Plant hyperspectral data processing method and device, equipment and medium

Abstract

The invention provides a plant hyperspectral data processing method and device, equipment and a medium, and belongs to the technical field of data processing.The method comprises the steps that first data and second data are preprocessed to obtain third data, the first data are plant canopy structure data, and the second data are plant spectral characteristic data; inputting the third data into the first neural network model to obtain a prediction result, and determining a corresponding target irrigation strategy according to the prediction result; the first neural network model is a neural network model for obtaining a growth trend predicted value and a moisture demand predicted value of the plant according to the third data. According to the plant hyperspectral data processing method and device, the equipment and the medium provided by the invention, the soil moisture content can be adjusted based on the change conditions of the plant canopy structure and the plant hyperspectral data, and the plant quality is improved while the normal growth of the plant is ensured.

Description

Technical Field

[0001] The present disclosure belongs to the technical field of data processing, and more specifically, relates to a method and device, equipment, and medium for processing plant hyperspectral data. Background Art

[0002] As one of the important environmental factors affecting plant growth and development, water plays an important role in the phenotype, yield, synthesis, and accumulation of secondary metabolites of medicinal plants. When the water supply is sufficient, medicinal plants can carry out normal metabolism and growth and development, which is beneficial to the formation of yield; when the water supply is insufficient, medicinal plants will respond to environmental stress by inducing the synthesis and accumulation of secondary metabolites, which is beneficial to the formation of the quality of medicinal materials. However, when the water stress is severe, the growth and development of medicinal plants will be hindered, and even the death of plants may occur. Long-term drought experiments have proved that mild and moderate drought stresses are beneficial to the synthesis of active ingredients for plant growth, while severe drought stress will inhibit the accumulation of active ingredients in plants and resist the invasion of the external environment by starting its own defense system and stimulating the synthesis of active ingredients synergistically. This is mainly because when the water content in the soil decreases, plant roots will form abscisic acid, which triggers the plant's adaptation mechanism to drought, reduces transpiration, promotes root growth, and changes physiological metabolic processes, etc., thus affecting the normal growth of plants and reducing the quality of plants at the same time. Summary of the Invention

[0003] The purpose of the present disclosure is to provide a method and device, equipment, and medium for processing plant hyperspectral data, so as to adjust the soil water content based on the changes in the plant canopy structure and plant hyperspectral data, and improve the quality of plants while ensuring the normal growth of plants.

[0004] In the first aspect of the embodiments of the present disclosure, a method for processing plant hyperspectral data is provided, including: Preprocessing the first data and the second data to obtain the third data, where the first data is plant canopy structure data and the second data is plant spectral characteristic data; Inputting the third data into the first neural network model to obtain a prediction result, and determining the corresponding target irrigation strategy according to the prediction result; the first neural network model is a neural network model that obtains the plant growth trend prediction value and water demand prediction value according to the third data.

[0005] Optionally, the method for processing plant hyperspectral data further includes: Obtaining a two-dimensional image sequence of the plant through a visible light camera, performing three-dimensional point cloud reconstruction to obtain plant canopy structure data; Sensing hyperspectral data at the plant canopy level through a hyperspectral sensor to obtain plant spectral characteristic data.

[0006] Optionally, preprocessing the first data and the second data to obtain third data includes: Performing temporal and spatial alignment on the plant canopy structure data and the plant spectral characteristic data to generate fusion data that matches in time and space; Performing multi-modal feature extraction on the fusion data to obtain third data.

[0007] Optionally, performing multi-modal feature extraction on the fusion data to obtain third data includes: Using multi-modal feature extraction technology to separately extract the plant canopy structure features of the plant canopy structure data and the spectral features of the plant spectral characteristic data; Splicing the plant canopy structure features and the spectral features to generate third data.

[0008] Optionally, the plant hyperspectral data processing method further includes: Before determining the target irrigation strategy, performing weighted fusion on the prediction result and the environmental characteristics to obtain a first weighted value; In response to the first weighted value being less than the first threshold, determining that the plant does not need to be irrigated; In response to the first weighted value being greater than or equal to the first threshold, determining that the plant needs to be irrigated; The environmental characteristics include soil humidity, light intensity, and air temperature.

[0009] Optionally, Determining the corresponding target irrigation strategy according to the prediction result includes: In response to the predicted value of water demand being less than the second threshold, determining that the irrigation method is the first irrigation method; In response to the predicted value of water demand being greater than or equal to the second threshold, determining that the irrigation method is the second irrigation method; the water supply methods of the first irrigation method and the second irrigation method are different.

[0010] Optionally, Determining the corresponding target irrigation strategy according to the prediction result further includes: Performing weighted fusion on the predicted value of growth trend and the predicted value of water demand; obtaining a second weighted value; In response to the second weighted value being less than the third threshold, determining that the irrigation amount is the first irrigation amount; In response to the second weighted value being greater than or equal to the third threshold, determining that the irrigation amount is the second irrigation amount; The first irrigation amount is less than the second irrigation amount.

