Unmanned aerial vehicle hyperspectral online calibration measurement method for agricultural remote sensing

Abstract

The invention relates to the technical field of agricultural remote sensing measurement, and discloses an unmanned aerial vehicle hyperspectral online calibration measurement method for agricultural remote sensing, which comprises the following steps: acquiring hyperspectral data of a target crop and reference vegetation in the same scene in real time through an unmanned aerial vehicle, and extracting a characteristic wave band reflectivity difference value delta R, according to the comparison result of the time sequence change rate of the delta R and the preset threshold value, crop physiological changes and environmental interference are dynamically distinguished, and online calibration is achieved. A self-consistent calibration mechanism is constructed by means of the time sequence stability of crop spectral characteristics, the illumination mutation influence can be eliminated without depending on an external reference substance, and the accuracy of the calibration result is improved. And meanwhile, a lodging scene is effectively identified through synchronous analysis of a gravity vector and spectrum drift, so that the system can still keep accurate tracking of a real physiological state of a crop under an extreme working condition.

Description

Technical Field

[0001] This invention relates to an online hyperspectral calibration measurement method for unmanned aerial vehicles (UAVs) used in agricultural remote sensing, belonging to the field of agricultural remote sensing measurement technology. Background Technology

[0002] In the field of agricultural remote sensing measurement, UAV hyperspectral technology monitors the physiological state of crops by capturing their reflectance spectra. Existing calibration methods mainly rely on two types of approaches: physical reference calibration (such as ground whiteboard calibration) and environmental model compensation (such as radiative transfer models). The former requires flight interruption to obtain reference data, while the latter requires real-time input of atmospheric parameters, such as aerosol optical thickness and solar zenith angle. Although these two methods are widely used, they have fundamental limitations, simplifying dynamic and continuous spectral monitoring into static spatial calibration and ignoring the temporal evolution characteristics of light conditions and crop physiological state in agricultural scenarios.

[0003] When the above methods are applied to actual farmland operations, their systemic defects are exposed in typical scenarios: whiteboard calibration requires the drone to hover or land, which leads to flight path interruptions. In farmland areas where clouds move rapidly, lighting conditions change 3-5 times per hour, and frequent interruptions make continuous monitoring impossible, resulting in loss of data timeliness. Radiation models rely on real-time data from weather stations, but the coverage of weather stations in farmland areas is less than 15%. When model parameters are missing or lagging, atmospheric scattering compensation errors are transmitted to reflectivity calculations, ultimately distorting the stress diagnosis results. To compensate for calibration errors, industry solutions combine multispectral sensors and edge AI chips. While this attempt to improve accuracy, it goes against the core trend of making agricultural drones more accessible and lightweight, making it difficult to deploy the technology on a large scale.

[0004] Industry attempts have focused on improvements through fusion of LiDAR point clouds or the addition of sensor frequency bands, but these solutions still face a new contradiction: increased external dependence and escalating computational complexity. LiDAR modeling requires high-precision RTK positioning, significantly increasing hardware costs; multi-band fusion algorithms rely on GPU computation, which edge devices struggle to handle; existing technologies treat calibration as a physical process independent of the monitored object, neglecting the fact that the persistent optical stability of self-generated vegetation in farmland scenarios can serve as a zero-cost dynamic benchmark. This blind spot has led to the industry's long-term reliance on high-cost external equipment, preventing the construction of a low-cost calibration mechanism intrinsically linked to agricultural ecological laws. Therefore, the technical problem this invention aims to solve is how to construct a lightweight hyperspectral online calibration method that does not rely on external references, can adapt to continuous flight conditions, and avoids environmental parameter coupling. Summary of the Invention

[0005] This invention provides a method for online hyperspectral calibration measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing. Its main purpose is to solve the problems of lack of dynamic continuity, strong environmental coupling, and hardware redundancy caused by the reliance on external equipment and complex models in existing spectral calibration technologies.

