A spectral-based method and system for predicting the deficiency of ammonia, phosphorus, and potassium in rice and its yield.

Abstract

This invention relates to the field of rice cultivation technology, specifically disclosing a spectral-based method and system for predicting rice ammonia, phosphorus, and potassium deficiencies and yield. The method includes the following steps: acquiring spectral data of a rice paddy; correcting the spectral data to obtain the reflectance data of the rice canopy; extracting spectral characteristic parameters related to the nutrient status of ammonia, phosphorus, and potassium based on the rice canopy reflectance data; performing spatial analysis on the spectral characteristic parameters to obtain spatial distribution characteristics of the spectral characteristic parameters relative to the rice paddy; combining the numerical values ​​of the spectral characteristic parameters with the spatial distribution characteristics to obtain nutrient deficiency diagnosis results; and predicting the yield of the rice paddy based on the nutrient deficiency diagnosis results. This method solves the problem of inaccurate diagnosis caused by easily confused spectral information in existing technologies, improves the accuracy of nutrient deficiency diagnosis, and thus enhances the accuracy of rice yield prediction.

Description

Technical Field

[0001] This application relates to the field of rice cultivation technology, and more specifically, to a method and system for predicting the deficiency of amino, phosphorus, and potassium in rice and its yield based on spectral analysis. Background Technology

[0002] As an important food crop, rice growth monitoring technology is developing towards non-contact and non-destructive methods. By using remote sensing spectroscopy to acquire the reflectance data of the rice canopy, vegetation indices and spectral characteristic parameters can be calculated. These parameters are closely related to the physiological and biochemical indicators of the plant and can be used to diagnose nutrient deficiencies such as ammonia, phosphorus, and potassium, and to predict yield.

[0003] However, in actual farmland environments, rice growth is affected by a variety of factors. Besides nutrient status, biotic and abiotic stresses such as diseases, pests, and drought are also prevalent. These stresses can also lead to changes in the physiological and biochemical state of rice plants, resulting in alterations in the canopy spectral response. The problem is that different types of stresses may exhibit similar spectral characteristics, leading to spectral confusion. This confusion makes it difficult for nutrient deficiency diagnostic models based solely on spectral data to accurately distinguish between genuine nutrient problems and anomalies caused by other stresses. The model may incorrectly identify spectral anomalies caused by non-nutrient stresses as nutrient deficiencies, leading to erroneous diagnoses and inappropriate fertilization recommendations. Conversely, when multiple stresses coexist, the spectral responses of interfering factors such as diseases may mask the characteristics of nutrient deficiencies, causing missed diagnoses and hindering accurate yield predictions, ultimately negatively impacting the healthy growth and yield potential of rice.

[0004] There is currently no effective technical solution to the above problems. Summary of the Invention

[0005] The purpose of this application is to provide a spectral-based method and system for predicting the deficiency of amino, phosphorus, and potassium in rice and its yield, in order to solve the problem of inaccurate diagnosis caused by the confusion of spectral information, improve the accuracy of nutrient deficiency diagnosis, and thus improve the accuracy of rice yield prediction.

[0006] Firstly, this application provides a spectral-based method for predicting the deficiency of amino, phosphorus, and potassium in rice and its yield, used for yield prediction in rice fields. The method includes the following steps:

[0007] S1. Obtain spectral data of the paddy field;

[0008] S2. Correct the spectral data to obtain the reflectance data of the rice canopy;

[0009] S3. Extract spectral characteristic parameters of ammonia, phosphorus, and potassium nutrient status based on the reflectance data of the rice canopy;

[0010] S4. Perform spatial analysis on the spectral feature parameters to obtain spatial distribution information of the spectral feature parameters with respect to paddy fields;

[0011] S5. Combining the numerical values ​​of the spectral characteristic parameters and the spatial distribution characteristic information, determine whether the paddy field has nutrient deficiency, and analyze and obtain the nutrient deficiency diagnosis results when nutrient deficiency exists;

[0012] S6. Estimate the yield of the paddy field based on the nutrient deficiency diagnosis results.

[0013] This application's spectral-based method for predicting rice ammonia, phosphorus, and potassium deficiencies and yield combines numerical information of spectral characteristic parameters with spatial distribution characteristics obtained through spatial analysis to determine whether nutrient deficiencies exist in rice fields. This effectively distinguishes between anomalies caused by nutrient deficiencies and spectral anomalies caused by other stress factors, solving the problem of inaccurate diagnosis due to spectral information being easily confused in existing technologies. It improves the accuracy of nutrient deficiency diagnosis, thereby enhancing the precision of rice yield prediction and providing a scientific basis for precise fertilization and management of rice.

[0014] The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectroscopy, wherein the correction process includes radiometric calibration and geometric correction; step S2 includes:

[0015] S21. Perform radiometric calibration and geometric correction on the spectral data to obtain preliminary reflectance data and corresponding location information;

[0016] S22. Based on the reflectance characteristics of pixels of different land cover types, the preliminary reflectance data is filtered to obtain reflectance data for rice canopy pixels.

[0017] The method in this application eliminates errors caused by sensors and observation geometry by radiometric calibration and geometric correction of the original spectral data. Furthermore, by screening based on the pixel reflectance characteristics of different land cover types, interference information from non-rice canopy land cover such as water bodies and soil is effectively removed, resulting in pure rice canopy reflectance data, which improves data quality and provides accurate and reliable basic data for subsequent nutrient diagnosis and yield prediction.

[0018] The method for predicting the deficiency of amino, phosphorus, and potassium in rice based on spectroscopy, wherein the correction process further includes atmospheric correction, and step S2 further includes:

[0019] S23. Perform atmospheric correction on the reflectance data of the rice canopy pixels.

[0020] The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectral analysis, wherein step S2 further includes:

[0021] S24. Obtain the observation geometric information corresponding to the spectral data;

[0022] S25. Based on the observed geometric information and the atmospherically corrected reflectance data, perform bidirectional reflectance distribution characteristic correction to obtain the reflectance data of the rice canopy under the standard observed geometry.

[0023] The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectral analysis, wherein step S3 includes:

[0024] S31. Based on the rice growth period information and / or variety information of the paddy field, obtain the range or combination of target bands related to the ammonia, phosphorus, and potassium nutrient status of rice;

[0025] S32. Based on the reflectance data of the rice canopy and the band range or band combination, obtain spectral characteristic parameters regarding the nutrient status of ammonia, phosphorus, and potassium. The spectral characteristic parameters include vegetation index, reflectance combination value of target band, and absorption peak characteristic value.

