Device and system for estimating cultivated land yield

Abstract

The invention relates to the technical field of remote sensing detection, in particular to a device and system for estimating cultivated land yield, and the system comprises a spectral data acquisition module, an image data acquisition module, a soil and physiological data acquisition module, an environmental parameter acquisition module, a pest and nutrition deficiency distinguishing module and a yield estimation and correction module. The device comprises a central controller and a power supply assembly for providing detection endurance, wherein the central controller is in signal connection with a spectral data acquisition assembly, an image data acquisition assembly, a soil and physiological data acquisition assembly and an environmental data acquisition assembly. According to the method, diseases and insect pests and nutrition deficiency are accurately distinguished through multi-source data acquisition and cross validation, the yield loss is corrected by dynamically integrating biological stress and environmental factors, and the defects of similar phenotype misjudgment and low estimation precision in the conventional technology are overcome.

Description

Technical Field

[0001] This invention relates to the field of remote sensing technology, and more specifically to a device and system for estimating arable land output. Background Technology

[0002] Accurate farmland yield estimation techniques can optimize irrigation, fertilization, and labor allocation. For example, global rice production is projected to reach a record high of 551.5 million tons in 2025, with significant increases in production in Asia, but demand growth in low-income, food-deficient African countries is expected to reach 2%, highlighting the importance of yield forecasting for resource allocation.

[0003] Conventional yield estimation techniques mainly consist of three parts: a data acquisition component, a data processing module, and a prediction model. The data acquisition component, centered on a multispectral camera (such as the GreenEdge-P multispectral camera) or hyperspectral camera, is mounted on a drone or mobile platform to collect spectral image data in the 400-1000nm band during key crop growth stages (such as the wheat jointing stage and rice grain-filling stage), simultaneously recording environmental parameters such as light intensity and temperature. The data processing module eliminates errors caused by shooting angle and illumination through radiometric and geometric corrections, and extracts key features such as vegetation indices (NDVI, red-edge chlorophyll index) and band reflectance. The prediction model, based on algorithms such as partial least squares regression (PLSR) and support vector machine (SVM), fits spectral features with historical yield data to establish a quantitative relationship model. The implementation process is as follows: multispectral images are acquired along a preset flight path during key crop growth stages; after preprocessing, features such as vegetation indices are extracted; these are input into the model to obtain yield estimates; and the model parameters are then optimized through actual yield verification.

[0004] Current yield estimation models often fail to dynamically integrate real-time pest and disease monitoring data (such as leaf chlorosis caused by aphid feeding and tissue necrosis caused by fungal diseases), making it impossible to quantify the immediate impact of biological stress on yield and leading to an overestimation of potential yield. Furthermore, when pest and disease data are introduced and inferences about crop growth status are based solely on indirect biophysical indicators such as spectral reflectance (e.g., NDVI, red-edge index) or image features (e.g., color moments, texture parameters), it becomes difficult to distinguish between similar phenotypes of pests and diseases and nutrient deficiencies (e.g., both can cause leaf yellowing), leading to misjudgment of stress types and ultimately a systematic underestimation of the actual yield reduction, significantly decreasing yield prediction accuracy. Therefore, it is necessary to propose a device and system for estimating arable land yield to address these problems. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides a device and system for estimating arable land yield. By collecting and cross-validating multi-source data, it accurately distinguishes between pests and diseases and nutrient deficiencies, and dynamically integrates biological stress and environmental factors to correct yield losses, thus overcoming the shortcomings of conventional technologies such as misjudgment of similar phenotypes and low estimation accuracy.

[0006] To achieve the above objectives, the technical solution of the present invention is as follows: a system for estimating arable land yield, comprising a spectral data acquisition module, an image data acquisition module, a soil and physiological data acquisition module, an environmental parameter acquisition module, a pest and nutrient deficiency differentiation module, and a yield estimation and correction module;

[0007] The spectral data acquisition module is used to collect spectral reflectance data and spectral vegetation index data of crop canopy or leaves, and to calculate vegetation index and characteristic band parameters.

[0008] The image data acquisition module is used to acquire image data and infrared thermal imaging data of crop leaves, to preliminarily locate the lesion areas on the surface of crop leaves based on the image data, and to generate a leaf surface temperature distribution map based on the infrared thermal imaging data.

[0009] The soil and physiological data acquisition module is used to collect soil nutrient data and crop physiological data of cultivated land, calculate the soil fertility index based on the soil nutrient data, and generate crop nutrient status parameters based on the crop physiological data.