[0011] In a second aspect of the embodiments of the present disclosure, there is provided a plant hyperspectral data processing device, including: A data processing module for preprocessing first data and second data to obtain third data, where the first data is plant canopy structure data and the second data is plant spectral characteristic data; An irrigation strategy determination module for inputting the third data into a first neural network model to obtain a prediction result, and determining a corresponding target irrigation strategy according to the prediction result; the first neural network model is a neural network model for obtaining a plant growth trend prediction value and a water demand prediction value based on the third data.

[0012] In a third aspect of the embodiments of the present disclosure, there is provided an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, and when the processor executes the computer program, the steps of the above-mentioned plant hyperspectral data processing method are implemented.

[0013] In a fourth aspect of the embodiments of the present disclosure, there is provided a computer-readable storage medium storing a computer program, and when the computer program is executed by a processor, the steps of the above-mentioned plant hyperspectral data processing method are implemented.

[0014] The beneficial effects of the plant hyperspectral data processing method, device, equipment, and medium provided by the embodiments of the present disclosure are as follows: Based on the changes in plant canopy structure data and plant hyperspectral data, a plant growth trend prediction value and a water demand prediction value are obtained, and the soil moisture content is further adjusted to avoid the adverse effects of excessive or insufficient water on plants while ensuring the water required for normal plant growth, thereby effectively improving the quality of plants. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] In order to more clearly illustrate the technical solutions in the embodiments of the present disclosure, the following will briefly introduce the drawings required for use in the embodiments or the description of the prior art. Obviously, the following drawings are only some embodiments of the present disclosure, and those of ordinary skill in the art can also obtain other drawings based on these drawings without creative efforts.

[0016] Figure 1 It is a flowchart of a plant hyperspectral data processing method provided by an embodiment of the present disclosure; Figure 2 It is a structural block diagram of a plant hyperspectral data processing device provided by an embodiment of the present disclosure; Figure 3 It is a schematic block diagram of an electronic device provided by an embodiment of the present disclosure. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0017] In the following description, specific details such as specific system architectures and technologies are presented for the purpose of illustration rather than limitation, so as to thoroughly understand the embodiments of the present disclosure. However, those skilled in the art should clearly understand that the present disclosure can also be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid unnecessary details from interfering with the description of the present disclosure.

[0018] To make the objectives, technical solutions, and advantages of the present disclosure clearer, the following will be described through specific embodiments in conjunction with the accompanying drawings.

[0019] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of a method for processing plant hyperspectral data provided for an embodiment of the present disclosure. The method includes: S101: Preprocess the first data and the second data to obtain the third data. The first data is plant canopy structure data, and the second data is plant spectral characteristic data.

[0020] In this embodiment, a two-dimensional image sequence of the plant is obtained through a visible light camera, and three-dimensional point cloud reconstruction is performed to obtain the plant canopy structure data; hyperspectral data at the plant canopy level is sensed through a hyperspectral sensor to obtain the plant spectral characteristic data.

[0021] Specifically, the plant spectral characteristic data reflects the changes in the physiological and biochemical information of the plant under different environmental conditions from multiple levels such as the leaf level, population level, and ecosystem, and is used for plant nutrition diagnosis, biomass estimation, growth status monitoring, soil nutrient content inversion, etc.

[0022] In this embodiment, a hyperspectral measurement sensor is used to measure the hyperspectral data at the plant canopy level, and the hyperspectral data range is 350 nm - 1100 nm. At different growth stages of the plant, according to the growth status and environmental conditions, a sunny, cloudless, or less cloudy day is selected. For example, the plant canopy reflection spectrum is collected from 10:00 am to 14:00 pm. During the measurement, the spectrometer probe is kept perpendicular downward, and the vertical height from the canopy is 1 m. To ensure the accuracy of the experiment, the calibration of the whiteboard needs to be carried out in a timely manner, and it is necessary to ensure that the measurement area is representative, avoiding measuring the edge or the part affected by the shadow.

[0023] In this embodiment, preprocessing the first data and the second data to obtain the third data includes time and space alignment of the plant canopy structure data and the plant spectral characteristic data to generate fused data that is time and space matched; and performing multi-modal feature extraction on the fused data to obtain the third data.

[0024] Specifically, there are differences in the acquisition time and space between the plant canopy structure data and the plant spectral characteristic data. In the time dimension, the growth process of plants is a dynamic process, and the data collected at different times may reflect different growth states of plants. For example, during the peak growth season of plants, significant changes may occur in the canopy structure, and the spectral characteristics will also change accordingly. Therefore, it is necessary to calibrate the data in terms of time according to the growth cycle of plants and the time interval of data acquisition, and adjust the data at different time points to the same time benchmark. In the space dimension, the spatial coordinates of the plant canopy structure data and the plant spectral characteristic data are uniformly transformed. For example, the coordinate systems of the plant canopy structure data and the plant spectral characteristic data are matched to ensure that the data at the same spatial position can correspond to each other.

[0025] In this embodiment, a fusion algorithm combining principal component analysis and Kalman filtering is used to extract the plant canopy structure features of the canopy structure data and the spectral features of the hyperspectral data respectively; the plant canopy structure features and the spectral features are spliced to generate the third data.