[0026] The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectral analysis, wherein step S31 includes:

[0027] S311. Obtain the current growth period information and / or variety information of the paddy field;

[0028] S312. Based on the current growth period information and / or variety information, search for the range or combination of target bands related to the ammonia, phosphorus, and potassium nutrient status from the preset correspondence data.

[0029] The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectral analysis, wherein step S4 includes:

[0030] S41. Obtain the spatial location information corresponding to the spectral feature parameters;

[0031] S42. Based on the spectral feature parameter values ​​and the spatial location information, calculate the global spatial autocorrelation index and the local spatial autocorrelation index of the spectral feature parameters;

[0032] S43. Based on the global spatial autocorrelation index and the local spatial autocorrelation index, obtain the spatial distribution characteristic information of the spectral feature parameters, wherein the spatial distribution characteristic information includes the overall spatial clustering characteristics of the spectral feature parameters and local salient regions.

[0033] The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectral analysis, wherein step S43 includes:

[0034] S431. Based on the value and statistical significance of the global spatial autocorrelation index, determine the type of overall spatial clustering characteristics of the spectral feature parameters;

[0035] S432. Based on the value and statistical significance of the local spatial autocorrelation index, identify the local salient regions of the spectral feature parameters.

[0036] The method for predicting rice ammonia, phosphorus, and potassium deficiency based on spectroscopy, wherein the nutrient deficiency diagnosis results include the deficiency type, deficiency region, and deficiency severity, and step S6 includes:

[0037] S61. Obtain the expected yield of the rice paddy under conditions of no nutrient deficiency;

[0038] S62. Based on preset rules or prediction models, according to the missing type and the severity of the missing, predict the missing output for the corresponding missing area and obtain the output loss value.

[0039] S63. Calculate the predicted yield of the paddy field based on the yield loss value and the expected yield.

[0040] Secondly, this application also provides a spectrum-based system for predicting the deficiency of ammonia, phosphorus, and potassium in rice and its yield, used for yield prediction in rice fields. The system includes:

[0041] The acquisition module is used to acquire spectral data of rice paddies;

[0042] The correction module is used to correct the spectral data to obtain the reflectance data of the rice canopy.

[0043] The feature extraction module is used to extract spectral feature parameters related to the nutrient status of ammonia, phosphorus, and potassium based on the reflectance data of the rice canopy.

[0044] The spatial analysis module is used to perform spatial analysis on the spectral feature parameters to obtain spatial distribution feature information of the spectral feature parameters with respect to paddy fields;

[0045] The diagnostic module is used to determine whether the paddy field has nutrient deficiency by combining the numerical value of the spectral feature parameters and the spatial distribution feature information, and to analyze and obtain the nutrient deficiency diagnosis results when nutrient deficiency exists.

[0046] The estimation module is used to estimate the yield of the paddy field based on the nutrient deficiency diagnosis results.

[0047] The spectral-based rice ammonia, phosphorus, and potassium deficiency and yield prediction system of this application determines whether there is nutrient deficiency in rice fields by combining the numerical information of spectral characteristic parameters with the spatial distribution characteristic information obtained by spatial analysis. This effectively distinguishes between abnormalities caused by nutrient deficiency and spectral abnormalities caused by other stress factors, thereby improving the accuracy of rice yield prediction.

[0048] As can be seen from the above, this application provides a spectral-based method and system for predicting the deficiency of amino, phosphorus, and potassium in rice and its yield. The spectral-based method for predicting the deficiency of amino, phosphorus, and potassium in rice and its yield in this application combines the numerical information of spectral characteristic parameters with the spatial distribution characteristic information obtained from spatial analysis to determine whether there is a nutrient deficiency in the paddy field. This effectively distinguishes between abnormalities caused by nutrient deficiency and spectral abnormalities caused by other stress factors, solves the problem of inaccurate diagnosis caused by the easy confusion of spectral information in the prior art, improves the accuracy of nutrient deficiency diagnosis, and thus improves the accuracy of rice yield prediction, providing a scientific basis for precision fertilization and management of rice. Attached Figure Description

[0049] Figure 1 A flowchart of a spectral-based method for estimating the deficiency of amino, phosphorus, and potassium in rice and its yield, provided in an embodiment of this application.

[0050] Figure 2 A schematic diagram of the structure of the spectrum-based rice ammonia, phosphorus, and potassium deficiency and yield prediction system provided in the embodiments of this application.

[0051] Figure reference numerals: 201, Acquisition module; 202, Correction module; 203, Feature extraction module; 204, Spatial analysis module; 205, Diagnosis module; 206, Prediction module. Detailed Implementation

[0052] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of the embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0053] It should be noted that similar reference numerals and letters in the following figures indicate similar items; therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. Furthermore, in the description of this application, terms such as "first," "second," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0054] Firstly, please refer to Figure 1 This application provides a spectral-based method for predicting the deficiency of amino, phosphorus, and potassium in rice and its yield, used for yield prediction in rice fields. The method includes the following steps:

[0055] S1. Obtain spectral data of the paddy field;

[0056] S2. Correct the spectral data to obtain the reflectance data of the rice canopy;

[0057] S3. Extract spectral characteristic parameters of ammonia, phosphorus, and potassium nutrient status based on the reflectance data of the rice canopy;

[0058] S4. Perform spatial analysis on the spectral characteristic parameters to obtain information on the spatial distribution characteristics of the spectral characteristic parameters in relation to paddy fields;

[0059] S5. Combining the numerical value and spatial distribution characteristics of spectral feature parameters, determine whether nutrient deficiency occurs in paddy fields, and analyze and obtain nutrient deficiency diagnosis results when nutrient deficiency exists.

[0060] S6. Estimate the yield of rice paddies based on the nutrient deficiency diagnosis results.