[0010] The environmental parameter acquisition module is used to collect farmland environmental data and receive real-time weather station data from the farmland location. Based on the farmland environmental data, it calculates crop growth environment suitability index and pest and disease transmission potential assessment parameters.

[0011] The module for distinguishing between pests and nutrient deficiencies is used to call upon a pest and nutrient feature library and a nutrient deficiency threshold model pre-input by the operator. First, based on the preliminary localization of lesion areas, fine-grained features of lesions are extracted from the image data. After matching with the pest and nutrient feature library, the module outputs the pest and nutrient type and the severity index. Simultaneously, soil fertility index and crop nutrient status parameters are input into the nutrient deficiency threshold model to determine the type and degree of nutrient deficiency. Then, based on the pest and nutrient type, the module identifies whether the crop deterioration symptoms caused by the pest and nutrient are similar to the symptoms of nutrient deficiency in the same crop. When similar symptoms are identified, cross-validation is performed based on spectral reflectance data and leaf temperature distribution map to output the quantitative results of pest and nutrient probability and nutrient deficiency probability. The pest and nutrient severity index and nutrient deficiency degree are adjusted according to the quantitative results of pest and nutrient probability and nutrient deficiency probability.

[0012] The yield estimation and correction module is used to first calculate the basic theoretical yield, calculate the yield loss coefficient of pests and diseases based on the type and severity index of pests and diseases, then calculate the yield loss coefficient of nutrient deficiency based on the type and degree of nutrient deficiency, and correct the theoretical yield value by combining the yield loss coefficient of pests and diseases and the yield loss coefficient of nutrient deficiency, and calculate the final yield estimate.

[0013] Furthermore, in the spectral data acquisition module, the characteristic band parameters are calculated by extracting the characteristic band absorption peaks caused by pests and diseases through the continuous spectral reflectance of the crop canopy or leaves.

[0014] Furthermore, the spectral vegetation index data collected by the spectral data acquisition module includes the normalized vegetation index, enhanced vegetation index, and red-edged chlorophyll index.

[0015] Furthermore, the preliminary localization of lesion areas in the image data acquisition module specifically involves extracting the texture and color features of lesions from the image data of crop leaves.

[0016] Furthermore, in the soil and physiological data acquisition module, crop nutrient status parameters include estimated nitrogen content and water stress index.

[0017] Furthermore, in the environmental parameter acquisition module, the farmland environmental data includes air temperature, relative humidity, light intensity, CO2 concentration, and wind speed;

[0018] Real-time weather station data includes daily precipitation, daily maximum temperature, daily minimum temperature, and severe weather warnings.

[0019] Furthermore, in the module for distinguishing between pests and diseases and nutritional deficiencies, the pest and disease feature library is constructed based on image feature templates and spectral fingerprints of common pests and diseases;

[0020] The nutrient deficiency threshold model uses the random forest algorithm. The input features include soil fertility index and crop nutrient status parameters, and the output is the type and degree of nutrient deficiency.

[0021] Furthermore, in the module distinguishing between pests and diseases and nutrient deficiencies, the specific operation of cross-validation based on spectral reflectance data is as follows:

[0022] The red edge position offset, visible light band reflectance standard deviation, near-infrared band reflectance mean, and disease-sensitive band ratio are extracted from the spectral reflectance data. The extracted parameters are then input into a pre-trained spectral discrimination model. Based on the spectral feature distribution of pests and diseases and nutrient deficiencies in historical data, the spectral discrimination model outputs the probability of pests and diseases and the probability of nutrient deficiency.

[0023] Furthermore, in the module distinguishing between pests / diseases and nutrient deficiencies, the specific operation of cross-validation based on leaf temperature distribution maps is as follows:

[0024] Extract the average leaf temperature, temperature standard deviation, area ratio of high-temperature region, and morphological characteristics of high-temperature region from the leaf temperature distribution map. High-temperature region is the region where the temperature is greater than the average temperature +2℃.

[0025] Several temperature parameters are input into the temperature differentiation model, which outputs the probability of pests and diseases and the probability of nutrient deficiency based on the temperature characteristics distribution of pests and diseases and nutrient deficiency in historical data.