[0006] To achieve the above objectives, the present invention provides a method for online hyperspectral calibration measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing, comprising the following steps: Step a: Obtain continuous hyperspectral data of the crop area. The continuous hyperspectral data contains spectral reflectance information of multiple bands. Step b involves extracting multiple spectral characteristic wavelengths from continuous hyperspectral data in real time, based on crop type and growth stage. These spectral characteristic wavelengths are directly related to the physiological state of the crop, and a spectral characteristic centroid representing the current hyperspectral morphology is constructed based on these multiple spectral characteristic wavelengths. Step c: Calculate the drift of the spectral feature centroid per unit time in real time; Step d: Compare the drift of the spectral feature centroid with the physiological change threshold determined based on historical data statistical analysis and modeling, and perform attribution determination of the spectral centroid drift based on the comparison results; Step e: When the drift of the spectral feature centroid does not exceed the physiological change threshold, it is determined that the drift of the spectral feature centroid is caused by the change in the physiological state of the crop, and the current spectral feature centroid is updated to the new calibration benchmark. At the same time, the drift is recorded as an indicator parameter of the change in the health status of the crop. Step f: When the drift of the spectral feature centroid exceeds the physiological change threshold, it is determined that the drift of the spectral feature centroid is caused by external illumination or changes in the attitude of the UAV. At this time, the calibration benchmark is not updated, and a correction factor is generated to perform real-time calibration on the subsequently acquired continuous hyperspectral data.

[0007] Preferably, the multiple spectral feature wavelengths used to construct the spectral feature centroid in step b include: the red edge inflection point wavelength, the green light reflection peak wavelength, and the near-infrared plateau initiation point wavelength.

[0008] Preferably, in step d, the physiological change threshold is determined based on statistical analysis and modeling of historical spectral data of the target crop type at different growth stages.

[0009] Preferably, the method enables online calibration of UAV hyperspectral data without relying on any ground physical reference whiteboard or real-time atmospheric parameter input.

[0010] Preferably, in step e, when it is determined that the centroid drift of the spectral features is caused by changes in the physiological state of the crop, the method further includes: using the amplitude and direction of the centroid drift as an indicator of the degree and type of physiological stress on the crop.

[0011] Preferably, in step f, when the drift of the spectral feature centroid exceeds the physiological change threshold, the method further includes the following sub-steps: step f1, acquiring the current flight attitude data from the UAV's onboard inertial measurement unit; step f2, performing crop lodging observation scenario determination based on the flight attitude data; step f3, calling the pre-established lodging attitude-spectral morphology response model, and correcting the current spectral feature centroid according to the model to adapt to the specific spectral response characteristics of lodged crops.

[0012] Preferably, in step f2, the determination of the crop lodging observation scenario is performed by comparing the instantaneous rate of change of the flight attitude angle with a predetermined maneuvering flight angle threshold. If the rate of change exceeds the maneuvering flight angle threshold and is synchronized with the drift of the spectral feature centroid in time, it is determined that the crop lodging observation scenario is in progress.

[0013] Preferably, the lodging posture-spectral morphology response model is a calibrated and stored lookup table or low-order polynomial function, which is established by data acquisition and model training of simulated lodged crops at different tilt angles.

[0014] Preferably, the method further includes temporal micro-oscillation energy analysis of the spectral feature centroid for early warning of early physiological stress in crops. The analysis includes the following steps: Step g1, buffering the position data of the spectral feature centroid over a continuous time period at a high sampling frequency to form a centroid time-series signal; Step g2, processing the centroid time-series signal to calculate the total energy index of the micro-oscillation, which is obtained by calculating the variance of the centroid time-series signal; Step g3, comparing the total energy index of the micro-oscillation with the physiological resting baseline in real time. The physiological resting baseline represents the normal energy level of crop micro-oscillations. When the index is consistently higher than the physiological resting baseline, early stress warning information is output.

[0015] Preferably, in step g2, the variance of the centroid timing signal... The calculation method is as follows: in, Indicates the first The centroids of spectral features collected at each time point, This represents the total number of data points in the buffered centroid timing signal. This represents the average value of the centroid timing signal of the cache.

