Tobacco bacterial wilt recognition method and device based on hyperspectrum, equipment and medium

Abstract

The application discloses a tobacco bacterial wilt recognition method and device based on hyperspectrum, equipment and medium, and the method comprises the following steps: acquiring tobacco hyperspectrum image data to be recognized; preprocessing the hyperspectrum image data; inputting the preprocessed data into a trained XGBoost model to obtain a tobacco bacterial wilt recognition result. The method can recognize the tobacco bacterial wilt in a safe, convenient, fast and nondestructive manner, and the accuracy and precision are high, the recognition efficiency is greatly improved, and the cost is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of hyperspectral image recognition, and in particular to a tobacco bacterial wilt recognition method and device based on hyperspectrum, equipment and medium. BACKGROUND

[0002] Tobacco bacterial wilt caused by Pseudomonas syringae pv. tabaci is a devastating soil-borne disease. Because the host range of Pseudomonas syringae is wide and difficult to control, it is called the "cancer" of plants. Therefore, early prevention and control is the key to reducing the damage of tobacco bacterial wilt.

[0003] The traditional tobacco bacterial wilt recognition method mainly relies on manual judgment based on its characteristic disease conditions and symptoms or pathological examination. Manual judgment is easy to introduce subjective factors, and the recognition accuracy and precision are affected. Pathological examination has the problems of long cycle, high equipment requirement, high cost and low recognition efficiency.

[0004] Hyperspectral technology has the characteristics of atlas integration, more and continuous wave bands, etc. The complete spectral curve is collected by a hyperspectral camera to describe the spectral characteristics of the ground object, and then the ground object with diagnostic spectral characteristics is recognized. Therefore, it has obvious advantages in crop nutrient diagnosis and disease and pest degree discrimination. The Extreme Gradient Boosting (XGBoost) model has excellent effect, simple use and fast speed for classification and regression problems. There is currently a blank in the application of XGBoost model in tobacco bacterial wilt recognition based on hyperspectrum. SUMMARY

[0005] The present application provides a tobacco bacterial wilt recognition method, device, equipment and medium based on hyperspectrum, which solves at least one technical problem in the prior art.

[0006] In a first aspect, the present application provides a tobacco bacterial wilt recognition method based on hyperspectrum, comprising:

[0007] Obtaining tobacco hyperspectral image data to be recognized;

[0008] Preprocessing the hyperspectral image data;

[0009] Inputting the preprocessed data into a trained XGBoost model to obtain a tobacco bacterial wilt recognition result.

[0010] The above technical solution can correctly recognize tobacco bacterial wilt by inputting the preprocessed tobacco hyperspectral image data into the XGBoost model trained by tobacco bacterial wilt data set, and has high recognition accuracy and precision.

[0011] Optionally, the preprocessing of the hyperspectral image data further comprises:

[0012] The ROI region is extracted from the hyperspectral image data of the tobacco to be identified, resulting in hyperspectral data containing only the ROI region.

[0013] Optionally, the extraction of the Region of Interest (ROI) from the hyperspectral image data of the tobacco to be identified further includes:

[0014] Extract the full-band raw images acquired by the hyperspectral camera;

[0015] A mask image is created by selecting a band image from the original image where the background and tobacco plants show significant differences.

[0016] The original image and the mask image are bitwise ANDed to obtain the ROI region.

[0017] Optionally, after extracting the ROI region from the hyperspectral image data of the tobacco to be identified, the process further includes:

[0018] The average reflectance of the ROI region is obtained based on the hyperspectral data and area of ​​the ROI region.

[0019] Optionally, after extracting the ROI region from the hyperspectral image data of the tobacco to be identified, the process further includes:

[0020] The first derivative of the average reflectance data of the ROI region for each band is calculated based on the following formula:

[0021]

[0022] Where, d i The first derivative of the i-band is represented by x. i+1 x is the average reflectivity of band i+1; i-1 The average reflectance of band i-1.