[0026] The above approach has the following beneficial effects:

[0027] 1. Conventional techniques do not dynamically integrate real-time pest and disease monitoring data, making it impossible to quantify the immediate impact of biological stress on yield and easily leading to overestimation of potential yield. This solution uses a pest and disease differentiation module to identify the type and severity of pests and diseases in real time, and combines this with a yield estimation and correction module to calculate the yield loss coefficient caused by pests and diseases. This enables the quantification of the immediate impact of biological stress on yield, overcoming the overestimation caused by conventional techniques ignoring the immediate impact of pests and diseases.

[0028] 2. This solution collects data collaboratively from multiple modules including spectroscopy, imaging, soil physiology, and environment. It uses cross-validation of spectral reflectance and leaf temperature distribution for similar symptoms to quantify the probability of pests and diseases and nutrient deficiencies, accurately distinguish stress types, and solve the problem of underestimating the extent of yield reduction due to misjudgment of similar phenotypes of pests and diseases and nutrient deficiencies in yield estimation.

[0029] 3. Conventional techniques often rely on fixed empirical values ​​to estimate yield loss, which deviates significantly from actual losses. This solution dynamically calculates the yield loss coefficients due to pests and diseases and nutrient deficiencies based on the severity index and nutrient deficiency level, rather than depending on static empirical values. This makes the yield loss assessment more closely reflect the actual damage status of the crops and improves the accuracy of the estimation.

[0030] 4. This solution integrates real-time farmland environmental data and meteorological station data through an environmental parameter acquisition module, calculates crop growth environment suitability index and pest and disease transmission potential assessment parameters, incorporates the indirect impact of the environment on yield into the estimation model, and solves the defect of one-sided estimation caused by missing environmental factors.

[0031] 5. In this solution, the spectral data acquisition module extracts the characteristic band absorption peaks caused by pests and diseases through continuous spectral reflectance, providing specific physical evidence for distinguishing between pests and diseases and nutritional deficiencies, thus making up for the problem that conventional techniques have strong spectral characteristics but insufficient distinguishability.

[0032] 6. The image data acquisition module extracts the texture and color features of lesions and generates a leaf surface temperature distribution map by combining infrared thermal imaging. Through the dual features of spatial morphology and temperature distribution, it distinguishes between "local tissue damage" and "overall metabolic abnormality", making up for the shortcomings of insufficient spectral spatial resolution and single image information utilization.

[0033] An apparatus for estimating arable land yield, based on the above-mentioned system construction for estimating arable land yield, includes a central controller and a power supply component for providing detection power supply. The central controller is signal-connected to a spectral data acquisition component, an image data acquisition component, a soil and physiological data acquisition component, and an environmental data acquisition component.

[0034] The spectral data acquisition components include a hyperspectral camera and a multispectral sensor;

[0035] The image data acquisition components include an RGB camera and an infrared thermal imager;

[0036] The soil and physiological data acquisition components include a soil nutrient sensor, a chlorophyll meter, and a crop physiological sensor;

[0037] The environmental parameter acquisition components include temperature and humidity sensors, light sensors, and wind speed and direction sensors.

[0038] Beneficial effects: The device is used to achieve integrated control and synchronous data transmission of various data acquisition components, the power supply component ensures continuous detection capability, and the hardware layer provides stable data acquisition support for the system, ensuring the high efficiency of multi-module collaborative work and field applicability.

[0039] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0040] Figure 1 This is a schematic diagram of the general operation of the system in an embodiment of the apparatus and system for estimating arable land yield of the present invention;

[0041] Figure 2 This is a partial operational schematic diagram of data acquisition in an embodiment of the device and system for estimating arable land yield according to the present invention;

[0042] Figure 3 This is a partial operational schematic diagram of the device and system embodiment for estimating arable land yield of the present invention, showing the distinction between pests and diseases and nutrient deficiencies;

[0043] Figure 4 This is a partial operational schematic diagram of the yield estimation and correction in an embodiment of the device and system for estimating arable land yield according to the present invention. Detailed Implementation

[0044] The technical solution of the present invention will now be clearly and completely described with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.

[0045] In the description of this invention, it should be noted that the terms "center," "upper," "lower," "left," "right," "vertical," "horizontal," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," and "third" are used for descriptive purposes only and should not be construed as indicating or implying relative importance.

[0046] In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "linking" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium; and they can refer to the internal connection of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0047] The following detailed description illustrates the specific implementation method:

[0048] Example 1:

[0049] This solution provides a system for estimating arable land yield, such as Figure 1 As shown, this system includes a spectral data acquisition module, an image data acquisition module, a soil and physiological data acquisition module, an environmental parameter acquisition module, a pest and nutrient deficiency differentiation module, and a yield estimation and correction module.

