Method and system for judging growth vigor of tea trees based on spectral remote sensing
By collecting and processing spectral images of tea tree growth using spectral remote sensing technology, a discrimination model is constructed, which solves the problems of low efficiency and destructiveness in traditional tea tree growth discrimination. This enables rapid, accurate, and non-destructive discrimination of tea tree growth, supporting precise management of tea gardens.
Patent Information
- Application Number
- CN202511132001.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-13
- Publication Date
- 2025-11-21
AI Technical Summary
Traditional methods for judging the growth of tea trees are inefficient and damage tea gardens, making it difficult to achieve rapid, non-destructive, large-scale monitoring.
Using spectral remote sensing technology, tea tree growth spectral images were collected through a DJI drone platform and an airborne hyperspectral instrument. After preprocessing and feature extraction, a tea tree growth discrimination model based on random forest or support vector machine was constructed to achieve non-destructive and rapid discrimination of tea tree growth.
It improves the accuracy and efficiency of tea tree growth assessment, provides a reliable basis for tea garden management, and reduces management costs.
Smart Images
Figure CN120997674A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of tea tree growth assessment technology, and more specifically to a method and system for assessing tea tree growth based on spectral remote sensing. Background Technology
[0002] Currently, accurate assessment of tea tree growth is a prerequisite for nutrient management and canopy cultivation in tea gardens, and a crucial aspect of tea production management. Traditional methods of assessing tea tree growth rely on manual observation and sampling, which is not only destructive but also inefficient. Spectral remote sensing technology, with its advantages of speed, non-destructive operation, and large-scale monitoring, can utilize spectral feature selection combined with machine learning to construct multi-level models for tea tree growth assessment. This allows for accurate assessment of tea garden growth, providing decision-making solutions for precise tea garden management and thereby reducing management costs.
[0003] Therefore, how to improve the accuracy and efficiency of judging the growth of tea trees without damaging the tea garden is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention
[0004] In view of this, the present invention provides a method and system for determining the growth status of tea trees based on spectral remote sensing, so as to solve the problems existing in the background art.
[0005] To achieve the above objectives, the present invention adopts the following technical solution:
[0006] A method for determining tea tree growth based on spectral remote sensing includes:
[0007] Collect spectral images of tea tree growth to obtain an initial spectral image dataset;
[0008] The initial spectral image dataset is preprocessed to obtain the preprocessed spectral image dataset.
[0009] Feature extraction is performed on the preprocessed spectral image dataset;
[0010] A tea tree growth discrimination model was constructed based on the extracted features;
[0011] A tea tree growth discrimination model is applied to process tea tree data in the region to be discriminated and obtain tea tree growth information.
[0012] Optionally, the acquisition of tea tree growth spectral images specifically includes:
[0013] Using the DJI drone platform and airborne hyperspectral imager, spectral images of tea tree growth were collected during the spring budding stage and summer growth period of mature tea gardens to obtain an initial spectral image dataset. The drone's flight altitude was controlled at 30 meters.
[0014] Optionally, the preprocessing includes preprocessing the obtained initial spectral image dataset through radiometric calibration, atmospheric correction, and image cropping.
[0015] Optionally, the feature extraction includes extracting the following features:
[0016] Vegetation index: The normalized vegetation index is calculated to reflect the growth status of tea trees;
[0017] Spectral reflectance characteristics: Analysis of the spectral reflectance characteristics of tea plants in different wavelength bands;
[0018] Red edge parameters: Extract the red edge position and red edge area parameters to evaluate the growth of tea trees.
[0019] Optionally, the construction of the tea tree growth discrimination model specifically involves using a random forest or support vector machine to establish the tea tree growth discrimination model.
[0020] Optionally, it also includes dividing the preprocessed spectral image dataset into a training set and a validation set according to a preset ratio; using the training set to train the model, and using the validation set to evaluate the model performance.
[0021] Optionally, the growth of tea trees can be classified into four levels—excellent, good, medium, and poor—using the normalized vegetation index (NDI) to determine the growth status of the tea trees.
[0022] A system for determining tea tree growth based on spectral remote sensing includes:
[0023] The data acquisition module collects spectral images of tea tree growth to obtain an initial spectral image dataset;
[0024] The data preprocessing module preprocesses the initial spectral image dataset to obtain the preprocessed spectral image dataset.
[0025] The feature extraction module extracts features from the preprocessed spectral image dataset;
[0026] The model building module constructs a tea tree growth discrimination model based on the extracted features;
[0027] The results output module applies a tea tree growth discrimination model to process tea tree data in the region to be discriminated and obtain tea tree growth information.