[0016] Compared with the prior art, the beneficial effects of the present invention are: 1. By constructing a time-series drift analysis mechanism for the centroid of the crop's own spectral characteristics, the decoupling of environmental disturbances and physiological changes is transformed into the identification of essential differences in drift rates. When the centroid change rate is lower than the physiological change threshold, the system absorbs it as a new calibration benchmark; when the rate abnormally jumps, instantaneous correction is triggered. This mechanism enables the spectral calibration process to break free from dependence on external references or atmospheric parameters, forming a stable measurement closed loop in the dynamic farmland environment.

[0017] 2. When a drone undergoes a drastic attitude change while observing lodged crops, the system actively identifies abrupt physical morphological changes by leveraging the strong spatiotemporal synchronization between flight attitude and spectral drift. It then invokes a pre-set lodging attitude-spectral response model to convert the gravity vector into a physical scale that corrects the spectral centroid, allowing the physiological state of the lodged area to be continuously tracked. This transforms the traditional measurement blind spot into a quantifiable post-disaster diagnostic window. By reusing the high-frequency calculated spectral centroid time-series data stream and analyzing the degree to which its micro-oscillation energy deviates from the healthy baseline, the system identifies signs of physiological regulatory dysfunction, such as water stress, even before the macroscopic morphology of the crop changes. This interpretation of the spectral heartbeat rhythm enables the system to provide early warning signals for the golden window of intervention before the conventional diagnostic process is initiated.

[0018] 3. The basic calibration mechanism, lodging adaptation mechanism, and early warning mechanism are not functionally superimposed, but rather form a three-layer cognitive leap: the first layer filters out environmental noise and locks into the essence of physiological changes; the second layer penetrates drastic changes in physical morphology and maintains diagnostic continuity; the third layer perceives latent instability precursors and expands the monitoring time boundary. The three layers form a low-resource-consumption synergistic enhancement network by sharing the centroid calculation data stream. The output data evolves from traditional reflectance maps to a multi-dimensional diagnostic set that integrates centroid drift trajectory, lodging correction parameters, and micro-oscillation energy indicators. This not only avoids the interference of redundant environmental information on decision-making, but also transforms complex spectral data into biological language that characterizes the spatiotemporal evolution logic of crops, lowering the interpretation threshold for agronomic experts. Attached Figure Description

[0019] Figure 1 This is a timing diagram of the online hyperspectral calibration and measurement method for crops according to the present invention; Figure 2 This is a graph showing the changes in spectral reflectance of crops during different growth stages according to the present invention. Figure 3 This is a flowchart of the crop hyperspectral data processing and analysis process of the present invention.

[0020] The objectives, features, and advantages of this invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0021] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below in conjunction with specific embodiments.

[0022] The UAV hyperspectral online calibration measurement method for agricultural remote sensing disclosed in this application aims to construct a closed-loop adaptive spectral data processing framework. Its fundamental purpose is to use the inherent temporal stability of the crop's own spectral characteristics as a dynamic benchmark to accurately distinguish and calibrate data fluctuations caused by external environmental factors (such as sudden changes in light intensity) in real time. At the same time, it deeply mines spectral temporal information to achieve accurate tracking of crop physiological state and early stress warning.