[0050] Combination Figure 1 and Figure 2As shown, the spectral data acquisition module collects spectral reflectance data (400-1000nm continuous band) and spectral vegetation index data of crop canopy or leaves, and calculates vegetation indices and characteristic band parameters. Among them, vegetation indices include normalized difference vegetation index (NDVI), enhanced vegetation index (EVI), and red-edge chlorophyll index (CIred-edge). The characteristic band parameters are extracted from the continuous spectral reflectance of crop canopy or leaves, specifically: identifying characteristic band absorption peaks caused by pests and diseases (such as the intensity attenuation of the 680nm chlorophyll absorption peak caused by aphid damage, and the abnormal water absorption peak at 960nm caused by fungal diseases), and simultaneously calculating the red-edge position offset (the wavelength corresponding to the maximum value of the first derivative of the reflectance in the 680-760nm band), the standard deviation of the visible light band reflectance (reflecting the uniformity of leaf color), the mean of the near-infrared band reflectance (R800, reflecting the integrity of cell structure), and the ratio of disease-sensitive bands (such as 680nm / 760nm). This design provides a core physical basis for subsequent stress differentiation because spectral reflectance is highly sensitive to changes in crop physiological structure (such as cell integrity) and chemical composition (such as chlorophyll and water). Different stress factors (diseases / pests / nutrient deficiencies) will trigger differentiated responses in characteristic band absorption peaks (such as nutrient deficiency causing overall spectral shift, and diseases and pests causing local band anomalies).

[0051] The image data acquisition module collects image data and infrared thermal imaging data of crop leaves to achieve preliminary location of lesion areas and generate leaf surface temperature distribution maps.

[0052] Preliminary localization of lesion areas is achieved by extracting lesion texture (such as the dotted texture formed by aphid aggregation and the reticulated texture of fungal lesions) and color features (such as the RGB value range of brown spots: R>150, G<100, B<80) based on image data. The lesion outline is located by using a threshold segmentation algorithm (such as the Otsu method), and the coordinates and area percentage of the lesion area are output. Image data can intuitively reflect the spatial morphological characteristics (texture, color, outline) of lesions, making up for the lack of spatial resolution of spectral data.

[0053] The leaf surface temperature distribution map is generated based on infrared thermal imaging data and can intuitively display the leaf surface temperature distribution pattern (such as local high temperature areas caused by pests and diseases, and overall temperature changes caused by nutrient deficiency); infrared thermal imaging can distinguish between "local tissue damage (pests and diseases)" and "overall metabolic abnormalities (nutrient deficiency)" by temperature distribution differences.

[0054] The soil and physiological data acquisition module collects soil nutrient data and crop physiological data from cultivated land, and calculates soil fertility index and crop nutrient status parameters. Since soil nutrient data directly reflects nutrient supply capacity, and crop physiological data (such as SPAD value and water potential) directly reflects nutrient absorption and metabolic status, combining the two can eliminate interfering factors such as "soil nutrients are sufficient but crop absorption is hindered."

[0055] Soil nutrient data includes N / P / K content and pH value, based on which soil fertility index is calculated (e.g., nitrogen supply level: high / medium / low, corresponding to available nitrogen >90mg / kg, 60-90mg / kg and <60mg / kg).

[0056] Crop physiological data include chlorophyll SPAD value and leaf water potential. Based on these, crop nutrient status parameters are generated, specifically the estimated nitrogen content (SPAD value and nitrogen content regression model: nitrogen content = 0.03 × SPAD value - 0.5) and the water stress index (leaf water potential < -1.5 MPa is considered moderate water stress).

[0057] The environmental parameter acquisition module collects farmland environmental data and receives real-time weather station data, calculating crop growth environment suitability index and pest and disease transmission potential assessment parameters. The main purpose of this design is to collect and calculate environmental factors that affect yield estimation. These environmental factors directly affect crop metabolic efficiency (e.g., temperature affects photosynthetic rate) and pest and disease activity (e.g., humidity affects fungal spore germination). Combining short-term field environmental data with long-term weather station data allows for a comprehensive assessment of the direct and indirect impacts of the environment on yield.

[0058] Farmland environmental data includes air temperature (°C), relative humidity (%), light intensity (μmol / m²·s), CO2 concentration (ppm), and wind speed (m / s), which are collected in real time by field sensors.

[0059] Real-time weather station data includes daily precipitation (mm), daily maximum temperature (°C), daily minimum temperature (°C), and severe weather warnings (such as rainstorm and drought warnings), which are received via a wireless communication module.