[0028] As can be seen from the above technical solution, compared with the prior art, this invention discloses a method and system for judging tea tree growth based on spectral remote sensing. The method involves acquiring spectral images of tea tree growth to obtain an initial spectral image dataset; preprocessing the initial spectral image dataset to obtain a preprocessed spectral image dataset; extracting features from the preprocessed spectral image dataset; constructing a tea tree growth discrimination model based on the extracted features; and applying the tea tree growth discrimination model to process the tea tree data in the area to be judged to obtain tea tree growth information. Compared with traditional methods of manual observation and sampling, this invention does not require damaging the tea garden, enables rapid and large-scale monitoring, effectively improves the accuracy and efficiency of tea tree growth judgment, and provides a reliable decision-making basis for precise tea garden management. Attached Figure Description
[0029] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on the provided drawings without creative effort.
[0030] Figure 1 This is a schematic diagram of the method flow provided by the present invention. Detailed Implementation
[0031] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0032] This invention discloses a method for determining tea tree growth based on spectral remote sensing, such as... Figure 1 As shown, it includes:
[0033] Collect spectral images of tea tree growth to obtain an initial spectral image dataset;
[0034] The initial spectral image dataset is preprocessed to obtain the preprocessed spectral image dataset.
[0035] Feature extraction is performed on the preprocessed spectral image dataset;
[0036] A tea tree growth discrimination model was constructed based on the extracted features;
[0037] A tea tree growth discrimination model is applied to process tea tree data in the region to be discriminated and obtain tea tree growth information.
[0038] In one specific embodiment, the acquisition of tea tree growth spectral images specifically includes:
[0039] Using the DJI drone platform and airborne hyperspectral imager, spectral images of tea tree growth (including five bands: red, green, blue, near-infrared, and red edge) were collected in mature tea gardens during the spring budding and summer growth periods to obtain an initial spectral image dataset. The drone's flight altitude was controlled at 30 meters.
[0040] In one specific embodiment, the preprocessing includes preprocessing the obtained initial spectral image dataset through radiometric calibration, atmospheric correction, and image cropping.
[0041] In one specific embodiment, the feature extraction includes extracting the following features:
[0042] Vegetation index: The normalized vegetation index is calculated to reflect the growth status of tea trees;
[0043] Spectral reflectance characteristics: Analysis of the spectral reflectance characteristics of tea plants in different wavelength bands;
[0044] Red edge parameters: Extract the red edge position and red edge area parameters to evaluate the growth of tea trees.
[0045] Specifically, using the remote sensing image processing platform (ENVI) software, the reflectance orthophotos of the five bands were combined into multispectral data.
[0046] Vegetation indices were obtained using remote sensing index methods: Normalized Difference Vegetation Index (NDVI) and Normalized Difference Red Edge Index (NDRE).
[0047] NDVI = (NIR - R) / (NIR + R)
[0048] NIR: Near-infrared reflectance (vegetation leaves strongly reflect near-infrared light due to scattering by cell structure);
[0049] R: Red light reflectance (chlorophyll strongly absorbs red light and reflects it weakly).
[0050] Physical meaning: By utilizing the absorption-reflection difference in the "red-near infrared" band, vegetation coverage and vitality can be quantified (the closer the value is to 1, the more lush the vegetation).
[0051] NDRE = (NIR - REG) / (NIR + REG)
[0052] The parameters are as follows: REG represents the red-edge reflectance (the transition zone from red light to near-infrared, which is more sensitive to changes in chlorophyll).
[0053] Advantages: It captures changes in chlorophyll content (such as nitrogen in tea trees and their health status) more accurately than NDVI, making it suitable for precise monitoring.
[0054] And it proposes the Tea Tree Specific Index (TCI):
[0055]
[0056] R 720 : 720nm band reflectivity (near the red edge, close to near-infrared, tea trees have unique reflectivity in this band);
[0057] R 680 : 680nm band reflectance (red light region, chlorophyll absorption peak).
[0058] Design logic: Based on the leaf structure and chlorophyll characteristics of tea trees, the 720nm and 680nm wavelength bands are selected to highlight the spectral differences of tea trees and improve the accuracy of growth monitoring (such as distinguishing new shoot density and leaf thickness).
[0059] Red edge parameters: Extract the red edge position (REP) and red edge slope (RES), and use the first derivative method to locate the REP.
[0060] Spectral reflectance characteristics: 680nm (red valley) and 720nm (red edge) were selected as key bands.
