connected layer.

[0092] S36: connecting the output end of the fully connected layer with the input end of the activation layer; and the output end of the activation layer outputs the tea round brown spot damage grade.

[0093] Figure 4 The gated recurrent module includes two GRU layers and a dropout layer, wherein the output end of one GRU layer is connected with the input end of the dropout layer, and the output end of the dropout layer is connected with the input end of the other GRU layer. Figure 4 In the figure, CNN represents a convolutional neural network, Fully connected layer represents a fully connected layer, Softmax represents a normalization exponential function, Conv.Block1 represents a first convolutional layer, and Dropout represents a random dropout layer.

[0094] The CNN-GRU neural network of the embodiment has the following characteristics:

[0095] (1) The size of the input spectral data is 256*1, that is, 256 bands and 1 channel.

[0096] (2) A 5-layer one-dimensional convolutional layer is adopted to extract feature values, the number of convolution kernels is 16, the size is 1*5, and the stride is 2. The spectral feature extraction operation is performed through the convolutional layer to mine the correlation between data.

[0097] (3) Adding a global average pooling layer to compress the number of parameters, reduce the complexity of the model, and improve the generalization ability.

[0098] (4) Adding a Dropout layer to randomly select neurons in the network to prevent overfitting.

[0099] (5) Using a GRU layer to process the reduced dimension data and learn the dependent relationship between different wavebands of spectral data.

[0100] (6) Flattening the data into a one-dimensional structure by a fully connected layer, and matching each feature data to the feature category with the maximum probability by a Softmax activation function.

[0101] Table 1 Classification standard of tea round brown spot

[0102] Standard Grade Number Healthy 0 285 Mild (1-30 number of lesion) 1 289 Moderate (31-60 number of lesion) 2 293 Severe (>60 number of lesion) 3 321

[0103] As shown in Table 1, the tea round brown spot damage grade of the present embodiment includes healthy, mild, moderate and severe, wherein healthy corresponds to 0 lesion spots, mild corresponds to 1-30 lesion spots, moderate corresponds to 31-60 lesion spots, and severe corresponds to more than 60 lesion spots.

[0104] In the present embodiment, since the content of chlorophyll in the leaf of tea round brown spot is more sensitive to the damage grade of tea round brown spot, in the two spectral bands with center wavelengths of 450 nm (blue light) and 650 nm (red light), chlorophyll absorbs most of the incident energy. Between the two chlorophyll absorption bands, a reflection peak is formed near 540 nm (green light) due to the smaller absorption. The gating cycle module includes multiple GRU layers (gating cycle layers), which can better capture the dependent relationship between different wavebands when the waveband gap is large. The GRU layer is introduced into the convolutional neural network to establish a tea round brown spot detection model that can output different tea round brown spot damage grades.

[0105] Embodiment 4

[0106] In the present embodiment, a computer device is also provided, which can be a server, wherein the computer device includes a processor, a memory, a network interface and a database connected by a system bus. The processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The network interface of the computer device is used to communicate with the external terminal through the network connection.

[0107] The embodiment also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the tea cercospora damage grade detection method based on the hyperspectral technology. It can be understood that the computer readable storage medium in the embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0108] It should be noted that in this document, the terms "comprising", "including", or any other variant thereof are intended to cover a non-exclusive inclusion, so that a process, device, article or method that includes a list of elements not only includes those elements, but also includes other elements not expressly listed, or inherent to such a process, device, article or method. Without more limitations, the element defined by the statement "comprising a" does not exclude the presence of additional identical elements in the process, device, article or method that includes the element.

[0109] The above only describes the preferred embodiments of the present application, and does not limit the patent scope of the present application, and any equivalent structure or equivalent process transformation using the content of the specification and drawings, or direct or indirect application in other related technical fields, are also included in the patent protection scope of the present application.