[0061] Specifically, spectral data refers to electromagnetic spectrum information acquired through sensors. This data can be obtained in various ways, such as using spectral sensors mounted on drones, satellites, or ground platforms, to provide raw spectral response information of the rice canopy. Correction processing refers to preprocessing the raw spectral data to eliminate or reduce the influence of external factors (such as atmosphere, sensor characteristics, and observation geometry) on the spectral information, used to obtain reflectance data that accurately reflects the characteristics of the rice canopy. Rice canopy reflectance data refers to the corrected reflectance values ​​of the rice canopy at different wavelengths. It is a physical quantity directly related to the physiological and biochemical state of rice and is mainly used as the basis for extracting spectral characteristic parameters. Spectral characteristic parameters are indicators related to the nutrient status of rice extracted from the rice canopy reflectance data. These can take the form of vegetation indices (such as NDVI, EVI), reflectance combinations of specific bands, or absorption peak characteristic values, and are mainly used to quantify the nutrient status information of rice. Spatial analysis refers to the analysis of the spatial distribution of spectral characteristic parameters within paddy fields using tools such as Geographic Information Systems (GIS). Methods include spatial autocorrelation analysis, cluster analysis, and hotspot analysis, primarily used to reveal the spatial variation patterns and characteristics of spectral characteristic parameters. Spatial distribution characteristic information refers to the spatial distribution characteristics of spectral characteristic parameters obtained through spatial analysis. This includes overall spatial aggregation characteristics (e.g., clustering, dispersion, or randomness) and locally significant areas (e.g., high- or low-value clusters), primarily used to provide spatial clues about potential nutrient deficiencies. Combining the numerical magnitude of spectral characteristic parameters with spatial distribution characteristic information involves simultaneously considering whether the values ​​of spectral characteristic parameters are abnormal and whether abnormal areas exhibit specific spatial distribution patterns. This is mainly used to distinguish between anomalies caused by nutrient deficiencies and those caused by other localized stresses, improving diagnostic accuracy. Nutrient deficiency diagnostic results refer to relevant data after determining the presence of nutrient deficiencies in the paddy field, primarily used to provide detailed information for subsequent yield forecasting and precision management. Yield forecasting refers to the process of estimating yields based on nutrient deficiency diagnoses. This can be achieved using loss value-based calculations or regression methods based on predictive models, and is primarily used to guide agricultural production decisions.

[0062] Specifically, the method of this application first acquires the raw spectral data of the paddy field, which includes the reflectance information of the rice canopy in different wavelength bands. Next, the raw spectral data undergoes necessary correction processing to eliminate the influence of the external environment and the sensor itself, obtaining reflectance data that truly reflects the characteristics of the rice canopy. Based on this reflectance data, spectral characteristic parameters related to the nutrient status of rice (ammonia, phosphorus, and potassium) are extracted; these parameters are sensitive indicators reflecting the nutrient level of rice. Subsequently, spatial analysis is performed on the extracted spectral characteristic parameters to obtain their spatial distribution characteristics throughout the paddy field, revealing their spatial aggregation or dispersion patterns. When judging nutrient deficiency, it no longer relies solely on whether the values ​​of the spectral characteristic parameters are below a threshold, but rather combines the parameter values ​​with the spatial distribution characteristics obtained through spatial analysis. For example, only when low-value areas exhibit specific spatial aggregation patterns is it more likely to be diagnosed as nutrient deficiency. This combined judgment method can effectively exclude random or localized spectral anomalies caused by non-nutrient stresses such as sporadic pests and diseases or localized drought, thereby more accurately identifying the true nutrient deficiency areas, types, and severity. Finally, based on the accurate nutrient deficiency diagnosis results, the yield of rice fields was estimated, and a more reliable yield prediction was obtained by taking into account the impact of nutrient deficiency on yield.

[0063] This application's spectral-based method for predicting rice ammonia, phosphorus, and potassium deficiencies and yield combines numerical information of spectral characteristic parameters with spatial distribution characteristics obtained through spatial analysis to determine whether nutrient deficiencies exist in rice fields. This effectively distinguishes between anomalies caused by nutrient deficiencies and spectral anomalies caused by other stress factors, solving the problem of inaccurate diagnosis due to spectral information being easily confused in existing technologies. It improves the accuracy of nutrient deficiency diagnosis, thereby enhancing the precision of rice yield prediction and providing a scientific basis for precise fertilization and management of rice.

[0064] In some preferred embodiments, the correction process includes radiometric calibration and geometric correction; step S2 includes:

[0065] S21. Perform radiometric calibration and geometric correction on the spectral data to obtain preliminary reflectance data and corresponding location information;

[0066] S22. Based on the reflectance characteristics of pixels of different land cover types, preliminary reflectance data are screened to obtain reflectance data for rice canopy pixels.

[0067] Specifically, different land cover types include rice canopy pixels, water body pixels, and soil pixels. Rice canopy pixels refer to ground pixels mainly covered by the canopy of rice plants. Water body pixels refer to ground pixels mainly covered by water bodies. Soil pixels refer to ground pixels mainly covered by soil.

[0068] More specifically, radiometric calibration refers to converting the raw digital signal received by the sensor into physically meaningful radiance or reflectance values. Geometric correction refers to correcting spatial distortions in an image so that points in the image accurately correspond to their actual positions on the ground. This can be achieved using methods based on ground control points or methods based on sensor orbital and attitude parameters.

[0069] More specifically, the reflectance characteristics of pixels of different land cover types refer to the differences in spectral reflectance of different land cover types (such as rice canopy, water body, and soil) in different bands, which can be characterized by analyzing the spectral curves of typical land cover types or calculating spectral indices.

[0070] Specifically, this scheme refines and improves the correction and processing steps for acquiring rice canopy reflectance data. First, radiometric calibration processes the raw digital signal received by the sensor, converting it into physically meaningful radiance or reflectance values. This eliminates the influence of differences in sensor response and ensures data comparability. Geometric correction corrects spatial distortions in the image, ensuring that points in the image accurately correspond to their actual locations on the ground, laying the foundation for subsequent spatial analysis. Building on this, step S22 further filters the preliminary reflectance data by analyzing the differences in spectral reflectance characteristics of different land cover types (such as rice canopy, water bodies, and soil). This effectively removes background interference, ensuring that the final reflectance data originates solely from the rice canopy itself. This precise filtering of rice canopy pixels improves the purity and accuracy of the reflectance data, providing more reliable basic data for extracting spectral feature parameters related to rice nutrient status, thereby contributing to improved accuracy in nutrient deficiency diagnosis and yield prediction.

[0071] Through the above scheme, the method of this application eliminates the errors caused by the sensor and observation geometry by radiometric calibration and geometric correction of the original spectral data; furthermore, by screening based on the pixel reflectance characteristics of different land cover types, the interference information of non-rice canopy land cover such as water bodies and soil is effectively removed, and pure rice canopy reflectance data is obtained, which improves the data quality and provides accurate and reliable basic data for subsequent nutrient diagnosis and yield prediction.