[0023] Optionally, the XGBoost model is trained based on the following objective function:

[0024]

[0025] Where L is the loss function; y i It is the true value of sample i; It is the model's predicted value after the (t-1)th iteration; f t (x i ) is the split node in the t-th iteration; Ω(f t ) is a regularization term.

[0026] Secondly, the present invention also provides a hyperspectral-based tobacco bacterial wilt identification device, comprising:

[0027] An acquisition module is configured to acquire tobacco hyperspectral image data to be identified;

[0028] A preprocessing module is configured to preprocess the hyperspectral image data;

[0029] An identification module is configured to input the preprocessed data into a trained XGBoost model to obtain a tobacco bacterial wilt identification result.

[0030] In a third aspect, the present application further provides an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and characterized in that the processor implements the hyperspectral-based tobacco bacterial wilt identification method according to the first aspect when executing the program.

[0031] In a fourth aspect, the present application further provides a computer-readable storage medium having a computer program stored thereon, and characterized in that the computer program is executable on a processor to implement the hyperspectral-based tobacco bacterial wilt identification method according to the first aspect.

[0032] The present application has the following beneficial effects over the prior art:

[0033] The tobacco hyperspectral image data is preprocessed and then input into the XGBoost model trained by the tobacco bacterial wilt data set to obtain a tobacco bacterial wilt identification result, so that the tobacco bacterial wilt is identified in a safe, convenient, fast, and non-destructive manner, the accuracy and precision are high, the identification efficiency is greatly improved, and the cost is reduced. BRIEF DESCRIPTION OF DRAWINGS

[0034] In order to more clearly illustrate the technical solutions in the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor based on these drawings.

[0035] Figure 1 is one of the flowcharts of the hyperspectral-based tobacco bacterial wilt identification method according to an embodiment of the present application;

[0036] Figure 2 is a flowchart of the ROI region reflectivity data processing according to an embodiment of the present application;

[0037] Figure 3 is a hyperspectral data original image according to an embodiment of the present application;

[0038] Figure 4 is a binary graph segmentation process according to an embodiment of the present application;

[0039] Figure 5 is a hyperspectral model prediction accuracy result according to an embodiment of the present application;

[0040] Figure 6 is a structural schematic diagram of a tobacco bacterial wilt recognition device based on hyperspectrum according to an embodiment of the present application;

[0041] Figure 7 is a structural schematic diagram of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0042] The technical solutions of the present application will be described clearly and completely in combination with the drawings in the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0043] As shown in Figure 1 The present application provides a tobacco bacterial wilt recognition method based on hyperspectrum, which comprises the following steps:

[0044] Step 100, acquiring tobacco hyperspectrum image data to be recognized;

[0045] Specifically, the hyperspectrum camera can be used to collect tobacco hyperspectrum image data. Since there are more night, indoor and closed environments in tobacco seedbed, additional light sources need to be added for light supplement to collect the reflection spectrum of tobacco plants, for example, tungsten filament lamp, halogen lamp, etc.

[0046] Exemplarily, the embodiments of the present application can adopt the intelligent crop information collection platform of Huazhong Agricultural University to collect data, and can realize full-automatic and non-destructive collection of trait parameters of greenhouse seedbed crops. The system mainly consists of a three-coordinate motion unit, an image collection unit and a control unit. The main body structure of the platform adopts aluminum alloy material, is installed on the seedbed, and has a size of 7000*2000*1950mm. The system adopts a three-axis motion unit driven by a stepping motor and a precise motion control card (ECI2000, ChinaZmotion), and the motion distances of the axes are 6100mm for the X axis, 950mm for the Y axis and 500mm for the Z axis. The image collection unit adopts an RGB camera (MARS-1230-23U3C, China Daheng), which is installed on the Y axis. The pixel resolution is 4096*3000 pixels, and a 16mm lens is used to obtain a field of view of 530mm*388mm. The spectral range of the hyperspectral camera (FX10e, Finland) is 400-1000nm, the spectral resolution is 5.5nm, the spatial pixels are 1024, the pixel size is 8*8um, the number of spectral bands is 224, the imaging speed is 330Hz for full-band acquisition, and the light source is a full-band halogen lamp. The seedbed equipment is manufactured by Guofeng Optoelectronics.