[0026] Specifically, using multi-modal feature extraction technology, for the plant canopy structure data, structural features such as canopy height, canopy width, and leaf area index are extracted. These features can reflect the morphology and spatial distribution of the plant canopy, and obtain the growth status and spatial distribution of the plants. For the plant spectral characteristic data, spectral features such as reflectance and absorbance in different bands are extracted. Through the multi-modal feature extraction algorithm, the plant canopy structure features and the spectral features are fused to extract the third data that can comprehensively reflect the growth state and ecological characteristics of plants.

[0027] S102: Input the third data into the first neural network model to obtain a prediction result, and determine the corresponding target irrigation strategy according to the prediction result; the first neural network model is a neural network model that obtains the predicted growth trend value and water demand predicted value of plants based on the third data.

[0028] In this embodiment, a depth neural network regression algorithm is used to construct the first neural network model, and the plant canopy structure features and spectral features are input into the first neural network model to predict the plant growth trend and water demand; the first neural network model is used to analyze the relationship between the plant canopy structure features and spectral features and the water content.

[0029] Specifically, through the vegetation index calculation formula in Table 1, indicators such as water content, photosynthetic traits, and pigment content reflecting the plant canopy structure are obtained.

[0030] Table 1 Vegetation index calculation formula

[0031] Among them, R xis the spectral reflectance at wavelength x, where x is the wavelength in nm.

[0032] In this embodiment, the first neural network model is obtained by training a neural network model based on an original data set, and includes: dividing the original data set into a training set and a test set, training the neural network model using the training set, and determining the corresponding hyperparameter combination when obtaining the best model effect by using the methods of grid search tuning and cross-validation; predicting the unknown data in the test set, and evaluating the accuracy of the first neural network model by calculating the coefficient of determination (R 2 ), root mean square error (RMSE), and mean absolute error (MAE) between the predicted value and the corresponding experimental value. The root mean square error (RMSE) represents the square root of the ratio of the sum of the squares of the differences between the predicted value and the true value to the number of observations n. The root mean square error is very sensitive to extremely large or small errors in a set of measurements, so it can well reflect the precision of the prediction results.

[0033] In this embodiment, before determining the target irrigation strategy, the prediction result is weighted and fused with the environmental characteristics to obtain a first weighted value, and it is determined whether the plant needs to be irrigated according to the first weighted value; in response to the first weighted value being less than the first threshold, it is determined that the plant does not need to be irrigated; in response to the first weighted value being greater than or equal to the first threshold, it is determined that the plant needs to be irrigated; the environmental characteristics include soil humidity, light intensity, and air temperature and humidity.

[0034] In this embodiment, before weighting and fusing the prediction result with the environmental characteristics, preprocessing is performed on the prediction result and the environmental characteristic data, and the preprocessing includes removing outliers and normalizing to improve the accuracy of the weighted fusion result. The acquisition of environmental characteristics includes: collecting data in real time through humidity sensors, light sensors, and temperature and humidity sensors deployed in the root zone of the plant, and uploading it to the cloud server for standardization processing every 5 minutes, and participating in the weighted calculation after removing noise.

[0035] In the embodiment, the calculation formula of the first weighted value is: W fusion = α× W pred +β× G trend +γ×( S dry - L real / L max + T air / T opt ) Wherein, Wfusion is the first weighted value (irrigation demand index), W pred is the predicted value of the water requirement of the plant, G trend is the predicted value of the growth trend of the plant, S dry is the difference between the soil moisture and the field capacity; L real / L max is the ratio of the real-time light intensity to the local historical maximum value; T air / T opt is the ratio of the air temperature to the optimum temperature of the crop; α is the first weight coefficient, β is the second weight coefficient, γ is the third weight coefficient, and α + β + γ = 1. Increase the γ weight during the seedling stage of the plant and increase the α weight during the reproductive stage of the plant.