[0060] Based on the above data, we calculated the crop growth environment suitability index (e.g., the suitable temperature for wheat grain filling is 20-25℃, and the suitability index decreases by 20% when it is higher than 28℃) and pest and disease transmission potential assessment parameters (e.g., the aphid transmission rate increases by 1.5 times when the temperature is 25℃ and the humidity is 60%-70%).

[0061] Because single data sources (such as images or soil data) are prone to misclassification due to similar phenotypes (such as yellowing), multi-source cross-validation (spectral, temperature, and soil-related physiological data) can leverage the differentiated characteristics of different stresses (such as spectral red-edge shift and temperature distribution patterns) to achieve complementarity and improve the reliability of differentiation. Therefore, combining... Figure 1 , Figure 2 and Figure 3 As shown, this embodiment designs a module to distinguish between pests and nutrient deficiencies. Through multi-dimensional data cross-validation, it accurately distinguishes between pests and nutrient deficiencies and outputs quantitative results to correct the severity index of pests and nutrient deficiencies. Specifically:

[0062] First, the pre-input pest and disease feature library (constructed based on image feature templates and spectral fingerprints of common pests and diseases) is called to extract fine-grained features of lesions in the image data (such as lesion texture, color, and edge morphology). After matching, the pest and disease type and pest and disease severity index are output.

[0063] At the same time, the nutrient deficiency threshold model (using the random forest algorithm, inputting soil fertility index and crop nutrient status parameters) is invoked to determine the type of nutrient deficiency (such as nitrogen deficiency, potassium deficiency) and the degree of deficiency.

[0064] When the deterioration symptoms caused by pests and diseases are similar to those of nutrient deficiency, verification is performed based on spectral reflectance data and leaf temperature distribution maps:

[0065] Spectral reflectance verification: Extract the red edge position offset, the standard deviation of reflectance in the visible light band, the mean reflectance in the near-infrared band, and the ratio of disease-sensitive bands, input them into the pre-trained spectral discrimination model, and output the probability of disease and pest and the probability of nutrient deficiency.

[0066] Leaf surface temperature validation: Extract the average leaf surface temperature, temperature standard deviation, area ratio of high-temperature regions (defined as regions with temperature > average temperature + 2℃) and morphological characteristics of high-temperature regions from the leaf surface temperature distribution map, input them into the temperature discrimination model, and output the probability of pests and diseases and the probability of nutrient deficiency.

[0067] Finally, based on the quantitative results of the disease and pest probability and nutrient deficiency probability obtained from cross-validation, the disease and pest severity index and nutrient deficiency degree are adjusted.

[0068] Combination Figure 1 , Figure 2 and Figure 4 As shown, the yield estimation and correction module calculates the final yield estimate based on the basic theoretical yield and incorporating yield loss coefficients due to pests, diseases, and nutrient deficiencies. Specifically:

[0069] Based on crop planting parameters (number of plants per mu, number of effective ears per plant, number of grains per ear, and weight of 1,000 grains, which can be calculated using historical databases or field measurements), the formula is: Basic theoretical yield = number of plants per mu × number of effective ears per plant × number of grains per ear × weight of 1,000 grains × 0.85 (moisture content correction coefficient).

[0070] The yield is determined based on the type and severity index of the pests and diseases. For example, a level 3 severity of wheat aphids (disease area accounting for 40%-60%) corresponds to a loss coefficient of 0.12 (yield loss of 12%). Then, the yield is determined based on the type and degree of nutrient deficiency. For example, a mild nitrogen deficiency (estimated nitrogen content <2.5%) corresponds to a loss coefficient of 0.05 (yield loss of 5%). The final basic theoretical yield is calculated as follows: (1 - yield loss coefficient due to pests and diseases) × (1 - yield loss coefficient due to nutrient deficiency). Simultaneously, environmental factors are incorporated for fine-tuning (e.g., when the average daily temperature during the grain-filling period is >30℃, the yield decreases by 1% for every 1℃ increase). The final yield estimate is then output.

[0071] The following experiments were conducted based on the system proposed in this embodiment:

[0072] (1) Experimental objective:

[0073] [1] Accurately distinguish between similar phenotypes of pests and diseases and nutrient deficiencies (such as leaf yellowing).

[0074] [2] Dynamically quantify the impact of biological stress (diseases and pests) and abiotic stress (nutrient deficiency) on yield, and reduce yield estimation error.