[0047] The acquired hyperspectral image data can be reflectance data, and the tobacco bacterial wilt identification method provided by the embodiments of the present application can determine whether the tobacco plant is suffering from bacterial wilt based on the reflectance data.

[0048] Step 101, pre-processing the hyperspectral image data;

[0049] Specifically, the pre-processing is to eliminate irrelevant information in the image, restore useful information, enhance the detectability of relevant information, maximize the simplification of data, and thus improve the reliability and accuracy of subsequent identification. The embodiments of the present application can use image pre-processing related algorithms (such as image enhancement algorithms) to enhance tobacco plant information and eliminate information other than tobacco plants (such as background information). The image data can also be filtered to remove clutter.

[0050] Step 102, inputting the pre-processed data into the trained XGBoost model to obtain a tobacco bacterial wilt identification result.

[0051] Specifically, the data after pre-processing is data that is convenient for reliable identification by the XGBoost model, for example, can be tobacco plant reflectance data or first-order derivative of reflectance data after removing background data, or can be average reflectance data or first-order derivative of average reflectance data after removing background tobacco plants. The data after pre-processing is used as input data of the XGBoost model.

[0052] The XGBoost model is a machine learning algorithm model based on gradient boosting decision trees. The contribution rate is obtained by calculating the sub-tree score, and the optimal feature band is selected. The preprocessed sample data is converted into a feature vector form, which is consistent with the features used in the training process. The trained model is loaded. The features of the sample to be predicted are mapped to the leaf nodes of each tree in the trained model one by one. For each tree, the feature values of the sample are matched step by step downward according to the tree splitting rules until the leaf node is reached. At the leaf node of each tree, there is a prediction value representing the probability or regression value of the leaf node belonging to a certain class. The prediction values of the leaf nodes of all trees are weighted and summed, and compared with the threshold value of the trained model to obtain the final prediction result.

[0053] The tobacco bacterial wilt dataset is annotated by researchers with rich experience in tobacco bacterial wilt research based on leaf scale. The XGBoost model is trained based on the dataset. After the training is completed, when the data after preprocessing of the hyperspectral image data collected by the hyperspectral camera is input, the model can correctly and quickly identify the tobacco plants infected with tobacco bacterial wilt.

[0054] The above technical solution identifies tobacco bacterial wilt by inputting the preprocessed tobacco hyperspectral image data into the XGBoost model trained based on the tobacco bacterial wilt dataset. This method is safe, convenient, fast, non-destructive, accurate and precise, greatly improves the identification efficiency, and reduces the cost.

[0055] Optionally, the preprocessing of the hyperspectral image data further comprises:

[0056] The ROI region extraction is performed on the tobacco hyperspectral image data to be identified to obtain hyperspectral data containing only the ROI region.

[0057] Specifically, the hyperspectral image data captured by the hyperspectral camera (such as the hyperspectral camera in the crop information collection platform) not only contains the hyperspectral data of the plant but also contains the hyperspectral data of the entire platform surface (background), so it is necessary to extract the hyperspectral data of the tobacco plant.

[0058] The ROI region refers to the region of interest in the image, which is usually the region that needs to be processed. In image processing, the extraction of the ROI region can greatly improve the processing efficiency and accuracy. The ROI region of the present application embodiment is the tobacco plant, and the extraction of the ROI region can obtain the hyperspectral data of the tobacco plant in the ROI region. For example, the threshold segmentation method can be used to obtain the ROI region, remove the background, and retain the hyperspectral data of the tobacco plant.

[0059] ROI region extraction is performed on the hyperspectral image data of tobacco to be identified, which facilitates data processing for the subsequent XGBoost recognition model.