[0036] Specifically, the predicted value of plant growth trend is positively correlated with the water demand of plants; the plant growth trend indicates whether the plant is in the vigorous growth period or the growth decline state, etc.; the water demand of plants is relatively large during the vigorous growth period, and when the plant is in the growth decline state, protective irrigation needs to be initiated and multiple irrigations are carried out, rather than directly irrigating according to the water demand, otherwise it will damage the roots of the plants. The first weighted value is obtained by weighted calculation of the predicted value of plant water demand, the predicted value of plant growth trend, and environmental characteristic values (the difference between soil humidity and field capacity, the ratio of real-time light intensity to the local historical maximum value, and the ratio of air temperature to the optimum temperature of the crop). Dynamically adjust the weights: During the seedling stage of the plant, since the root system is not yet developed and is more sensitive to soil moisture, the value of the weight γ can be appropriately increased, and more consideration can be given to the predicted value of plant growth trend and soil humidity conditions to adjust the irrigation strategy; during the reproductive stage, the water demand of the plant is closely related to the growth stage, and the value of the weight α is increased, and the irrigation amount is determined according to the predicted value of plant water demand to meet the growth needs of the plant. The functions of environmental characteristic values include: the difference between soil humidity and field capacity reflects the current water surplus or deficit status of the soil; when the difference is negative, it indicates that the soil moisture exceeds the field capacity, and irrigation needs to be reduced to prevent soil waterlogging and avoid root hypoxia and disease breeding; when the difference is positive, it means that the soil moisture is insufficient, and the irrigation amount needs to be increased to meet the growth needs of the plant. Light intensity directly affects the transpiration and photosynthesis of plants. When the light intensity approaches or exceeds the local historical maximum value, the transpiration of plants will increase sharply, water loss will accelerate, and photosynthesis may also be inhibited. When a high transpiration demand caused by abnormal light intensity is detected, water supply needs to be increased to ensure that the plant has sufficient water to maintain physiological activities; at the same time, the logic set by the negative sign is used to inhibit excessive irrigation to avoid excessive soil moisture caused by blind water replenishment and damage the roots of the plants. The ratio of air temperature to the optimum temperature of the crop; when the ratio of air temperature to the optimum temperature of the crop is greater than 1, it indicates that the air temperature is higher than the optimum temperature of the plant, which may lead to an increase in plant respiration, excessive energy consumption, affect its growth and development, and at the same time will also intensify the transpiration of the plant and increase the water demand; if the ratio is less than 1, it means that the air temperature is lower than the optimum temperature of the plant, which may inhibit the metabolism and physiological activity of the crop and reduce the photosynthesis efficiency, etc. The present disclosure more comprehensively and accurately predicts whether irrigation is needed for plants through multiple factors.

[0037] In this embodiment, the method for determining the first threshold includes: Step 1: Establish a crop water stress response curve Obtain the change rate of canopy spectral reflectance, the inflection point of stomatal conductance decline, and the sudden drop threshold of photosynthetic rate of samples at different growth stages under gradient water stress through controlled experiments.

[0038] Step 2: Construct a multi-factor coupling model

[0039] Among them, is the threshold value, LAI is the leaf area index (from canopy structure data), VPD is the vapor pressure deficit (kPa), Ψ is the soil water potential (kPa), and a, b, and c are the first plant type coefficient, the second plant type coefficient, and the third plant type coefficient, respectively.

[0040] Step 3: Dynamic environment correction Introduce an adjustment factor λ according to real-time meteorological data:

[0041] Among them, is the reference evapotranspiration, is the plant water requirement, and RH is the ratio of day-night relative humidity.

[0042] Step 4: Final threshold

[0043] Among them, is the final threshold (the first threshold).

[0044] In this embodiment, according to the prediction result, determine the corresponding target irrigation strategy, including: in response to the predicted value of water demand being less than the second threshold, determine the irrigation method as the first irrigation method; in response to the predicted value of water demand being greater than or equal to the second threshold, determine the irrigation method as the second irrigation method.

[0045] Specifically, the first irrigation method is drip irrigation, and the drip irrigation rate is determined according to the mapping relationship between the predicted value of water demand and the preset drip irrigation rate range. The second irrigation method is sprinkler irrigation, and the sprinkler irrigation range is determined according to the mapping relationship between the predicted value of water demand and the preset sprinkler irrigation range. The preset sprinkler irrigation range is divided according to the area and layout of the plant planting area. After determining the irrigation method, the irrigation amount is also corrected according to the permeability coefficient of the soil. If the soil permeability coefficient is large, the irrigation amount is increased; if the soil permeability coefficient is small, the irrigation amount is decreased.

[0046] In this embodiment, according to the prediction result, determining the corresponding target irrigation strategy further includes: performing weighted fusion on the predicted value of growth trend and the predicted value of water demand; obtaining a second weighted value; in response to the second weighted value being less than the third threshold, determining the irrigation amount as the first irrigation amount; in response to the second weighted value being greater than or equal to the third threshold, determining the irrigation amount as the second irrigation amount.

[0047] Specifically, the formula for calculating the second weighted value is: V second = w1 W pred +w 2 G trend Among them, W pred is the predicted value of plant water demand, G trend is the predicted value of growth trend, V second is the second weighting value, w 1 is the fourth weighting coefficient, w 2 is the fifth weighting coefficient.

[0048] The calculation formula for the irrigation amount is: I

[0049] W pred is the predicted value of plant water demand, G trend is the predicted value of growth trend, V third is the third threshold; is the basic irrigation amount; is the first irrigation amount adjustment coefficient, is the second irrigation amount adjustment coefficient; is the plant growth stage coefficient; is the first irrigation amount, is the second irrigation amount.

[0050] The present disclosure introduces a plant growth stage coefficient to determine the final irrigation amount, which is adjusted according to different stages of the plant, such as the seedling stage, vegetative growth stage, flowering stage, fruiting stage, etc., to reflect the different degrees of water demand in different growth stages.

[0051] In this embodiment, the target strategy of the present disclosure also includes a drainage measure to detect the soil moisture in real time; if the soil moisture value is greater than the preset moisture threshold, an alarm is given; the user is reminded to drain the water in time to reduce the soil moisture content.