[0023] In its specific implementation, this method framework mainly follows core stages such as real-time acquisition and feature extraction of hyperspectral data, attribution analysis of temporal drift of spectral feature centroids, and adaptive calibration and physiological state diagnosis based on attribution results. Its complete operation can be illustrated in a typical farmland inspection scenario. Imagine a drone equipped with a hyperspectral imager conducting routine monitoring of a rice paddy in its jointing stage. The specific execution steps of this invention are as follows: First, the drone flies along a preset route above the crop canopy. Its onboard hyperspectral imager continuously acquires image data of the rice canopy below at a predetermined frequency, generating real-time images covering the visible to near-infrared spectrum. A continuous hyperspectral data cube, encompassing hundreds of narrow-band spectral reflectance information, forms the raw input for all subsequent analyses. Next, an airborne or ground-based processing system analyzes this data stream in real time, extracting the spectral characteristic wavelengths most closely related to the physiological state of rice at its current growth stage. Based on prior agronomic knowledge, for rice in the jointing stage, these characteristic wavelengths are precisely identified as the red-edge inflection point wavelength, the green light reflectance peak wavelength, and the near-infrared plateau initiation point wavelength. These three wavelengths respectively characterize the crop's physiological status from three dimensions: chlorophyll content, cell structure health, and canopy water content. The system is not isolated. Instead of using these three discrete wavelength values, the system weights and fuses them along the spectral dimension to construct a multi-dimensional spectral feature centroid. Conceptually, this centroid can be viewed as a focal point in the spectral space representing the current overall physiological state of the crop. Its dynamic positional changes can macroscopically and sensitively characterize the evolution of the crop's overall spectral morphology. Subsequently, the system enters a continuous monitoring cycle, continuously locating the spectral feature centroid within extremely short timescales (e.g., seconds or sub-seconds) and calculating its drift per unit time. This drift, as a vector, precisely quantifies the instantaneous rate and specific trend of spectral morphology change. At this point, the system initiates its core judgment... The logic compares the real-time calculated centroid shift of the spectral features with a preset physiological change threshold to determine the root cause of the shift. This physiological change threshold is established based on the aforementioned principles through in-depth statistical analysis and modeling of a large amount of historical spectral data of the target crop at a specific growth stage. For example, by learning the range of natural rhythmic fluctuations in the spectrum of rice caused by normal photosynthesis, water transpiration, and other physiological activities during the jointing stage, the model provides a reasonable upper limit for the drift rate. This threshold scientifically defines the maximum rate of change in spectral morphology that can be caused by the physiological evolution of the crop itself under stable external conditions.

[0024] Based on the comparison results, the system will automatically split into two distinct processing paths: First, if the drift of the spectral feature centroid does not exceed a preset physiological change threshold, the system determines that the drift originates from the slow and real physiological evolution of the crop itself, such as a gradual decrease in leaf water content due to continuous diurnal transpiration. In this case, the system will update the current position of the spectral feature centroid, which has experienced a slight drift, as the new calibration benchmark. This means that the zero point of the measurement system can proactively and slowly adapt to the actual growth rhythm of the crop, thereby ensuring the long-term accuracy of the measurement. Simultaneously, as a further application defined by this invention, the system will also record and analyze this... The amplitude and direction of the centroid drift are important indicators. A persistent, small drift trajectory towards a specific spectral region can be precisely interpreted as an early signal of a specific physiological stress (such as mild nitrogen deficiency) and output as an indicator parameter of changes in crop health. Secondly, if the drift of the spectral characteristic centroid significantly exceeds the physiological change threshold, for example, if the light intensity drops sharply within seconds due to the rapid movement of clouds in the sky, causing a drastic and synchronous non-physiological jump in reflectivity across all bands, the system determines that this drastic drift is caused by non-physiological factors such as external light or sudden changes in drone attitude. In this case, the system will lock the calibration benchmark and not update it, and immediately adjust the output based on the amplitude and direction of the drift. A dynamic correction factor is calculated in reverse, which is applied to all subsequently acquired hyperspectral data until external interference disappears. This enables online real-time calibration of ambient light abrupt changes without relying on any ground-based physical reference whiteboard or real-time atmospheric parameter input, ensuring the continuity and inherent consistency of the monitoring data stream. Furthermore, in specific scenarios, the system will activate a more refined sub-process to handle special observation conditions such as crop lodging: when the system detects a severe spectral drift exceeding a threshold, it acquires the current flight attitude data from the UAV's onboard inertial measurement unit (IMU); subsequently, the system analyzes the instantaneous changes in the flight attitude angle. If the rate of change exceeds a preset flight angle threshold representing violent maneuvering, and the dramatic change in attitude is strictly synchronized with the dramatic shift of the spectral feature centroid in terms of timestamp, the system will determine with high confidence that it is currently observing a crop lodging scenario. Once the determination is made, the system will call a pre-established lodging attitude-spectral morphology response model, which is a lookup table or low-order polynomial function calibrated and stored after a large amount of data collection and model training on the spectral response of simulated lodged crops at different tilt angles. It accurately maps the inherent laws of spectral morphology changes (such as increased shadow ratio and changes in multiple scattering paths) caused by physical tilt of the crop canopy.Based on the current UAV attitude angle, the system instantly queries or calculates the corresponding spectral correction value from the model and uses it to accurately correct the current spectral feature centroid. This effectively eliminates structural artifacts introduced by observing the lodging attitude, enabling the system to penetrate appearances and accurately track the true physiological health status of crops even under extreme conditions of drastic physical changes.