[0061] In one specific embodiment, the construction of the tea tree growth discrimination model specifically involves using a random forest or support vector machine to establish the tea tree growth discrimination model.
[0062] In one specific embodiment, the method further includes dividing the preprocessed spectral image dataset into a training set and a validation set according to a preset ratio; using the training set to train the model; and using the validation set to evaluate the model performance.
[0063] In one specific embodiment, the tea tree growth is further classified into four levels—excellent, good, medium, and poor—using the normalized vegetation index (NDI) to determine the growth status of the tea trees.
[0064] The following specific example further illustrates the method of the present invention.
[0065] Taking the growth assessment of tea trees in a mature tea garden (approximately 50 acres, with trees aged 10 years) as an example:
[0066] 1. Data Collection
[0067] Time: In March (spring tea budding period) and July (summer growing period) of a certain year, select sunny and cloudless weather with wind force ≤3, and collect samples from 10:00 am to 2:00 pm (stable light period).
[0068] Equipment: DJI Matrice 300RTK drone platform, equipped with GaiaSky-mini hyperspectral imager (spectral range 400-1000nm, 256 bands, spectral resolution 3.5nm).
[0069] Flight parameters: flight altitude 30 meters, flight speed 5 m / s, forward overlap 80%, lateral overlap 70% to ensure complete image stitching; GPS position information is recorded simultaneously for later image geometric correction.
[0070] Data output: Obtain the initial spectral image dataset (including 200 images of spring tea and 300 images of summer growing season, in TIFF format).
[0071] 2. Data Preprocessing
[0072] Radiometric calibration: Using the calibration coefficients built into the hyperspectral imager, the DN values (digital quantization values) of the original image are converted into apparent reflectance to eliminate differences in instrument response.
[0073] Atmospheric correction: Using the FLAASH module of ENVI 5.6 software, the meteorological data collected (temperature 25℃, humidity 60%, air pressure 1013hPa) was input to remove the effects of atmospheric scattering and absorption, and obtain the true surface reflectance image.
[0074] Image cropping: The boundaries of the tea garden were manually delineated using ArcGIS software, and non-tea tree areas such as surrounding roads and buildings were cropped out of the image to retain the effective tea garden area, resulting in a preprocessed spectral image dataset (450 effective images in total).
[0075] 3. Feature Extraction
[0076] Normalized Difference Vegetation Index (NDVI): Calculated based on the near-infrared band (800nm) and the red band (650nm), the formula is NDVI = (NIR - R) / (NIR + R). It is used to reflect the chlorophyll content of tea leaves. The higher the value, the more vigorous the growth.
[0077] Spectral reflectance characteristics: Analysis of tea plant reflectance in key wavelength bands:
[0078] Visible light band (400-700nm): A green reflection peak appears at 550nm (the reflectance of a healthy tea tree is about 30%), and a trough appears at 670nm red light due to chlorophyll absorption (the reflectance is about 10%).
[0079] Near-infrared band (700-1000nm): The reflectance is significantly increased at 750-900nm (over 60% in healthy tea trees), which is related to the leaf cell structure.
[0080] Red edge parameter: Red edge position: The wavelength corresponding to the maximum value of the first derivative of spectral reflectance (approximately 720-730nm for healthy tea trees, and the blue shift of the red edge position to around 710nm for tea trees with poor growth).
[0081] Red edge area: The area enclosed by the reflectance curve and the baseline within the red edge band (680-760nm) (healthy tea trees have a larger red edge area, about 15-20 (%) nm).
[0082] 4. Model Building and Training
[0083] Dataset partitioning: The 450 preprocessed images were divided into a training set (360 images) and a validation set (90 images) in an 8:2 ratio. Each image corresponds to feature data of 100 tea tree sample points (a total of 36,000 training samples and 9,000 validation samples).
[0084] Model selection: Random forest model (implemented by Python scikit-learn library) was adopted, with parameters set as follows: n_estimators = 100 (number of decision trees), max_depth = 10 (maximum tree depth), and random states = 42.
[0085] Training and Validation: The training process used NDVI, keyband reflectance, and red-edge parameters as input features, and the crop growth grades (excellent, good, medium, and poor) from manual field surveys as labels. Validation set evaluation results: The accuracy was 91.2%, and the confusion matrix showed that the misclassification rate for "good" and "medium" was the lowest (<5%).
[0086] 5. Growth assessment and result output
[0087] Spectral images of a 10-mu (approximately 1.65 acres) area of the tea garden to be identified were collected in August of a certain year. After preprocessing and feature extraction, the images were input into the model, and the growth level was output.