[0052] The plant hyperspectral data processing method provided by the embodiment of the present disclosure adjusts the soil moisture content based on the plant canopy structure and the change of plant hyperspectral data, and improves the quality of the plant while ensuring the normal growth of the plant.

[0053] In one embodiment of the present disclosure, a two-dimensional image sequence of Atractylodes chinensis (DC.) Koidz. is obtained by a high-resolution visible light camera. The segmentation of Atractylodes chinensis (DC.) Koidz. from the background environment is achieved based on the HSV model threshold segmentation method. A three-dimensional point cloud is generated by combining the Structure from Motion (SfM) algorithm, and the distance conversion between coordinate systems is carried out using a checkerboard, finally realizing the three-dimensional reconstruction of Atractylodes chinensis (DC.) Koidz., and then completing the extraction of key phenotypes. The phenotypic parameters of plants can be quickly and accurately extracted, and non-destructive measurement can be achieved, meeting the requirements for measuring the three-dimensional phenotypic parameters and growth observation of Atractylodes chinensis (DC.) Koidz. The three-dimensional point cloud of Atractylodes chinensis (DC.) Koidz. is generated using the open-source software VisualSFM based on multi-view stereo Structure from Motion. The sparse point cloud of the plant is reconstructed by importing two sets of two-dimensional images of the plant from different perspectives. The sparse three-dimensional point cloud obtained by SfM reconstruction needs to be reconstructed into a dense three-dimensional point cloud of the target for better reconstruction effect. The dense point cloud of the plant is generated using a three-dimensional multi-view stereo vision algorithm based on patches. The generated three-dimensional dense point cloud is pre-filtered to remove noise points and outliers, obtaining a relatively smooth dense point cloud; then, a calibration plate is used for scale scaling, and the size of each square in the calibration plate is 50mm×50mm, and the overall size is 150mm×150mm. The scaling relationship of the coordinate position is calculated through the scaling relationship calculation formula K , and by using K , the conversion between the distance in the point cloud coordinate system and the true distance in the world coordinate system can be carried out.

[0054]

[0055] In the formula, K represents the distance conversion coefficient, O is the origin coordinate (0, 0, 0) of the world coordinate system, is the coordinate (0, 50, 0) of the upper right corner point of the center white square of the calibration plate plane, is the coordinate (50, 0, 0) of the lower left corner point of the center white square, is the coordinate (50, 50, 0) of the lower right corner of the center white square; Q is the origin coordinate (x, y, z) of the generated point cloud coordinate system, and the points (x 1 , y 1 , z 1 ), (x 2 , y 2 , z 2 ) and (x 3 , y 3 , z 3 ) are the corresponding coordinates in the world coordinate system.

[0056] Error analysis shows that the measurement error of machine vision mainly comes from the influence of noise on the three-dimensional point cloud data. There is noise interference at the top of the plant, resulting in a certain error in the generated bounding box. New leaves usually germinate at the growth point of the plant, and there will be leaf occlusion at the growth point, making the point cloud reconstruction effect at the growth point unsatisfactory and affecting the size of the bounding box. As shown in Table 2, the mean absolute percentage error of the measurement values obtained by the machine vision system is relatively large. This is mainly because the size of Atractylodes chinensis is small, and the main source of the system measurement error is the certain deviation between the selected points of the basal diameter and the manually selected points. In addition, the deviation of the selected points of the basal diameter is also the main reason for the error, and the interference of the noise at the edge of the basal diameter will also cause the measurement values to be generally larger than the manually measured values. Compared with the manually measured values, the leaf area values extracted by this system are generally on the low side. There are wrinkles and burr structures on the leaf surface of the plant. The three-dimensional point cloud of the plant generated by the SFM and CMVS algorithms cannot well reflect these details, and fitting by the least squares method will smooth the leaf point cloud, resulting in a smaller calculated value of the leaf area.

[0057] Table 2 Estimation of the errors between the manual measurement and the machine vision measurement of each phenotypic parameter of Atractylodes chinensis

[0058] In this embodiment, different soil moisture contents have an impact on the reflectance spectrum of the plant canopy.

[0059] Specifically, select annual plants with consistent growth status and good development (Atractylodes chinensis seedlings as the experimental materials), and carry out planting experiments under controlled conditions. The bulk density of the planting soil is 1.58 g / cm3, pH is 7.7, organic matter content is 3.46 g / kg, available nitrogen content is 62.6 mg / kg, available phosphorus content is 20.3 mg / kg, and available potassium content is 112.2 mg / kg. During this period, ensure that the cultivation conditions are consistent. At the flowering stage, randomly divide them into two treatments: a moisture control group (MCG) and a control group (CTG). Among them, the moisture control group undergoes natural water consumption to form a multi-level soil moisture gradient, and the control group maintains the soil moisture content unchanged during the experiment by artificial water replenishment (once every 2 days). Measure the reflectance spectrum of the canopy on the 0th, 2nd, 4th, 6th, and 10th days respectively. The results show that as the soil moisture content decreases, the reflectance of the Atractylodes chinensis canopy in the visible light band (450 - 680 nm) gradually increases, and the reflectance in the near-infrared region (740 - 1000 nm) also shows an increasing trend, and the change range is more significant than that in the visible light band. The reflectance spectrum curve of the Atractylodes chinensis in the control treatment does not change significantly.