[0025] Finally, the method of this invention also integrates a high-level analysis module for early warning of physiological stress. This module achieves prospective diagnosis by analyzing the temporal micro-oscillation energy of the spectral feature centroid: First, the system caches high-density, fine-grained positional data of the spectral feature centroid over continuous time periods at a sampling frequency far higher than conventional monitoring, forming a centroid temporal signal that reflects its subtle dynamics. Subsequently, the system processes this temporal signal to calculate the total energy index of its micro-oscillations. Specifically, this index is calculated by measuring the variance of the centroid temporal signal. To quantify it, its calculation method strictly follows the formula. In progress, among which Indicates the first The centroids of spectral features collected at each time point, This represents the total number of data points for buffered timing signals, while This is the average value of the buffered timing signal; ultimately, the system will calculate the total energy index of the micro-oscillations in real time. Continuous comparisons are made against a pre-defined physiological resting baseline, representing the normal energy level of healthy crops in a resting state, characterized by small, regularly fluctuating spectral centroids. Once the indicators are monitored... If the stress level remains consistently and significantly above the baseline, even if no stress symptoms are yet apparent in the macroscopic morphology, the system will immediately output early stress warning information, as it indicates that the physiological regulatory functions within the crop may have become unstable or disordered.

[0026] Before entering the online operation phase, all the core parameters upon which this method depends are deterministically generated through an offline calibration procedure. This procedure first targets the spectral feature centroids. Calculation formula Weighting coefficients in Calibration is performed by applying specific stresses to the target crop under controlled conditions and simultaneously collecting spectral and physiological indicators, identifying the characteristic wavelengths that provide the most sensitive response. Initial weights are assigned, and the remaining weights are equally distributed. Then, using maximizing the correlation coefficient as the objective function, the weights are iteratively adjusted until convergence, thus obtaining the weight combination most correlated with a specific stress. Secondly, physiological change thresholds are determined. The determination of the centroid is achieved by applying a slow physiological stress to healthy crop samples, collecting continuous time series data on their centroid positions, calculating the statistical distribution of the centroid drift rate, and then taking the distribution... The upper limit serves as a threshold; furthermore, the threshold for the maneuvering flight angle in crop lodging scenarios is obtained by statistically analyzing historical data on the rate of change of attitude angles of UAVs during standard flight routes, taking a specific multiple of its 95th percentile value, while the lodging attitude-spectral morphology response model... This involves using a mechanical platform to simulate a series of tilt angles for crops from vertical to horizontal. At each angle, spectral data is collected to calculate the centroid position, and the angle-centroid position data pairs are finally fitted to a low-order polynomial function; finally, a physiological resting baseline is used for early warning. The variance of the centroid time-series signal is directly calculated from the variance of the signal obtained by continuous monitoring of healthy, stress-free crops under stable conditions for 24 hours; during online operation, the centroid drift of the spectral features is... Exceeding the threshold Furthermore, when the drone's attitude is stable, the system initiates a sudden change in illumination correction process, which generates a correction factor. It is based on the spectral data of two frames before and after the illumination change, and the stable spectral curve before the change is calculated. Average reflectivity across all bands Spectral curve after mutation Average reflectivity across all bands and take their ratio. Thus, the factor is obtained. As a multiplicative scalar, it is applied to all subsequently acquired spectral data until the drift regression threshold is reached; if the spectral drift exceeds the threshold and the rate of change of attitude angle simultaneously exceeds the maneuvering flight angle threshold, it is determined to be lodged crop, and the system immediately calls the offline calibrated response model. The current attitude angle provided by the airborne inertial measurement unit As input, the pseudo-error value of spectral centroid shift caused by physical tilt is calculated. and from the current real-time measured centroid position Subtracting this spurious difference value yields the corrected centroid, free from physical morphological interference. The corrected centroid was then used for subsequent assessments of the crop's true physiological state.