[0075] (2) Experimental materials and design

[0076] [1] Test crop: winter wheat (variety: Jimai 44, growth period: grain-filling stage)

[0077] [2] Experimental groups (each group was repeated 3 times, and the plot area was 10m²):

[0078] Group Handling method Control group (CK) Healthy plants, normal water and fertilizer management, free from pests and diseases. Pests and diseases group (P) <![CDATA[Artificially inoculated aphids (density 5 aphids / plant) + wheat leaf rust pathogen (spore concentration 1×10 5 per mL)]]> Nutritional deficiency group (N) Nitrogen deficiency treatment (soil available nitrogen content 40 mg / kg, normal control 90 mg / kg) resulted in no pest or disease infection. Mixed stress group (P+N) Aphids + leaf rust infection (same as group P) + nitrogen deficiency treatment (same as group N)

[0079] [3] Experiment period: April 1, 2025 (initial stage of grouting) - May 20, 2025 (final stage of grouting)

[0080] (3) Experimental steps

[0081] [1] Data Acquisition

[0082] The system data acquisition described in this embodiment:

[0083] Spectral data were collected using the device's spectral module to assess canopy reflectance in the 400-1000 nm band, and to calculate NDVI, red-edge chlorophyll index, and characteristic absorption peaks (680 nm and 960 nm).

[0084] Image data: Leaf images captured by an RGB camera (extracting lesion texture / color features), and leaf surface temperature distribution maps generated by an infrared thermal imager (high temperature areas are defined as >average temperature +2℃).

[0085] Soil and physiological data: Soil nutrient sensors were used to measure available nitrogen content, and chlorophyll meters were used to measure SPAD values ​​(to calculate estimated nitrogen content).

[0086] Environmental data include air temperature (25-30℃), humidity (60-70%), and daily precipitation (5-10mm) from the meteorological station.

[0087] Conventional acquisition techniques (control group): NDVI and red-edge index were acquired using only a multispectral camera, and yield was estimated based on the traditional PLSR model.

[0088] [2] Data analysis and yield estimation

[0089] The system data acquisition and analysis described in this embodiment:

[0090] The module for distinguishing between pests and diseases and nutrient deficiencies outputs the type of pest or disease (aphids, leaf rust browning RGB values) and the degree of nutrient deficiency (N group: moderate nitrogen deficiency; P+N group: mild nitrogen deficiency) by matching lesion features (aphid dotted texture, leaf rust browning RGB values) and cross-validating the spectrum and temperature (red edge offset, high temperature area percentage).

[0091] Yield estimation module: Calculates the basic theoretical yield (based on parameters such as number of plants per acre and thousand-grain weight), combines the disease and pest loss coefficient (P group: 0.15; P+N group: 0.12) and the nutrient deficiency loss coefficient (N group: 0.08; P+N group: 0.05), and outputs the final estimated value.

[0092] Conventional technical analysis: Directly input NDVI and red edge index into the PLSR model to output production estimates.

[0093] [3] Actual yield determination

[0094] The actual yield of each plot was measured at harvest time (after drying, the yield was converted to kg / mu).

[0095] (4) Experimental data:

[0096] Group Judgment result Estimated yield (kg / mu) Actual yield (kg / mu) Estimation error (in this example) Estimation error (conventional technique) CK Free from pests and diseases, and free from nutrient deficiencies 580 590 -1.7% -3.4% P (pests and diseases) Aphids (severity 0.6) + Leaf rust (0.5) 493 485 +1.6% +6.2% (overestimated) N (deficient element) Moderate nitrogen deficiency (severity 0.4) 532 520 +2.3% -5.1% (underestimation) P+N (mixture) Aphids (0.5) + Mild nitrogen deficiency (0.2) 468 470 -0.4% +3.3% (Misdiagnosed as nutrient deficiency due to pests and diseases)

[0097] (5) Experimental conclusions:

[0098] [1] Solving the problem of misjudgment due to similar phenotypes:

[0099] The system described in this embodiment distinguishes between pests and diseases and nutrient deficiencies in the mixed stress group (P+N) through spectral-temperature cross-validation (e.g., aphids cause attenuation of the 680nm chlorophyll absorption peak, and leaf rust causes a local high temperature zone to account for 15%). It effectively distinguishes between "local damage caused by pests and diseases" and "overall metabolic abnormalities due to nutrient deficiency", thus solving the defect of conventional technology that "relies on a single spectral index to misjudge similar phenotypes".