[0072] In some preferred embodiments, the correction process further includes atmospheric correction, and step S2 further includes:

[0073] S23. Perform atmospheric correction on the reflectance data of rice canopy pixels.

[0074] Specifically, atmospheric correction refers to eliminating or reducing the influence of the atmosphere on the electromagnetic wave signals received by remote sensing sensors, converting the apparent reflectance data received by the sensors into the true reflectance data of ground objects. Atmospheric correction can be implemented using various models or algorithms, such as methods based on radiative transfer models (e.g., MODTRAN, 6S models), methods based on image information (e.g., dark target method, histogram matching method), or methods based on ground synchronous measurement data. The specific execution logic will not be elaborated here.

[0075] Specifically, based on step S22, this scheme further adds an atmospheric correction step. By performing atmospheric correction on the selected rice canopy pixel reflectance data, the effects of atmospheric absorption and scattering can be removed, transforming the apparent reflectance data into data that is closer to the true reflectance of the rice canopy. This improves the accuracy of subsequent spectral feature parameter extraction, thereby enhancing the accuracy of nutrient deficiency diagnosis and yield prediction, and solving the problem of inaccurate spectral data caused by atmospheric influence.

[0076] In some preferred embodiments, step S2 further includes:

[0077] S24. Obtain the observation geometric information corresponding to the spectral data;

[0078] S25. Based on the observation geometry information and the atmospherically corrected reflectance data, perform bidirectional reflectance distribution characteristic correction to obtain the reflectance data of the rice canopy under the standard observation geometry.

[0079] Specifically, observation geometry information refers to the relative positional relationship between the remote sensing sensor, the target ground object, and the sun when the remote sensing sensor acquires an image. It can be described using parameters such as solar zenith angle, solar azimuth angle, sensor zenith angle, sensor azimuth angle, and relative azimuth angle. Bidirectional reflectance distribution characteristic correction refers to eliminating or reducing the influence of observation geometry on ground object reflectance measurement based on the bidirectional reflectance distribution characteristic model or empirical methods, normalizing reflectance data acquired under different observation geometries to a preset standard observation geometry. The standard observation geometry refers to the reference observation angle used to normalize reflectance data; for example, it can be set as the observation conditions when the sensor zenith angle is 0 degrees (nadir observation) and the solar zenith angle is a fixed angle.

[0080] Specifically, remote sensing images typically contain the observation geometry parameters acquired at the time of acquisition in their metadata. These parameters can be used to quantify the impact of different observation angles on reflectance. Step S25 performs bidirectional reflectance distribution characteristic correction based on the acquired observation geometry information and the atmospherically corrected rice canopy reflectance data. This correction process applies a bidirectional reflectance distribution characteristic model or algorithm to adjust the reflectance values ​​measured at different observation angles to the equivalent reflectance values ​​under a preset standard observation geometry. For example, all data can be corrected to reflectance under nadir observation conditions. In this way, reflectance variations introduced by differences in observation angles are eliminated, making rice canopy reflectance data acquired at different times, with different sensors, or from different observation angles more comparable. The rice canopy reflectance data under the standard observation geometry obtained after bidirectional reflectance distribution characteristic correction provides a more stable and reliable input for subsequent extraction of spectral feature parameters related to nutrient status.

[0081] Step S25, combined with preprocessing steps such as radiometric calibration, geometric correction, ground cover screening, and atmospheric correction, works together on the raw spectral data to progressively eliminate external interference factors and gradually extract information from the raw signal that better reflects the true reflectance characteristics of the rice canopy. Radiometric calibration and geometric correction provide basic spatial and radiometric accuracy; ground cover screening ensures that only rice-growing areas are analyzed; atmospheric correction eliminates the effects of atmospheric scattering and absorption; and two-way reflectance distribution characteristic correction further eliminates the influence of observation angle. This comprehensive correction process results in purer and more accurate reflectance data, thereby significantly improving the reliability of rice nutrient diagnosis and yield prediction based on this data.

[0082] In some preferred embodiments, the spectral characteristic parameters include multiple of the following: vegetation index, red edge parameter, narrow band spectral index, reflectance combination value of vegetation pigment-related bands, and absorption peak characteristic value.

[0083] Specifically, vegetation indices reflect vegetation biomass and chlorophyll content, and are commonly used parameters to reflect vegetation growth status and pigment content. By extracting vegetation indices, we can understand the overall growth vigor and chlorophyll level of plants, which are closely related to nutrient status. Vegetation indices include the Normalized Difference Vegetation Index (NDVI), calculated based on red and near-infrared reflectance. NDVI is based on red and near-infrared reflectance and is sensitive to vegetation cover and chlorophyll content. Calculating NDVI based on red and near-infrared reflectance utilizes the typical characteristics of these two bands in the vegetation spectrum, reflecting the health status and biomass of vegetation, and providing basic information for nutrient diagnosis.

[0084] More specifically, the red edge parameter reflects vegetation health and amino acid status. The red edge is a rapidly rising region in the vegetation spectrum between the red and near-infrared bands, and its position and slope are highly sensitive to the physiological state of the vegetation. Extracting the red edge parameter allows for more precise capture of changes in vegetation in this key band, providing supplementary information for assessing vegetation health and amino acid status, and helping to differentiate amino acid deficiency from other stresses. Changes in the red edge position are related to chlorophyll content and amino acid levels, while the red edge slope reflects the rate of change in chlorophyll content. Calculating the red edge position and slope based on reflectance data utilizes the sensitivity of the red edge region to vegetation physiological status, providing more nuanced physiological information than simple vegetation indices, and enhancing the ability to assess amino acid status and health.

[0085] More specifically, narrow-band spectral indices reflect specific nutrient states. These indices are calculated based on the reflectance of specific narrow bands, which may be related to the absorption or reflection characteristics of specific nutrients. Extracting narrow-band spectral indices allows for the targeted capture of spectral information related to specific nutrients such as phosphorus and potassium, compensating for the shortcomings of vegetation indices and red-edge parameters, which mainly focus on amino acids and biomass, thus providing a more comprehensive basis for nutrient diagnosis.

[0086] More specifically, the reflectance combination values ​​and absorption peak characteristic values ​​of vegetation pigment-related bands also provide spectral information related to the physiological and biochemical components of vegetation, further enriching the feature set for diagnosis.