[0060] Optionally, ROI region extraction is performed on the hyperspectral image data of the tobacco to be identified, further including:

[0061] Extract the full-band raw images acquired by the hyperspectral camera;

[0062] A mask image is created by selecting a band image from the original image where the background and tobacco plants have a significant difference.

[0063] The ROI region is obtained by performing a bitwise AND operation between the original image and the mask image.

[0064] Specifically, such as Figure 2 As shown, the first step involves using a program to read and preview data captured by a hyperspectral camera, extracting the original images across all 224 bands. The second step involves selecting two bands from the original image set that show a clear difference between the background and the tobacco plants to create mask images. For example, using... Figure 3 Mask images were created using two bands: the darker band (numbered 117) and the brighter band (numbered 222). The process for creating the mask images is as follows: Figure 4 As shown: Two images are divided using digital image processing techniques, followed by threshold segmentation. The image is then denoised using the small region removal module in LabVIEW to obtain a binary image of the Region of Interest (ROI). This binary image is an image where every pixel is either black or white. In this embodiment, threshold segmentation is used to assign a value of 1 to the plant region and 0 to the background region. The third step involves mapping the obtained binary image back to the original image. The original image is a high-dimensional array containing reflectance values ​​for each band. The binary image (ROI region as 1, background region as 0) is filtered and multiplied by the original image (i.e., bitwise AND operation) to obtain hyperspectral data containing only the ROI region.

[0065] Optionally, after extracting the ROI region from the hyperspectral image data of the tobacco to be identified, the process further includes:

[0066] The average reflectance of the ROI region is obtained based on the hyperspectral data and area of ​​the ROI region.

[0067] Specifically, when the hyperspectral data is reflectance data, the average reflectance of the ROI region can be obtained by weighting and summing the hyperspectral data of each band ROI region and then dividing by the area of ​​the ROI region.

[0068] Optionally, after extracting the ROI region from the hyperspectral image data of the tobacco to be identified, the process further includes:

[0069] The first derivative of the average reflectance data of the ROI region for each band is calculated based on the following formula:

[0070]

[0071] Where, d i The first derivative of the i-band is represented by x. i+1 x is the average reflectivity of band i+1; i-1 The average reflectance of band i-1.

[0072] Specifically, before constructing a hyperspectral-based tobacco bacterial wilt identification model, to make the dataset features more apparent and the classifier easier to learn, the data is further processed after obtaining the average reflectance. The first derivative of the average reflectance data for each ROI region is calculated. The first derivative provides information on the rate of change of spectral data. By analyzing the first derivative spectrum, characteristic peaks, troughs, and transition regions in the spectrum can be identified and analyzed, thereby better understanding the properties and composition of the spectral data and finding characteristic bands.

[0073] Optionally, the XGBoost model is trained based on the following objective function:

[0074]

[0075] Where L is the loss function; y i It is the true value of sample i; It is the model's predicted value after the (t-1)th iteration; f t (x i ) is the split node in the t-th iteration; Ω(f t ) is a regularization term.

[0076] Specifically, the objective function in the XGBoost method consists of two parts: a loss function and a regularization term. The loss function measures the model's fit, while the regularization term prevents overfitting. The loss function uses a weighted version of the gradient boosting decision tree algorithm, and the regularization term is a linear combination of L1 and L2 regularization terms. The basic idea is to sort the feature values, traverse the split points, and use the best splitting reward as the splitting reward for that feature. The feature with the best splitting reward is then selected as the splitting feature for the current node, and a binary split is performed based on its optimal splitting point to obtain left and right subtrees.

[0077] Split node score calculation formula:

[0078] Where gain represents the splitting reward; G L G R They are the left and right nodes respectively. cumulative sum of the first-order partial derivatives of H L , H R are the cumulative sum of the second-order partial derivatives of the left and right nodes respectively; λ is a coefficient; γ is a penalty term, which means if the gain is too small, it is better to choose not to split, and greater than γ to split.