Claims

1. A method for detecting the damage level of tea cercospora leaf spot based on hyperspectral technology, characterized in that, The application relates to a tea brown spot disease detection method and device. The application comprises the following steps: Obtaining original hyperspectral data of tea leaves in a tea garden; wherein the tea leaves in the tea garden include tea leaves suffering from tea brown spot disease and healthy tea leaves; Preprocessing the original hyperspectral data to obtain preprocessed data; the preprocessing includes filtering and first-order derivation; Constructing a tea brown spot disease detection model, inputting the preprocessed data into the tea brown spot disease detection model, performing hazard grade judgment, and obtaining a tea brown spot disease hazard grade; the tea brown spot disease detection model is based on a convolutional neural network and a gated recurrent module; The application comprises the following steps: Using a CNN-GRU neural network as the network architecture of the tea brown spot disease detection model; Using five one-dimensional convolutional layers as a feature extraction module of the tea brown spot disease detection model; Connecting an input end of the feature extraction module with an input module and connecting an output end of the feature extraction module with an input end of a global average pooling layer; After connecting the output end of the feature extraction module with the input end of the global average pooling layer, the following steps are further included: Connecting an output end of the global average pooling layer with an input end of a gated recurrent module; Connecting an output end of the gated recurrent module with an input end of a full connection layer; Connecting an output end of the full connection layer with an input end of an activation layer; an output end of the activation layer outputs a tea brown spot disease hazard grade; 2. The method for detecting the damage level of tea brown spot disease based on hyperspectral technology according to claim 1, characterized in that, The gated recurrent module comprises two GRU layers and a dropout layer; an output end of one of the GRU layers is connected with an input end of the dropout layer, and an output end of the dropout layer is connected with an input end of the other GRU layer. The application comprises the following steps: Performing standard normal transformation on the original hyperspectral data to obtain transformed spectral data; Performing filtering on the transformed spectral data to obtain filtered data; 3. The method for detecting the damage level of tea brown spot disease based on hyperspectral technology according to claim 2, characterized in that, Performing first-order derivation on the filtered data to obtain preprocessed data. The application comprises the following steps: ； wherein x i is the average value of the spectral reflectance of the i-th sample, is the spectral reflectance of the k-th spectral band of the i-th sample; m is the total number of spectral bands of the i-th sample, X i,SNV is the transformed spectral data of the i-th sample, 1≤i≤n, n is the total number of samples; Performing standard normal transformation on the original hyperspectral data according to the following formula:

4. The method for detecting the damage level of tea brown spot disease based on hyperspectral technology according to claim 3, characterized in that, Combining spectral data of all the transformed samples to obtain transformed spectral data. The application comprises the following steps: ； wherein a j is the jth polynomial coefficient, r is the size of the SG window, m1 is the reference coefficient, , X i,SG is the ith SG filtered spectral data; Performing SG filtering on the transformed spectral data according to the following formula:

5. The method for detecting the damage level of tea brown spot disease based on hyperspectral technology according to claim 2, characterized in that, Combining all the SG-filtered spectral data to obtain filtered data. The application comprises the following steps: ； wherein, is the wavelength Reflectance after standard normal variate transformation and SG filtering, is the wavelength Reflectance after standard normal variate transformation and SG filtering, is the wavelength to the wavelength interval, is the pre-processed data.

6. The method for detecting the damage level of tea brown spot disease based on hyperspectral technology according to claim 1, characterized in that, Performing first-order derivation on the filtered data according to the following formula: The application comprises the following steps: Obtaining the tea leaves suffering from tea brown spot disease and the healthy tea leaves; Using a hyperspectral imager to collect spectral reflectivity data of the tea leaves suffering from tea brown spot disease and the healthy tea leaves to obtain original hyperspectral data of the tea leaves in the tea garden; wherein the spectral band range of the hyperspectral imager is 400-1000 nm.

7. The method for detecting the damage level of tea brown spot disease based on hyperspectral technology according to claim 1, characterized in that, The tea round brown spot disease damage grade includes healthy, mild, moderate and severe, wherein the healthy corresponds to 0 of the lesion spot number, the mild corresponds to 1-30 of the lesion spot number, the moderate corresponds to 31-60 of the lesion spot number, and the severe corresponds to more than 60 of the lesion spot number.

Images