[0087] More specifically, the extraction of these multiple spectral characteristic parameters, combined with the acquisition of rice canopy reflectance data and subsequent spatial analysis, nutrient deficiency assessment, and yield prediction steps, provides richer and more discriminative information. This helps improve the accuracy of nutrient deficiency diagnosis, overcomes the problem that single or general spectral parameters are easily interfered with by other stress factors, and thus enhances the robustness and accuracy of the entire diagnosis and prediction method.

[0088] In some preferred embodiments, step S3 includes:

[0089] S31. Based on the rice growth period information and / or variety information of the paddy field, obtain the range or combination of target bands related to the nutrient status of rice ammonia, phosphorus, and potassium.

[0090] S32. Based on the reflectance data of the rice canopy and the band range or band combination, obtain the spectral characteristic parameters of ammonia, phosphorus and potassium nutrient status. The spectral characteristic parameters include vegetation index, reflectance combination value of target band and absorption peak characteristic value.

[0091] Specifically, step S31 can be implemented by consulting a pre-established correspondence table, using a machine learning model for prediction, or dynamically determining the spectral parameters through spectral analysis. Its purpose is to enable the subsequently extracted spectral feature parameters to more effectively reflect nutrient information and reduce interference from non-nutrient factors.

[0092] More specifically, step S32 is used to calculate or extract spectral indicators that can characterize the nutrient status of rice, with the aim of converting the raw spectral reflectance data into quantitative indicators that are more closely related to the nutrient status.

[0093] Specifically, the method of this application, when extracting spectral characteristic parameters of nutrient status from rice canopy reflectance data, first obtains the growth stage information and / or variety information of the current rice field. This is because rice has different physiological and biochemical characteristics at different growth stages, and there are also differences between different varieties. These differences lead to different canopy spectral responses and sensitive bands to nutrient deficiency. Based on this specific growth stage and / or variety information, the method of this application determines the target band range or combination that is more closely related to the current ammonia, phosphorus, and potassium nutrient status of the rice. This means that fixed bands or band combinations are no longer used, but rather the spectral information that best reflects the nutrient status is selected according to the specific situation of the rice. Subsequently, using the corrected rice canopy reflectance data, and in combination with the determined target band range or combination, spectral characteristic parameters of ammonia, phosphorus, and potassium nutrient status are calculated and obtained. These parameters can be vegetation indices, reflectance combination values ​​of target bands, or absorption peak characteristic values. By extracting feature parameters within the selected target band range or using a selected combination of bands, spectral signals related to the nutrient status of rice (ammonia, phosphorus, and potassium) can be captured more accurately, reducing the influence of irrelevant spectral information or interference signals. This method of dynamically selecting bands and extracting features based on growth stage and / or variety information improves the correlation between the extracted feature parameters and nutrient status, providing a more reliable data foundation for subsequent nutrient deficiency diagnosis and yield prediction.

[0094] Through the above scheme, the method of this application can select spectral bands or combinations that are more correlated with nutrient status to extract characteristic parameters according to the specific growth stage and variety of rice. This allows the extracted spectral characteristic parameters to more accurately reflect the true nutrient status of rice, reduce the interference of other non-nutrient factors on the diagnostic results, and improve the reliability of nutrient deficiency diagnosis and yield prediction.

[0095] In some preferred embodiments, step S31 includes:

[0096] S311. Obtain information on the current growth stage and / or variety of rice paddies;

[0097] S312. Based on the current growth period information and / or variety information, find the range or combination of target bands related to the nutrient status of ammonia, phosphorus, and potassium from the preset correspondence data.

[0098] Specifically, growth period information refers to the growth and development stages of rice from sowing to maturity, such as the greening stage, tillering stage, jointing stage, heading stage, grain-filling stage, and maturity stage. Variety information refers to the specific variety type of rice. Different varieties differ in genetic characteristics, morphological features, and growth habits, and this information can be obtained through farmer records, field marking, and variety database searches.

[0099] More specifically, the pre-defined correspondence data refers to a set of data that is established in advance and stores the spectral band ranges or combinations associated with different rice growth stages and / or varieties and specific nutrient (ammonia, phosphorus, potassium) states. It can be stored in the form of tables, databases, lookup tables, etc.

[0100] Through the above processing, the method of this application can dynamically select the most suitable spectral bands or combinations to reflect nutrient status based on the specific growth stage and / or variety information of rice, thereby improving the correlation between the extracted spectral feature parameters and the actual nutrient status, reducing the interference of other non-nutrient stress factors, and making the subsequent nutrient deficiency diagnosis and yield prediction more accurate and reliable.

[0101] In some preferred embodiments, step S4 includes:

[0102] S41. Obtain the spatial location information corresponding to the spectral feature parameters;

[0103] S42. Based on the spectral characteristic parameter values ​​and spatial location information, calculate the global spatial autocorrelation index and local spatial autocorrelation index of the spectral characteristic parameters;

[0104] S43. Based on the global spatial autocorrelation index and the local spatial autocorrelation index, obtain the spatial distribution characteristics of the spectral feature parameters. The spatial distribution characteristics include the overall spatial clustering characteristics of the spectral feature parameters and the local salient regions.

[0105] Specifically, spatial location information refers to the specific coordinates or location identifiers of spectral feature parameters in geographic space, which can be realized using the row and column numbers of pixels, geographic coordinates (such as latitude and longitude), or planar coordinates based on a specific projection system.

[0106] More specifically, the global spatial autocorrelation index is a statistical indicator that measures the degree of spatial correlation of a certain variable (in this case, a spectral characteristic parameter) across the entire study area. It can be calculated using Moran's I index, Geary's C index, or the Getis-Ord General G index. The local spatial autocorrelation index is a statistical indicator that measures the degree of spatial correlation of a certain variable (in this case, a spectral characteristic parameter) between each spatial unit and its neighboring units. It is used to identify spatial clustering or outlier patterns in local areas and can be calculated using the Local Moran's I index, the Getis-Ord Gi index, or... The autocorrelation index is used for calculation. The global spatial autocorrelation index and the local spatial autocorrelation index are commonly used indices in spatial analysis, and their specific calculation methods will not be elaborated here.

[0107] More specifically, spatial distribution characteristic information refers to the spatial distribution patterns and laws of spectral characteristic parameters revealed by spatial analysis methods, which can be represented by the results of spatial autocorrelation analysis, spatial clustering analysis, or spatial interpolation analysis.