[0100] [2] Dynamically correct production losses to avoid overestimation / underestimation:

[0101] Pests and diseases group (P): The system estimation error described in this embodiment (+1.6%) is significantly lower than that of conventional technology (+6.2%), proving that by monitoring the severity of pests and diseases in real time and calculating the dynamic loss coefficient, the problem of "overestimating yield due to ignoring biological stress" is solved.

[0102] Nutrient deficiency group (N): The system estimation error described in this embodiment (+2.3%) is lower than that of conventional technology (-5.1%), indicating that by using soil-physiological data to collaboratively determine the degree of nutrient deficiency, the problem of "misjudging nutrient deficiency leading to an underestimation of the yield reduction" has been solved.

[0103] Mixed stress group (P+N): The system estimation error described in this embodiment is only -0.4%, which verifies the effect of multi-source data cross-validation on improving the accuracy of production estimation under complex stress.

[0104] [3] Improved overall accuracy: The average estimation error (1.5%) of the system described in this embodiment for all processing groups is lower than that of conventional technology, which proves that it effectively solves the core defects in the background technology and provides a reliable technical means for accurate farmland yield estimation.

[0105] Example 2:

[0106] An apparatus for estimating arable land yield, based on the system architecture for estimating arable land yield described in Embodiment 1, includes a central controller as a core control unit, the central controller being configured with a power supply component for providing continuous detection power, and the central controller being connected to a spectral data acquisition component, an image data acquisition component, a soil and physiological data acquisition component, and an environmental data acquisition component via a CAN bus signal.

[0107] The spectral data acquisition component includes a hyperspectral camera and a multispectral sensor, used to simultaneously acquire spectral reflectance and vegetation index data of crop canopy / leaf.

[0108] The image data acquisition components include an RGB camera and an infrared thermal imager. The RGB camera and the infrared thermal imager work together to locate the diseased areas on crop leaves and monitor the temperature distribution on the leaf surface.

[0109] The soil and physiological data acquisition components include a soil nutrient sensor for detecting N / P / K content and pH value in cultivated land, a chlorophyll meter for measuring SPAD value, and a crop physiological sensor for collecting and monitoring changes in stem diameter and leaf water potential, enabling real-time acquisition of soil fertility and crop nutrient status.

[0110] The environmental parameter acquisition components include temperature and humidity sensors, light sensors, and wind speed and direction sensors. These sensors are used to collect data on environmental factors such as field air temperature and humidity, light intensity, and wind speed and direction, providing hardware-level data support for the system's subsequent differentiation between pests and diseases and nutrient deficiencies, as well as yield estimation.

[0111] Obviously, the above embodiments are merely illustrative examples for clear explanation and are not intended to limit the implementation. Those skilled in the art will recognize that other variations or modifications can be made based on the above description. It is neither necessary nor possible to exhaustively list all possible implementations here. However, obvious variations or modifications derived therefrom are still within the scope of protection of this invention.

Claims

1. A system for estimating arable land yield, characterized in that, It includes a spectral data acquisition module, an image data acquisition module, a soil and physiological data acquisition module, an environmental parameter acquisition module, a pest and nutrient deficiency differentiation module, and a yield estimation and correction module; The spectral data acquisition module is used to collect spectral reflectance data and spectral vegetation index data of crop canopy or leaves, and to calculate vegetation index and characteristic band parameters. The image data acquisition module is used to acquire image data and infrared thermal imaging data of crop leaves, to preliminarily locate the lesion areas on the surface of crop leaves based on the image data, and to generate a leaf surface temperature distribution map based on the infrared thermal imaging data. The soil and physiological data acquisition module is used to collect soil nutrient data and crop physiological data of cultivated land, calculate the soil fertility index based on the soil nutrient data, and generate crop nutrient status parameters based on the crop physiological data. The environmental parameter acquisition module is used to collect farmland environmental data and receive real-time weather station data from the farmland location. Based on the farmland environmental data, it calculates crop growth environment suitability index and pest and disease transmission potential assessment parameters. The module for distinguishing between pests and diseases and nutrient deficiencies is used to call the pest and disease feature library and nutrient deficiency threshold model pre-input by the operator. First, it performs preliminary localization based on the lesion area, extracts fine-grained features of lesions from the image data, and outputs the pest and disease type and severity index after matching with the pest and disease feature library. At the same time, it inputs the soil fertility index and crop nutrient status parameters into the nutrient deficiency threshold model to determine the type and degree of nutrient deficiency. Then, based on the type of pests and diseases, it is identified whether the crop deterioration symptoms caused by pests and diseases are similar to the symptoms of nutrient deficiency in the same crop. When similar symptoms are identified, cross-validation is performed based on spectral reflectance data and leaf temperature distribution map to output the quantitative results of pest and disease probability and nutrient deficiency probability. The severity index of pests and diseases and the degree of nutrient deficiency are adjusted according to the quantitative results of pest and disease probability and nutrient deficiency probability. The yield estimation and correction module is used to first calculate the basic theoretical yield, calculate the yield loss coefficient of pests and diseases based on the type and severity index of pests and diseases, then calculate the yield loss coefficient of nutrient deficiency based on the type and degree of nutrient deficiency, and correct the theoretical yield value by combining the yield loss coefficient of pests and diseases and the yield loss coefficient of nutrient deficiency, and finally calculate the final yield estimate.