[0060] Corresponding to the plant hyperspectral data processing method in the above embodiment, Figure 2The following is a structural block diagram of a plant hyperspectral data processing device provided by an embodiment of the present disclosure. For ease of description, only parts related to the embodiments of the present disclosure are shown. Refer to Figure 2 , the plant hyperspectral data processing device 20 includes: a data processing module 21 and an irrigation strategy determination module 22.

[0061] Among them, the data processing module 21 is used to preprocess the first data and the second data to obtain the third data. The first data is plant canopy structure data, and the second data is plant spectral characteristic data; the irrigation strategy determination module 22 is used to input the third data into the first neural network model to obtain a prediction result, and determine the corresponding target irrigation strategy according to the prediction result; the first neural network model is a neural network model for obtaining the plant growth trend prediction value and the water demand prediction value based on the third data.

[0062] In an embodiment of the present disclosure, the data processing module 21 is specifically used to: obtain a two-dimensional image sequence of the plant through a visible light camera, perform three-dimensional point cloud reconstruction to obtain plant canopy structure data; sense hyperspectral data at the plant canopy level through a hyperspectral sensor to obtain plant spectral characteristic data.

[0063] In an embodiment of the present disclosure, the data processing module 21 is specifically used to: perform temporal and spatial alignment on the plant canopy structure data and the plant spectral characteristic data to generate temporally and spatially matched fusion data; Perform multi-modal feature extraction on the fusion data to obtain the third data.

[0064] In an embodiment of the present disclosure, the data processing module 21 is specifically used to: adopt a multi-modal feature extraction technique to respectively extract the plant canopy structure features of the plant canopy structure data and the spectral features of the plant spectral characteristic data; splice the plant canopy structure features and the spectral features to generate the third data.

[0065] In an embodiment of the present disclosure, the irrigation strategy determination module 22 is specifically used to: before determining the target irrigation strategy, perform weighted fusion on the prediction result and the environmental features to obtain a first weighted value; in response to the first weighted value being less than the first threshold, determine that the plant does not need irrigation; in response to the first weighted value being greater than or equal to the first threshold, determine that the plant needs irrigation; the environmental features include soil humidity, light intensity, and air temperature and humidity.

[0066] In an embodiment of the present disclosure, the irrigation strategy determination module 22 is specifically used to: The target irrigation strategy includes: irrigation method; In response to the water demand prediction value being less than the second threshold, determine that the irrigation method is the first irrigation method; When the predicted value of water demand is greater than or equal to the second threshold, determine that the irrigation method is the second irrigation method; the water supply methods of the first irrigation method and the second irrigation method are different.

[0067] In an embodiment of the present disclosure, the irrigation strategy determination module 22 is specifically configured to: The target irrigation strategy further includes: irrigation amount; Perform weighted fusion on the predicted value of growth trend and the predicted value of water demand; obtain the second weighted value; When the weighted value is less than the third threshold, determine that the irrigation amount is the first irrigation amount; When the weighted value is greater than or equal to the third threshold, determine that the irrigation amount is the second irrigation amount; the first irrigation amount is less than the second irrigation amount.

[0068] In one embodiment, a robot includes a wheeled chassis, a main body frame, a solar power module, a GPS navigation module, a visible light camera, a hyperspectral determination sensor, a data transmission module, a plant hyperspectral data processing device, and a water regulation module; Among them, the wheeled chassis is used to move on the ground and carry the main body frame; the main body frame is the support structure of the robot and is used to provide an installation basis for other components such as the solar power module, sensors, etc., the solar power module, the GPS navigation module, the visible light camera, the hyperspectral determination sensor, the data transmission module, the plant hyperspectral data processing device, and the water regulation module; the solar power module is used to convert solar energy into electrical energy to provide kinetic energy for the wheeled chassis; the GPS navigation module is used to determine the position, speed, and time information of the robot by receiving satellite signals, and can also enable the robot to travel along a preset path, accurately reach a specified location for operation, and can also be used to record the driving trajectory of the robot for subsequent data analysis and operation evaluation; the visible light camera is used to obtain a two-dimensional image sequence of the plant, perform three-dimensional point cloud reconstruction, and obtain plant canopy structure data; the hyperspectral determination sensor is used to sense the hyperspectral data at the plant canopy level to obtain plant spectral characteristic data; the data transmission module is used to transmit the plant canopy structure data, plant spectral characteristic data, GPS position data, and target irrigation strategy collected by the robot to a remote control center or a data processing platform. The plant hyperspectral data processing device is used to determine the corresponding target irrigation strategy according to the plant canopy structure data and the plant spectral characteristic data; the water regulation module is used to irrigate the target area according to the target irrigation strategy output by the plant hyperspectral data processing device.