[0027] In another implementation, the weight coefficients of the spectral feature centroids are not preset fixed values, but are deterministically generated through an offline calibration procedure. This procedure first applies a pre-trained semantic segmentation model to the raw data cube acquired by the airborne hyperspectral camera. This model is a deep learning network whose function is to classify each pixel into a predefined category, thereby accurately dividing the image into the target crop canopy region and background regions such as soil and shadows. Subsequently, using only the pure pixel spectral data extracted from the target crop canopy region, combined with true physiological indicators of the crop ground collected synchronously in a controlled experiment, such as leaf water potential measured by the leaf pressure chamber, an optimization problem is constructed. The objective function is to maximize the value obtained by the formula... Calculated spectral characteristic centroid Pearson correlation coefficient between the time series and the corresponding physiological indicator time series This coefficient is a statistic ranging from -1 to 1, used to measure the degree of linear correlation between two variables, and is solved using numerical optimization algorithms under constraints. Furthermore, with all weight components being non-negative, the objective function... Weight combination that reaches the maximum value .

[0028] Furthermore, to ensure accurate judgment of the synchronization between the UAV's attitude changes and spectral drift, the system integrates a data stream alignment mechanism based on hardware timestamps. The pulse-per-second (PPS) signal output by the UAV's GPS module, a high-precision time reference provided by the GPS satellite system, is used as a unified clock source to simultaneously trigger data sampling by the inertial measurement unit (IMU) and image exposure by the hyperspectral imager, adding nanosecond-level precision synchronization timestamps to the data packets of both. During online processing, the system uses a fixed-size time window, such as 50 milliseconds, to match and associate the IMU data points with the closest timestamps with the spectral feature centroid data points, thus forming synchronization event pairs. The maneuvering flight angle threshold used to determine the collapse scenario is a variable dynamically adjusted based on real-time wind speed. Its data comes from an offline-built lookup table, which is established by analyzing historical attitude data of the UAV flying along a standard flight path at different wind speed levels. This table maps the real-time wind speed input to a specific angle change rate threshold output, which is set to the 99th percentile of the statistical distribution of historical attitude angle change rates under the corresponding wind speed conditions. These are all extended implementation methods known to those skilled in the art.

[0029] Example 1: In a large, automated irrigated corn planting base that is highly sensitive to weather and crop growth, the technical solution of this invention was deployed to support precision fertilization and disaster assessment based on crop growth. The climate characteristics of this area are frequent localized short-term strong winds and drastic changes in sunlight caused by severe convective weather in the summer afternoons, which poses a severe challenge to any continuous remote sensing monitoring. When a UAV equipped with the measurement method of this invention performs routine monitoring tasks over the tens of thousands of acres of cornfields in this base, its inherent technical architecture advantages are demonstrated through its response to a series of continuous events. In the early stages of the UAV's flight, a rapidly moving cumulonimbus cloud causes a drastic fluctuation in ground sunlight intensity in a short period of time. Traditional methods that rely on external references or fixed time interval calibration will suffer from severe distortion in reflectivity calculations due to the inability to respond synchronously, or sacrifice monitoring operations due to frequent interruptions of the flight path for physical calibration. The continuity of data accuracy and monitoring continuity is a long-standing inherent contradiction in the field of dynamic remote sensing. The present invention provides a fundamental solution by comparing the drift rate of the centroid of spectral features with the threshold of physiological change. When the system detects that the centroid drift rate far exceeds the upper limit of the normal physiological rhythm of corn, it definitively attributes it to a sudden change in external light and instantaneously generates a correction factor to compensate for subsequent data streams. The entire process is automatically closed during flight without affecting the continuity of monitoring, thus avoiding the aforementioned inherent contradiction within a single architecture. The essence of this mechanism is to redefine the calibration problem from how to accurately measure and compensate for an external physical variable to whether the current spectral change rate conforms to biological laws. It no longer seeks external dependence on complex environmental models, but instead turns to the internal verification of the system's endogenous biological rhythms, thus making the calibration process simple, self-consistent, and robust.