[0108] More specifically, overall spatial clustering characteristics refer to the overall spatial distribution trend of spectral characteristic parameters throughout the study area. That is, whether regions with similar values ​​tend to cluster together, are randomly distributed, or are dispersed. This can be determined based on the value and statistical significance of the global spatial autocorrelation index. Locally significant regions refer to local spatial units or areas within the study area where the spectral characteristic parameter values ​​exhibit statistically significant differences from the surrounding areas. Examples include regions with significant clustering of high values, significant clustering of low values, or significant outliers between high and low values. These can be identified based on the value and statistical significance of the local spatial autocorrelation index.

[0109] Specifically, step S41 obtains the spatial location information corresponding to the spectral characteristic parameters, providing basic data for subsequent spatial analysis. In step 42, the global spatial autocorrelation index is used to assess the average spatial correlation of the spectral characteristic parameters across the entire field, determining whether their overall distribution is clustered, dispersed, or random; the local spatial autocorrelation index, on the other hand, assesses the correlation between each spatial unit and its neighboring units, thereby identifying statistically significant local clusters or outliers. Step S43, through a method based on spatial autocorrelation analysis, concretizes the spatial analysis of the spectral characteristic parameters, expanding the focus from merely numerical magnitude to their spatial distribution patterns. This spatial distribution information, especially the identification of low-value clustered areas, provides specific and effective spatial evidence for subsequent steps to combine numerical and spatial information to determine nutrient deficiency.

[0110] The method of this application, by combining numerical information of spectral characteristic parameters with spatial distribution characteristic information, can more effectively identify abnormal areas caused by nutrient deficiency. This is because nutrient deficiency usually manifests as poor plant growth within a specific area, resulting in low values ​​of spectral characteristic parameters, and these low-value areas often exhibit spatial clustering. This spatial clustering pattern helps distinguish nutrient deficiency from other randomly distributed stress factors (such as sporadic pests or diseases or accidental mechanical damage), thereby overcoming the problem of spectral information confusion mentioned in the background art and improving the accuracy of nutrient deficiency diagnosis.

[0111] In some preferred embodiments, step S43 includes:

[0112] S431. Based on the value and statistical significance of the global spatial autocorrelation index, determine the type of overall spatial clustering characteristics of the spectral feature parameters;

[0113] S432. Identify the local salient regions of spectral characteristic parameters based on the values ​​and statistical significance of the local spatial autocorrelation index.

[0114] Specifically, statistical significance refers to the degree to which a spatial pattern is determined by statistical tests to determine whether it is generated by a random process. It can be determined by using statistical measures such as p-value and Z-value in combination with a preset significance level.

[0115] More specifically, in the embodiments of this application, the types of overall spatial clustering characteristics include spatial clustering, spatial dispersion, or spatial random distribution; local salient regions include high-value clustering regions, low-value clustering regions, or high-low value outlier regions.

[0116] Specifically, the method of this application, through step S431, can clearly determine whether the spectral characteristic parameters exhibit a spatial clustering, spatial dispersion, or spatial random distribution pattern throughout the paddy field based on the magnitude of the global spatial autocorrelation index and its statistical test results. For example, a positive and statistically significant global index indicates spatial clustering, a negative and statistically significant global index indicates spatial dispersion, and an insignificant global index indicates spatial random distribution. This judgment provides a foundation for understanding the overall spatial pattern of nutrient status. Secondly, the method of this application, through step S432, can identify local areas that exhibit patterns significantly different from the surrounding areas based on the values ​​of the local spatial autocorrelation index and its statistical test results. For example, a positive and statistically significant local index, coupled with a high parameter value at that location, is identified as a high-value clustering area; a positive and statistically significant local index, coupled with a low parameter value at that location, is identified as a low-value clustering area; and a negative and statistically significant local index is identified as a high-low value outlier area. These locally significant areas correspond to places where the spectral characteristic parameters exhibit abnormal spatial clustering or outliers. By identifying these regions and distinguishing their types, potential areas of nutrient excess, nutrient deficiency, or abnormal boundaries can be precisely located. Determining overall spatial aggregation characteristics provides macroscopic spatial pattern information, while identifying locally significant regions provides microscopic key location information. The combination of these two types of information constitutes complete spatial distribution characteristics of spectral features, providing accurate and geographically significant spatial evidence for subsequent nutrient deficiency diagnosis.

[0117] In some preferred embodiments, step S5 includes:

[0118] S51. Based on a preset feature threshold, identify abnormal areas in paddy fields where the values ​​of spectral feature parameters are lower than the preset threshold.

[0119] S52. Based on the abnormal area and the local significant area, determine whether the abnormal area conforms to the spatial distribution pattern of nutrient deficiency;

[0120] S53. When the spatial distribution pattern of nutrient deficiency conforms to the spatial distribution pattern, determine whether there is nutrient deficiency in the paddy field based on the type of overall spatial aggregation characteristics.

[0121] S54. When it is determined that there is a nutrient deficiency, the area of ​​nutrient deficiency shall be determined based on the abnormal area;

[0122] S55. Analyze the degree of numerical deviation of different types of spectral characteristic parameters in the missing region;

[0123] S56. Determine the type and severity of the deficiency based on the type of spectral characteristic parameters whose values ​​are lower than the preset threshold and the corresponding degree of numerical deviation, and generate a nutrient deficiency diagnosis result based on the deficiency region.

[0124] Specifically, the preset feature threshold refers to the critical value used to distinguish between normal and abnormal spectral feature parameter values, which can be determined based on historical data statistical analysis, expert experience and knowledge, rice variety information, or based on a specific diagnostic model.

[0125] More specifically, abnormal areas refer to the spatial range of spectral characteristic parameters in paddy fields where the values ​​are below a preset threshold. These areas indicate potential nutrient deficiencies or other stress effects. The spatial distribution pattern of nutrient deficiencies refers to the specific spatial morphology or aggregation characteristics that nutrient deficiencies typically exhibit in paddy fields. For example, it may manifest as continuous patchy areas or a distribution related to specific environmental factors. This can be determined by analyzing and summarizing spatial data from known nutrient-deficient fields or based on agronomic principles. The deficient area refers to the spatial range identified as having actual nutrient deficiencies after spatial distribution pattern analysis; it is the portion screened and confirmed from the initially identified abnormal areas. The degree of numerical deviation refers to the degree of deviation of the actual values ​​of different types of spectral characteristic parameters within the identified deficient area from the normal level or preset threshold. This can be calculated using methods such as percentage deviation or standardized difference. The deficiency type refers to the specific type of nutrient deficiency determined based on the types of spectral characteristic parameters whose values ​​are below the preset threshold. The severity of the deficiency refers to the severity level of the nutrient deficiency determined based on the degree of numerical deviation of the spectral characteristic parameters. The nutrient deficiency diagnosis result refers to the final diagnostic information that integrates the deficient area, deficiency type, and deficiency severity.