[0069] Specifically, the wheeled chassis includes forms such as front-wheel steering and rear-wheel differential drive, two-wheel drive + omnidirectional wheels, and four-wheel drive. The form of front-wheel steering + rear-wheel drive has low cost and simple control, but a large turning radius; the form of two-wheel drive + omnidirectional wheels has strong flexibility and easy algorithm control; the four-wheel drive has strong straight-line walking ability and driving force, but high cost and complex motor control. The main frame is the support structure of the robot, similar to the human skeleton, providing an installation basis for other components such as the solar power module, sensors, etc., and having sufficient strength, stiffness, and stability. The solar power module consists of solar panels, a charge controller, a storage battery, etc. The solar panels are used to convert solar energy into electrical energy, and the charge controller controls the charging process to prevent overcharging or over-discharging of the storage battery; the storage battery is used to store electrical energy and supply power to various components of the robot, enabling the robot to work continuously without an external power source, reducing energy costs, and having the advantages of environmental protection and renewable. The GPS navigation module determines the precise position, speed, and time information of the robot on the earth by receiving satellite signals, providing precise positioning and navigation services for the robot, enabling the robot to travel along a preset path and accurately reach the designated location for operation, and can also be used to record the driving trajectory of the robot for subsequent data analysis and operation evaluation. The visible light camera can capture visible light images of the environment around the robot, providing intuitive visual information, which can be used to identify the morphology, color, growth status, pest and disease signs, etc. of plants. The hyperspectral determination sensor can obtain spectral reflection or radiation information of plants in multiple narrow bands. These detailed spectral data contain physiological, biochemical, and other characteristic information of plants, and can detect information such as the nutritional status, water content, and early symptoms of pests and diseases that are difficult to detect with the naked eye, providing a scientific basis for the precise management of plants. The data transmission module can use wireless communication technologies such as Wi-Fi, 4G / 5G, Bluetooth, etc., or can also use wired communication methods to ensure the stability and timeliness of data transmission, so that the operator can monitor the working status of the robot and the collected data in real time. The water regulation module can include components such as water pumps, sprinklers, solenoid valves, water pipes, etc., and can achieve precise irrigation of plants, reasonably control the water supply according to the actual needs of plants, ensure that plants grow in a suitable water environment, improve the utilization efficiency of water resources, and promote the growth and development of plants.

[0070] See Figure 3 , Figure 3 is a schematic block diagram of an electronic device provided by an embodiment of the present disclosure. As Figure 3The electronic device 300 in the present embodiment shown may include: one or more processors 301, one or more input devices 302, one or more output devices 303, and one or more memories 304. The above-mentioned processors 301, input devices 302, output devices 303, and memories 304 communicate with each other through a communication bus 305. The memory 304 is used to store a computer program, and the computer program includes program instructions. The processor 301 is used to execute the program instructions stored in the memory 304. Among them, the processor 301 is configured to call the program instructions to execute the functions of each module in the above-mentioned device embodiments, for example Figure 2 the functions of the modules 21 to 22 shown.

[0071] It should be understood that in the embodiments of the present disclosure, the so-called processor 301 may be a central processing unit (CPU), and this processor may also be other general-purpose processors, digital signal processors (DSPs), application specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or this processor may also be any conventional processor, etc.

[0072] The input device 302 may include a touchpad, a fingerprint acquisition sensor (for acquiring the fingerprint information and the direction information of the fingerprint of the user), a microphone, etc., and the output device 303 may include a display (such as an LCD), a speaker, etc.

[0073] The memory 304 may include a read-only memory and a random access memory, and provide instructions and data to the processor 301. A part of the memory 304 may also include a non-volatile random access memory. For example, the memory 304 may also store information about the device type.

[0074] In specific implementation, the processors 301, input devices 302, and output devices 303 described in the embodiments of the present disclosure may implement the implementation manners described in the first embodiment and the second embodiment of the plant hyperspectral data processing method provided by the embodiments of the present disclosure, and may also implement the implementation manner of the electronic device described in the embodiments of the present disclosure, which will not be elaborated here.

[0075] In another embodiment of the present disclosure, a computer-readable storage medium is provided. The computer-readable storage medium stores a computer program, and the computer program includes program instructions. When the program instructions are executed by a processor, all or part of the processes in the method of the above embodiment are implemented. It can also be completed by instructing relevant hardware through the computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in the form of source code, object code, executable file or some intermediate form, etc. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard disk, magnetic disk, optical disc, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium, etc.

[0076] The computer-readable storage medium can be the internal storage unit of the electronic device in any of the foregoing embodiments, such as the hard disk or memory of the electronic device. The computer-readable storage medium can also be an external storage device of the electronic device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc. equipped on the electronic device. Further, the computer-readable storage medium can also include both the internal storage unit and the external storage device of the electronic device. The computer-readable storage medium is used to store the computer program and other programs and data required by the electronic device. The computer-readable storage medium can also be used to temporarily store the data that has been output or will be output.