[0088] Advantages: 28% (NDVI>0.7, red edge position 725nm);
[0089] Good: 45% (NDVI 0.5-0.7, red edge position 720nm);
[0090] China: 20% (NDVI 0.3-0.5, red edge position 715nm);
[0091] Poor: 7% (NDVI<0.3, red edge position 710nm).
[0092] Generate a tea garden growth distribution map to guide precise management: apply nitrogen fertilizer to areas with "poor" ratings and increase irrigation frequency for areas with "medium" ratings.
[0093] A system for determining tea tree growth based on spectral remote sensing includes:
[0094] The data acquisition module collects spectral images of tea tree growth to obtain an initial spectral image dataset;
[0095] The data preprocessing module preprocesses the initial spectral image dataset to obtain the preprocessed spectral image dataset.
[0096] The feature extraction module extracts features from the preprocessed spectral image dataset;
[0097] The model building module constructs a tea tree growth discrimination model based on the extracted features;
[0098] The results output module applies a tea tree growth discrimination model to process tea tree data in the region to be discriminated and obtain tea tree growth information.
[0099] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the apparatus disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the description is relatively simple; relevant parts can be referred to the method section.
[0100] The above description of the disclosed embodiments enables those skilled in the art to make or use the invention. Various modifications to these embodiments will be readily apparent to those skilled in the art, and the general principles defined herein may be applied without departing from the invention.
[0101] In other embodiments, the spirit or scope of the invention is considered. Therefore, the invention will not be...
[0102] The embodiments shown herein are intended to be applicable, rather than limited to, the principles and new ideas disclosed herein.
[0103] The widest range that is consistent with the characteristics of Ying.
Claims
1. A method for determining tea tree growth based on spectral remote sensing, characterized in that, include: Collect spectral images of tea tree growth to obtain an initial spectral image dataset; The initial spectral image dataset is preprocessed to obtain the preprocessed spectral image dataset. Feature extraction is performed on the preprocessed spectral image dataset; A tea tree growth discrimination model was constructed based on the extracted features; A tea tree growth discrimination model is applied to process tea tree data in the region to be discriminated and obtain tea tree growth information.
2. The method for determining tea tree growth based on spectral remote sensing according to claim 1, characterized in that, The acquisition of tea tree growth spectral images specifically includes: Using the DJI drone platform and airborne hyperspectral imager, spectral images of tea tree growth were collected during the spring budding stage and summer growth period of mature tea gardens to obtain an initial spectral image dataset. The drone's flight altitude was controlled at 30 meters.
3. The method for determining tea tree growth based on spectral remote sensing according to claim 1, characterized in that, The preprocessing includes preprocessing the obtained initial spectral image dataset through radiometric calibration, atmospheric correction, and image cropping.
4. The method for determining tea tree growth based on spectral remote sensing according to claim 1, characterized in that, The feature extraction includes extracting the following features: Vegetation index: The normalized vegetation index is calculated to reflect the growth status of tea trees; Spectral reflectance characteristics: Analysis of the spectral reflectance characteristics of tea plants in different wavelength bands; Red edge parameters: Extract the red edge position and red edge area parameters to evaluate the growth of tea trees.
5. The method for determining tea tree growth based on spectral remote sensing according to claim 1, characterized in that, The specific method for constructing a tea tree growth discrimination model is to use random forest or support vector machine to establish the tea tree growth discrimination model.
6. The method for determining tea tree growth based on spectral remote sensing according to claim 1, characterized in that, It also includes dividing the preprocessed spectral image dataset into a training set and a validation set according to a preset ratio; using the training set to train the model; and using the validation set to evaluate the model performance.
7. The method for determining tea tree growth based on spectral remote sensing according to claim 1, characterized in that, It also includes using the normalized vegetation index (NDI) to classify tea tree growth into four levels: excellent, good, medium, and poor, in order to determine the growth status of tea trees.
8. A system for determining tea tree growth based on spectral remote sensing, characterized in that, The method for determining tea tree growth based on spectral remote sensing according to any one of claims 1-7 includes: The data acquisition module collects spectral images of tea tree growth to obtain an initial spectral image dataset; The data preprocessing module preprocesses the initial spectral image dataset to obtain the preprocessed spectral image dataset. The feature extraction module extracts features from the preprocessed spectral image dataset; The model building module constructs a tea tree growth discrimination model based on the extracted features; The results output module applies a tea tree growth discrimination model to process tea tree data in the region to be discriminated and obtain tea tree growth information.