[0126] Specifically, the method of this application first identifies areas with low spectral characteristic parameter values ​​in paddy fields based on preset feature thresholds and the magnitude of these values. These areas may be potential nutrient deficiency areas or areas affected by other stresses, and are defined as anomalous areas. Next, the method utilizes locally significant region information of spectral characteristic parameters obtained from spatial analysis to compare the initially identified anomalous areas with these statistically significant local clusters or outliers. By determining whether the anomalous areas conform to the spatial distribution patterns typically exhibited by nutrient deficiency, such as whether they are located within low-value clusters, it is possible to effectively distinguish between anomalies caused by nutrient deficiency and those caused by other random or different spatial stress patterns, thereby reducing misdiagnosis. After confirming that the anomalous areas conform to the spatial distribution patterns of nutrient deficiency, the method further combines the overall spatial clustering characteristics of the spectral characteristic parameters obtained from spatial analysis, such as whether the overall structure exhibits low-value clustering, to determine whether nutrient deficiency exists in the entire paddy field based on the type of overall spatial clustering characteristics. This step further verifies the possibility of nutrient deficiency at a macroscopic level. When nutrient deficiency is determined, the method of this application utilizes previously identified abnormal regions to precisely determine the specific spatial range of the nutrient deficiency, i.e., the deficiency region. Subsequently, different types of spectral characteristic parameters within the determined deficiency region are analyzed, and their deviations from normal levels or preset thresholds are calculated. Finally, based on the types of spectral characteristic parameters with values ​​below the preset threshold and their corresponding deviations, combined with the determined deficiency region, a nutrient deficiency diagnosis result containing the deficiency type, deficiency region, and deficiency severity is generated. Through this diagnostic process that combines numerical thresholds, local spatial patterns, and overall spatial characteristics, the method of this application can more accurately identify and diagnose amino, phosphorus, and potassium nutrient deficiencies in rice, providing a more reliable basis for subsequent yield prediction.

[0127] In some preferred embodiments, based on the foregoing, the nutrient deficiency diagnosis results include the deficiency type, the deficiency area, and the deficiency severity. Step S6 includes:

[0128] S61. Obtain the expected yield of a paddy field under conditions of no nutrient deficiency;

[0129] S62. Based on preset rules or prediction models, predict the missing output for the corresponding missing area according to the missing type and the severity of the missing output, and obtain the output loss value.

[0130] S63. Calculate the predicted yield of the paddy field based on the yield loss value and the expected yield.

[0131] Specifically, expected yield refers to the potential yield of a paddy field under conditions of sufficient nutrient supply and no nutrient stress. It can be obtained using historical yield data, simulation results based on crop growth models, or expert experience. Preset rules or prediction models refer to algorithms or models used to establish a correlation between the type and severity of nutrient deficiency and yield loss. These can be implemented using empirical formulas, lookup tables, regression models, machine learning models, etc.; preset rules or prediction models are trained based on historical spectral characteristic parameters and yield data. Deficit yield prediction refers to estimating the yield reduction caused by nutrient deficiency based on nutrient deficiency diagnosis results. Yield loss value refers to the specific yield reduction obtained through deficient yield prediction. Predicted yield refers to the estimated yield derived by comprehensively considering both the potential yield of the field and the loss caused by nutrient deficiency.

[0132] Specifically, the method of this application obtains the expected yield of a paddy field under nutrient-free conditions in step S61. This provides a benchmark for yield estimation, namely the potential yield of the field under ideal nutrient supply, enabling subsequent yield estimation to reflect the actual yield reduction caused by nutrient deficiency. Next, step S62, based on preset rules or a prediction model, predicts the yield loss for these areas according to the type, severity, and corresponding deficiency region in the nutrient deficiency diagnosis results, thereby obtaining specific yield loss values. This step transforms the qualitative or semi-quantitative nutrient deficiency diagnosis results into quantitative yield losses. The rules or model consider the different impacts of different nutrient deficiency types and severity on yield, and the calculation is limited to the areas where deficiencies actually occur, making the yield loss calculation more accurate and targeted. Finally, step S63, based on the expected yield obtained in step S61 and the yield loss value calculated in step S62, obtains the predicted yield of the paddy field through simple subtraction. This method combines potential yield with the loss caused by nutrient deficiency, resulting in a predicted yield that is closer to the actual situation and improving the reliability of yield estimation.

[0133] Secondly, please refer to Figure 2 Some embodiments of this application also provide a spectrum-based system for predicting the deficiency of ammonia, phosphorus, and potassium in rice and its yield, used for yield prediction in rice fields. The system includes:

[0134] The acquisition module 201 is used to acquire spectral data of paddy fields;

[0135] The correction module 202 is used to correct the spectral data in order to obtain the reflectance data of the rice canopy.

[0136] The feature extraction module 203 is used to extract spectral feature parameters about the nutrient status of ammonia, phosphorus, and potassium based on the reflectance data of the rice canopy;

[0137] Spatial analysis module 204 is used to perform spatial analysis on spectral feature parameters to obtain spatial distribution feature information of spectral feature parameters with respect to paddy fields;

[0138] The diagnostic module 205 is used to combine the numerical value and spatial distribution characteristics of spectral feature parameters to determine whether there is nutrient deficiency in paddy fields, and to analyze and obtain the nutrient deficiency diagnosis results when there is nutrient deficiency.

[0139] The estimation module 206 is used to estimate the yield of rice fields based on the nutrient deficiency diagnosis results.

[0140] This application's spectral-based rice ammonia, phosphorus, and potassium deficiency and yield prediction system combines numerical information of spectral characteristic parameters with spatial distribution characteristic information obtained through spatial analysis to determine whether nutrient deficiency exists in rice fields. This effectively distinguishes between abnormalities caused by nutrient deficiency and spectral abnormalities caused by other stress factors, solving the problem of inaccurate diagnosis caused by easily confused spectral information in existing technologies. It improves the accuracy of nutrient deficiency diagnosis, thereby enhancing the precision of rice yield prediction and providing a scientific basis for precise fertilization and management of rice.