[0079] The recognition effect evaluation of the embodiments of the present application is as follows:

[0080] The tobacco bacterial wilt data set is used for testing, the training set has 225 samples, the test set has 75 samples, and the leaves are scaled, and are labeled by researchers with rich experience in tobacco bacterial wilt research.

[0081] The accuracy (Accuracy) is the most common evaluation index, that is, the number of samples correctly divided is divided by the total number of samples, and the formula is as follows:

[0082]

[0083] Generally speaking, the higher the accuracy, the better the tobacco bacterial wilt recognition model (classifier) based on hyperspectral. TP: predicted as 1, predicted correctly, that is, actually 1; FP: predicted as 1, predicted incorrectly, that is, actually 0; FN: predicted as 0, predicted correctly, that is, actually 1; TN: predicted as 0, predicted correctly, that is, actually 0.

[0084] The precision (Precision) is also called the recall rate, which is for the predicted result, that is, the probability of actually positive samples in all predicted positive samples, that is, how many can be predicted correctly in the predicted positive sample results, and the formula is as follows:

[0085]

[0086] Among them, TP: predicted as 1, predicted correctly, that is, actually 1; FP: predicted as 1, predicted incorrectly, that is, actually 0;

[0087] Table 1. Prediction results

[0088] Human annotation results Hyperspectral model prediction results Diseased leaves 57 54 Healthy leaves 18 21

[0089] According to the prediction results in the above table, the accuracy of the model can be calculated by the above formula as 96.00%, and the precision is 94.73%.

[0090] As shown in Figure 5 , the data set is divided by time for testing. The detection accuracy of the hyperspectral model on the first day can reach 69.57%, the accuracy on the third day can reach 86.36%, and the detection accuracy on the sixth day at the end of the disease can reach 99.21%.

[0091] The tobacco bacterial wilt recognition device based on hyperspectrum provided by the present application is described below, and the tobacco bacterial wilt recognition device based on hyperspectrum described below can be correspondingly referred to the tobacco bacterial wilt recognition method based on hyperspectrum described above.

[0092] As shown in the figure, the tobacco bacterial wilt recognition device based on hyperspectrum comprises: Figure 6

[0093] The acquisition module 600 is configured to acquire tobacco hyperspectrum image data to be recognized.

[0094] The preprocessing module 610 is configured to preprocess the hyperspectrum image data.

[0095] The recognition module 620 is configured to input the preprocessed data into the trained XGBoost model to obtain a tobacco bacterial wilt recognition result.

[0096] Optionally, the preprocessing of the hyperspectrum image data further comprises:

[0097] The ROI region extraction is performed on the tobacco hyperspectrum image data to be recognized to obtain hyperspectrum data containing only the ROI region.

[0098] Optionally, the ROI region extraction on the tobacco hyperspectrum image data to be recognized further comprises:

[0099] The full-band original image collected by the hyperspectrum camera is extracted.

[0100] The band image with obvious difference between the background and the tobacco plant in the original image is selected to make a mask image.

[0101] The original image and the mask image are calculated by bitwise AND to obtain the ROI region.

[0102] Optionally, after the ROI region extraction on the tobacco hyperspectrum image data to be recognized, further comprising:

[0103] The average reflectivity of the ROI region is obtained according to the hyperspectrum data of the ROI region and the area of the ROI region.

[0104] Optionally, after the ROI region extraction on the tobacco hyperspectrum image data to be recognized, further comprising:

[0105] The first derivative of the average reflectivity data of each band ROI region is calculated based on the following formula:

[0106]

[0107] Wherein, d i ​denotes the first derivative of the i-th band; x i+1 is the average reflectance of the i+1-th band; x i-1 is the average reflectance of the i-1-th band.

[0108] Optionally, the XGBoost model is trained based on the following objective function:

[0109]

[0110] wherein L is a loss function; y i is a true value of the sample i; is a predicted value of the model after the t-1th iteration; f t (x i ) is a split node in the tth iteration; Ω(f t ) is a regularization term.