[0030] Subsequently, an area ahead of the drone's flight path encountered a downburst brought by the aforementioned severe convective weather, causing some corn to lodging. When the drone flew over this area, its onboard inertial measurement unit data and spectral data simultaneously exhibited drastic instantaneous jumps. At this critical juncture, the synergistic effect between different technical features in this invention was triggered. On the one hand, the drastic drift of the spectral feature centroid exceeded the physiological change threshold, triggering anomaly diagnosis. On the other hand, the instantaneous change rate of the flight attitude angle was strictly synchronized with the spectral drift in time, allowing the system to definitively attribute the event to the lodging observation scenario, rather than catastrophic biological stress, according to preset rules. The direct result of this attribution was to prevent the huge spectral artifacts caused by physical lodging from being incorrectly included in the time-series micro-oscillation energy analysis model used for early stress warning, thereby ensuring the long-term stability and reliability of the model's diagnostic baseline. Simultaneously, the lodging attitude-spectral morphology response model invoked by the system began to analyze the lodged area. Spectral correction makes it possible to assess the post-disaster physiological state of lodged corn. Therefore, the precise physical identification of lodged scenarios and the continuous monitoring of physiological states form a logical positive feedback loop: the former provides uncontaminated data input for the latter, and the latter, based on this input, endows the former with practical value for post-disaster diagnosis. The system as a whole demonstrates a complete analytical capability for complex and compound disaster scenarios. The measurement method, as demonstrated in the entire application example, embodies the fundamental principle of its system architecture: its reliability and accuracy are not based on dependence on external high-precision sensors or complex physical models, but rather on the deep utilization of the intrinsic life rhythms of the monitored object itself. By switching the measurement benchmark from an externally uncertain physical environment to an internally relatively stable biological temporal logic, this method constructs a measurement closed loop capable of self-calibration and recognizing the true physiological state with extremely low hardware redundancy, achieving a high degree of adaptability to real agricultural ecosystems.

[0031] Example 2: To quantitatively verify the effectiveness of the online calibration method proposed in this invention in distinguishing between crop physiological evolution and light-induced abrupt changes, this example was designed and executed. The purpose of the experiment is to demonstrate that, by analyzing the temporal drift rate of the spectral characteristic centroid, this method can accurately attribute and isolate apparent spectral distortions caused by sudden changes in environmental light without relying on any external reference, while maintaining accurate tracking of the slow evolution of the crop's true physiological state. To this end, we built a highly controllable indoor simulation test platform, which consists of a quadcopter drone equipped with a Cucumber S185 hyperspectral imager, The system consists of a programmable full-spectrum LED array lighting system covering a 10-square-meter area, and an observation area containing 20 potted rice samples at the jointing stage with uniform growth. A drone is fixed to a gantry that can move precisely along a preset trajectory in three-dimensional space, operating at a height of 2 meters above the ground at a speed of 1 meter per second. This avoids interference from changes in flight attitude on spectral acquisition, ensuring the purity of experimental variables. The experimental parameters are set according to a prudent engineering trade-off logic. The hyperspectral data acquisition cycle is set to 100 milliseconds, a technical consideration aimed at achieving an optimal balance between the ability to instantly capture sudden changes in illumination and the data processing bandwidth of the onboard processor. The rule for setting this period is as follows: its reciprocal, i.e., the sampling frequency of 10 Hz, must be significantly higher than the Nyquist frequency of the fastest environmental change frequency in the observed scene. This ensures that the complete form of the illumination change process can be reconstructed without distortion, providing high-density data support for the accurate calculation of the centroid drift rate. The time window used for calculating the drift of the spectral feature centroid is 1 second, corresponding to 10 consecutive sampling points. This parameter aims to balance the smoothing ability of the sensor's inherent noise with the response delay to the target event. Too short a window will amplify noise artifacts, while too long a window will dull real instantaneous changes. This value is determined based on the sensor's noise power spectrum and the duration of the target illumination change. After analysis, the signal-to-noise ratio optimization engineering trade-off achieved, and the setting logic of the physiological change threshold are rooted in the prior knowledge modeling of the rate of physiological change in crops. It is the core benchmark for distinguishing normal physiological evolution from abnormal external disturbances. The threshold is established based on the statistical analysis of historical big data on the daily maximum natural drift of key spectral features of Jimai 22 caused by water deficit during the jointing stage. It is set as an unrestricted example value, namely the equivalent centroid drift rate of 0.8 nanometers per minute. This value represents the statistical limit of the centroid drift rate that can be caused by the physiological changes of the crop itself under the condition of no severe external disturbance.