2. The system for estimating arable land yield according to claim 1, characterized in that, In the spectral data acquisition module, the characteristic band parameters are calculated by extracting the characteristic band absorption peaks caused by pests and diseases through the continuous spectral reflectance of the crop canopy or leaves.

3. The system for estimating arable land yield according to claim 2, characterized in that, The spectral vegetation index data collected by the spectral data acquisition module includes the Normalized Difference Vegetation Index (NDVI), Enhanced Vegetation Index (EDI), and Red-edged Chlorophyll Index (CRI).

4. The system for estimating arable land yield according to claim 3, characterized in that, The initial localization of lesion areas in the image data acquisition module is specifically achieved by extracting the texture and color features of lesions from the image data of crop leaves.

5. The system for estimating arable land yield according to claim 4, characterized in that, In the soil and physiological data acquisition module, crop nutrient status parameters include estimated nitrogen content and water stress index.

6. The system for estimating arable land yield according to claim 5, characterized in that, In the environmental parameter acquisition module, farmland environmental data includes air temperature, relative humidity, light intensity, CO2 concentration, and wind speed; Real-time weather station data includes daily precipitation, daily maximum temperature, daily minimum temperature, and severe weather warnings.

7. The system for estimating arable land yield according to claim 6, characterized in that, In the module for distinguishing between pests and diseases and nutrient deficiencies, the pest and disease feature library is constructed based on image feature templates and spectral fingerprints of common pests and diseases; The nutrient deficiency threshold model uses the random forest algorithm. The input features include soil fertility index and crop nutrient status parameters, and the output is the type and degree of nutrient deficiency.

8. The system for estimating arable land yield according to claim 7, characterized in that, In the module distinguishing between pests and diseases and nutrient deficiencies, the specific operation of cross-validation based on spectral reflectance data is as follows: The red edge position offset, visible light band reflectance standard deviation, near-infrared band reflectance mean, and disease-sensitive band ratio are extracted from the spectral reflectance data. The extracted parameters are then input into a pre-trained spectral discrimination model. Based on the spectral feature distribution of pests and diseases and nutrient deficiencies in historical data, the spectral discrimination model outputs the probability of pests and diseases and the probability of nutrient deficiency.

9. The system for estimating arable land yield according to claim 8, characterized in that, In the module distinguishing between pests / diseases and nutrient deficiencies, the specific operation of cross-validation based on leaf temperature distribution maps is as follows: Extract the average leaf temperature, temperature standard deviation, area ratio of high-temperature region, and morphological characteristics of high-temperature region from the leaf temperature distribution map. High-temperature region is the region where the temperature is greater than the average temperature +2℃. Several temperature parameters are input into the temperature differentiation model, which outputs the probability of pests and diseases and the probability of nutrient deficiency based on the temperature characteristics distribution of pests and diseases and nutrient deficiency in historical data.

10. An apparatus for estimating arable land yield, based on the system for estimating arable land yield according to any one of claims 1-9, characterized in that, It includes a central controller and a power supply component for providing continuous power for detection. The central controller is signal-connected to the spectral data acquisition component, the image data acquisition component, the soil and physiological data acquisition component, and the environmental data acquisition component. The spectral data acquisition components include a hyperspectral camera and a multispectral sensor; The image data acquisition components include an RGB camera and an infrared thermal imager; The soil and physiological data acquisition components include a soil nutrient sensor, a chlorophyll meter, and a crop physiological sensor; The environmental parameter acquisition components include temperature and humidity sensors, light sensors, and wind speed and direction sensors.

Images