[0111] Figure 7 An example of an entity structure diagram of an electronic device is shown in Figure 7 The electronic device can include a processor 710, a communications interface 720, a memory 730, and a communications bus 740, wherein the processor 710, the communications interface 720, and the memory 730 communicate with each other through the communications bus 740. The processor 710 can invoke a logical instruction in the memory 730 to execute a tobacco bacterial wilt identification method based on hyperspectral data, which includes:

[0112] Obtaining tobacco hyperspectral image data to be identified;

[0113] Pretreating the hyperspectral image data;

[0114] Inputting the pretreated data into a trained XGBoost model to obtain a tobacco bacterial wilt identification result.

[0115] In addition, the logic instructions in the memory 730 described above can be implemented in the form of software function units and sold or used as independent products, and can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the parts that contribute to the prior art or parts of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0116] In another aspect, the present application also provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the tobacco bacterial wilt identification method based on hyperspectral provided by the above method, the method comprises:

[0117] Obtaining tobacco hyperspectral image data to be identified;

[0118] Pretreating the hyperspectral image data;

[0119] Inputting the pretreated data into the trained XGBoost model to obtain a tobacco bacterial wilt identification result.

[0120] The device embodiments described above are only schematic, wherein the units illustrated as separate components can or can not be physically separate, and the components illustrated as units can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment. Those skilled in the art can understand and implement without creative labor.

[0121] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be implemented by means of software plus necessary universal hardware platforms, and of course, can also be implemented by hardware. Based on such understanding, the above technical solutions essentially or the parts that contribute to the prior art can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in various embodiments or some parts of the embodiments.

[0122] Finally, it should be noted that the above examples are only used to illustrate the technical solutions of the present application, and are not intended to limit the same; although the present application has been described in detail with reference to the foregoing examples, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A tobacco bacterial wilt recognition method based on hyperspectrum, characterized in that, The method comprises: obtaining to-be-identified tobacco hyperspectral image data; preprocessing the hyperspectral image data, further comprising: extracting an ROI region from the to-be-identified tobacco hyperspectral image data to obtain hyperspectral data containing only the ROI region; wherein the ROI region extraction from the to-be-identified tobacco hyperspectral image data comprises: extracting a full-band original image collected by a hyperspectral camera; selecting a band image in which the difference between the background and the tobacco plant is obvious to make a mask image; and performing a bitwise AND operation on the original image and the mask image to obtain the ROI region; after the ROI region extraction from the to-be-identified tobacco hyperspectral image data, further comprising: obtaining the average reflectivity of the ROI region according to the hyperspectral data of the ROI region and the area of the ROI region; and calculating the first derivative of the ROI region average reflectivity data of each band based on the following formula: ； wherein, represents the first derivative of the i-th band; is the average reflectance of the i+1-th band; is the average reflectance of the i-1-th band; inputting the preprocessed data into a trained XGBoost model to obtain a tobacco bacterial wilt identification result; the XGBoost model is trained based on the following objective function: ； where L is a loss function; is the true value of sample i; is the predicted value of XGBoost model after t-1th iteration; is the split node in the tth iteration; is a regularization term; Split node score calculation formula: ; where represents the split gain; , are the cumulative sums of the first order partial derivatives of the left and right nodes , respectively; are the cumulative sums of the second order partial derivatives of the left and right nodes , respectively; is a coefficient; is a penalty term that represents the trade-off between choosing not to split if the gain is too small versus the side effect of increasing the model complexity, and is greater than if the split is performed.

2. A hyperspectral-based tobacco bacterial wilt recognition device for implementing the method of claim 1, characterized in that, The method comprises: an acquisition module configured to obtain to-be-identified tobacco hyperspectral image data; a preprocessing module configured to preprocess the hyperspectral image data; an identification module configured to input the preprocessed data into a trained XGBoost model to obtain a tobacco bacterial wilt identification result.

3. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor executes the program to implement the tobacco bacterial wilt identification method based on hyperspectra according to claim 1.

4. A computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the tobacco bacterial wilt identification method based on hyperspectra according to claim 1.

Images