[0141] Furthermore, the units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0142] Furthermore, the functional modules in the various embodiments of this application can be integrated together to form an independent part, or each module can exist independently, or two or more modules can be integrated to form an independent part.

[0143] In this document, relational terms such as first and second are used only to distinguish one entity or operation from another entity or operation, without necessarily requiring or implying any such actual relationship or order between these entities or operations.

[0144] The above description is merely an embodiment of this application and is not intended to limit the scope of protection of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of protection of this application.

Claims

1. A spectral-based method for predicting the deficiency of amino, phosphorus, and potassium in rice and its yield, used for yield prediction in rice fields, characterized in that... The steps of this method include: S1. Obtain spectral data of the paddy field; S2. Correct the spectral data to obtain the reflectance data of the rice canopy; S3. Extract spectral characteristic parameters of ammonia, phosphorus, and potassium nutrient status based on the reflectance data of the rice canopy; S4. Perform spatial analysis on the spectral feature parameters to obtain spatial distribution information of the spectral feature parameters with respect to paddy fields; S5. Combining the numerical values ​​of the spectral characteristic parameters and the spatial distribution characteristic information, determine whether the paddy field has nutrient deficiency, and analyze and obtain the nutrient deficiency diagnosis results when nutrient deficiency exists; S6. Estimate the yield of the paddy field based on the nutrient deficiency diagnosis results; Step S3 includes: S31. Based on the rice growth period information and / or variety information of the paddy field, obtain the range or combination of target bands related to the ammonia, phosphorus, and potassium nutrient status of rice; S32. Based on the reflectance data of the rice canopy and the band range or band combination, obtain spectral characteristic parameters regarding the nutrient status of ammonia, phosphorus, and potassium. The spectral characteristic parameters include vegetation index, reflectance combination value of target band, and absorption peak characteristic value. Step S5 includes: S51. Based on a preset feature threshold, identify abnormal areas in paddy fields where the values ​​of spectral feature parameters are lower than the preset threshold. S52. Based on the abnormal area and the local significant area, determine whether the abnormal area conforms to the spatial distribution pattern of nutrient deficiency; S53. When the spatial distribution pattern of nutrient deficiency conforms to the spatial distribution pattern, determine whether there is nutrient deficiency in the paddy field based on the type of overall spatial aggregation characteristics. S54. When it is determined that there is a nutrient deficiency, the area of ​​nutrient deficiency shall be determined based on the abnormal area; S55. Analyze the degree of numerical deviation of different types of spectral characteristic parameters in the missing region; S56. Determine the type and severity of the deficiency based on the type of spectral characteristic parameters whose values ​​are lower than the preset threshold and the corresponding degree of numerical deviation, and generate a nutrient deficiency diagnosis result based on the deficiency region.

2. The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectral analysis according to claim 1, characterized in that, The correction process includes radiometric calibration and geometric correction; step S2 includes: S21. Perform radiometric calibration and geometric correction on the spectral data to obtain preliminary reflectance data and corresponding location information; S22. Based on the reflectance characteristics of pixels of different land cover types, the preliminary reflectance data is filtered to obtain reflectance data for rice canopy pixels.

3. The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectroscopy according to claim 2, characterized in that, The correction process also includes atmospheric correction, and step S2 further includes: S23. Perform atmospheric correction on the reflectance data of the rice canopy pixels.

4. The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectroscopy according to claim 3, characterized in that, Step S2 also includes: S24. Obtain the observation geometric information corresponding to the spectral data; S25. Based on the observation geometry information and the atmospherically corrected reflectance data, perform bidirectional reflectance distribution characteristic correction to obtain the reflectance data of the rice canopy under the standard observation geometry.

5. The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectroscopy according to claim 1, characterized in that, Step S31 includes: S311. Obtain the current growth period information and / or variety information of the rice field; S312. Based on the current growth period information and / or variety information, search for the range or combination of target bands related to the ammonia, phosphorus, and potassium nutrient status from the preset correspondence data.

6. The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectroscopy according to claim 1, characterized in that, Step S4 includes: S41. Obtain the spatial location information corresponding to the spectral feature parameters; S42. Based on the spectral feature parameter values ​​and the spatial location information, calculate the global spatial autocorrelation index and the local spatial autocorrelation index of the spectral feature parameters; S43. Based on the global spatial autocorrelation index and the local spatial autocorrelation index, obtain the spatial distribution characteristic information of the spectral feature parameters, wherein the spatial distribution characteristic information includes the overall spatial clustering characteristics of the spectral feature parameters and local salient regions.

7. The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectroscopy according to claim 6, characterized in that, Step S43 includes: S431. Based on the value and statistical significance of the global spatial autocorrelation index, determine the type of overall spatial clustering characteristics of the spectral feature parameters; S432. Based on the value and statistical significance of the local spatial autocorrelation index, identify the local salient regions of the spectral feature parameters.

8. The method for predicting rice ammonia, phosphorus, and potassium deficiency and yield based on spectroscopy according to claim 1, characterized in that, The nutrient deficiency diagnosis results include the deficiency type, deficiency area, and deficiency severity. Step S6 includes: S61. Obtain the expected yield of the rice paddy under conditions of no nutrient deficiency; S62. Based on preset rules or prediction models, according to the missing type and the severity of the missing, predict the missing output for the corresponding missing area and obtain the output loss value. S63. Calculate the predicted yield of the paddy field based on the yield loss value and the expected yield.

9. A spectral-based system for predicting the deficiency and yield of amino acids, phosphorus, and potassium in rice, used for predicting the yield of rice fields, and for executing the spectral-based method for predicting the deficiency and yield of amino acids, phosphorus, and potassium in rice as described in any one of claims 1-8, characterized in that... The system includes: The acquisition module is used to acquire spectral data of rice paddies; The correction module is used to correct the spectral data to obtain the reflectance data of the rice canopy. The feature extraction module is used to extract spectral feature parameters related to the nutrient status of ammonia, phosphorus, and potassium based on the reflectance data of the rice canopy. The spatial analysis module is used to perform spatial analysis on the spectral feature parameters to obtain spatial distribution feature information of the spectral feature parameters with respect to paddy fields; The diagnostic module is used to determine whether the paddy field has nutrient deficiency by combining the numerical value of the spectral feature parameters and the spatial distribution feature information, and to analyze and obtain the nutrient deficiency diagnosis results when nutrient deficiency exists. The estimation module is used to estimate the yield of the paddy field based on the nutrient deficiency diagnosis results.

Images