[0077] Those of ordinary skill in the art can realize that the units and algorithm steps of the examples described in combination with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two. To clearly illustrate the interchangeability of hardware and software, the components and steps of the examples have been generally described according to functions in the above description. Whether these functions are executed in a hardware or software manner depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present disclosure.

[0078] Those skilled in the art can clearly understand that for the convenience and brevity of description, the specific working processes of the above-described electronic devices and units can refer to the corresponding processes in the foregoing method embodiments and will not be elaborated herein.

[0079] In several embodiments provided in this application, it should be understood that the disclosed electronic devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of units is only a logical function division. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the displayed or discussed couplings or direct couplings or communication connections to each other can be indirect couplings or communication connections through some interfaces or units, or can be electrical, mechanical or other forms of connection.

[0080] The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or may be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of the embodiments of the present disclosure.

[0081] In addition, each functional unit in various embodiments of the present disclosure can be integrated in a processing unit, or each unit can exist physically alone, or two or more units can be integrated in one unit. The above integrated units can be implemented in the form of hardware or in the form of software functional units.

[0082] The above is only the specific implementation manner of the present disclosure, but the protection scope of the present disclosure is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present disclosure can easily think of various equivalent modifications or substitutions, and these modifications or substitutions should all be covered within the protection scope of the present disclosure. Therefore, the protection scope of the present disclosure should be subject to the protection scope of the claims.

Claims

1. A method for processing plant hyperspectral data, characterized in that: include: Preprocessing the first data and the second data to obtain third data, wherein the first data is plant canopy structure data, and the second data is plant spectral characteristic data; Inputting the third data into the first neural network model to obtain a prediction result, and determining a corresponding target irrigation strategy according to the prediction result; the first neural network model is a neural network model that obtains a growth trend prediction value and a water demand prediction value of the plant according to the third data; Plant hyperspectral data processing methods also include: Before determining the target irrigation strategy, the prediction results are weighted and fused with the environmental characteristics to obtain a first weighted value; Determining whether the plant needs irrigation according to a comparison relationship between the first weighted value and the first threshold; The environmental characteristics include soil moisture, light intensity and air temperature.

2. The plant hyperspectral data processing method according to claim 1, characterized in that: Also includes: The visible light camera is used to obtain a 2D image sequence of the plant, and 3D point cloud reconstruction is performed to obtain plant canopy structure data; The hyperspectral data at the plant canopy level is sensed by a hyperspectral sensor to obtain the plant spectral characteristic data.

3. The plant hyperspectral data processing method according to claim 2, characterized in that: The preprocessing of the first data and the second data to obtain the third data includes: Align plant canopy structure data and plant spectral characteristic data in time and space to generate fused data that matches time and space; Perform multimodal feature extraction on the fused data to obtain third data.

4. The plant hyperspectral data processing method according to claim 3, characterized in that: Performing multimodal feature extraction on the fused data to obtain third data includes: The multimodal feature extraction technology is used to extract the plant canopy structure features of the plant canopy structure data and the spectral features of the plant spectral characteristic data; The plant canopy structure characteristics and the spectral characteristics are spliced ​​to generate third data.

5. The plant hyperspectral data processing method according to claim 1, characterized in that: The step of determining whether the plant needs irrigation based on the comparison relationship between the first weighted value and the first threshold value includes: In response to the first weighted value being less than a first threshold, determining that the plant does not require irrigation; In response to the first weighted value being greater than or equal to a first threshold, it is determined that the plant needs irrigation.

6. The method for processing plant hyperspectral data according to claim 5, characterized in that: Determining a corresponding target irrigation strategy according to the prediction result includes: In response to the predicted water demand value being less than a second threshold, determining the irrigation mode to be a first irrigation mode; In response to the predicted water demand value being greater than or equal to a second threshold, determining the irrigation method to be a second irrigation method; the first irrigation method and the second irrigation method have different water supply methods.

7. The method for processing plant hyperspectral data according to claim 6, characterized in that: Determining a corresponding target irrigation strategy according to the prediction result also includes: Performing weighted fusion on the growth trend prediction value and the water demand prediction value to obtain a second weighted value; In response to the second weighted value being less than a third threshold, determining the irrigation amount to be a first irrigation amount; In response to the second weighted value being greater than or equal to a third threshold, determining the irrigation amount to be a second irrigation amount; The first irrigation amount is smaller than the second irrigation amount.

8. A plant hyperspectral data processing device, characterized in that: include: A data processing module, used for preprocessing the first data and the second data to obtain third data, wherein the first data is plant canopy structure data, and the second data is plant spectral characteristic data; an irrigation strategy determination module, configured to input the third data into a first neural network model to obtain a prediction result, and determine a corresponding target irrigation strategy according to the prediction result; the first neural network model is a neural network model that obtains a plant growth trend prediction value and a water demand prediction value according to the third data; The irrigation strategy determination module is specifically used to, before determining the target irrigation strategy, perform weighted fusion of the prediction result and the environmental characteristics to obtain a first weighted value; Determining whether the plant needs irrigation according to a comparison relationship between the first weighted value and the first threshold; The environmental characteristics include soil moisture, light intensity and air temperature.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images