[0032] The experimental procedure was carried out within a continuous 3-hour monitoring cycle. In the first stage, from 0 to 2 hours, the LED array continuously irradiated the wheat samples with constant power to simulate clear, cloudless weather. Irrigation was stopped to induce a natural, slow-onset drought stress process. The system continuously collected hyperspectral data and calculated the position and drift rate of the spectral feature centroid in real time. The observed phenomenon was that the centroid position exhibited a slow but continuous directional drift towards the shortwave direction. In the second stage, from 2 to 3 hours, while maintaining the drought stress, at 2 hours 30 minutes 0 seconds, the light intensity of the LED array was programmed to drop by 70% within 500 milliseconds to simulate rapid obstruction by thick clouds for 30 seconds, and then returned to the original intensity within 500 milliseconds. This process aimed to introduce a violent external light disturbance unrelated to physiological changes. The data records at key time points are shown in the table below. Table 1: Comparison of spectral feature centroid drift rate and system judgment action time points.

[0033] The above experimental data directly corroborates the inherent calibration mechanism of this invention. In the first two hours of the experiment, the observed centroid drift rate was consistently lower than the preset physiological change threshold. Based on this, the system correctly attributed the slow drift to the actual evolution of the crop's internal physiological state caused by water deficit and continuously incorporated the current centroid position as the new dynamic calibration benchmark. However, at the instant of the sudden change in light intensity at the 150-minute mark, the centroid drift rate instantly jumped to an order of magnitude far exceeding the threshold. This dramatic change was accurately captured by the system's inherent comparison logic and immediately triggered the decision branch. As a result, the system refused to update the calibration benchmark and generated an instantaneous correction factor to offset the impact of this disturbance on subsequent spectral data, thereby ensuring the continuity and authenticity of the measurement results.

[0034] Example 4: This example combines Figures 1 to 3 This paper describes the implementation of an online hyperspectral calibration measurement method for agricultural remote sensing using unmanned aerial vehicles (UAVs). Figure 1As shown, at 8:00 AM, under normal light conditions, the crops (taking rice as an example) were in a stable physiological state. The hyperspectral imager began collecting the first set of spectral data. Based on this, the processing system calculated the current spectral characteristic centroid C0 of the crop and set it as the initial calibration benchmark. As time progressed for 30 minutes, the moisture content of the crop leaves decreased slightly by about 2%, and the chlorophyll concentration increased slightly, indicating that the crop was in a normal transpiration and photosynthesis process. At time T1, the system collected the second set of spectral data and calculated the centroid C1. The centroid drift rate at this time was 0.45 nm / min, which was lower than the preset physiological change threshold of 0.8 nm / min, and was determined to be caused by physiological changes in the crop itself. The system updates the calibration benchmark to C1 and records the current centroid change trajectory, initially identifying the signal trend of water stress. After another 30 minutes, the crop leaf water content continues to decrease, and the canopy structure undergoes slight adjustments. At time T2, the system collects the third set of spectral data, and the centroid is updated to C2, with a corresponding drift rate of 0.68 nm / min, still within the physiological threshold range. The processing system continues to update the calibration benchmark and performs trend analysis on the data from multiple time points before and after. After receiving continuous centroid change data, the diagnostic module, combined with changes in microscopic drift direction and energy indicators, identifies that the crop is under continuous water stress, and finally outputs a diagnostic conclusion, recommending the initiation of irrigation intervention measures.

[0035] like Figure 2 As shown in the figure, three main growth stages are marked: the technical stage, the heading stage, and the maturity stage, and represented by different line types. The horizontal axis represents wavelength (nm), and the vertical axis represents reflectance (%). The reflectance change during the technical stage shows a relatively gradual trend, especially when the wavelength is greater than 600nm, the reflectance begins to gradually increase. During the heading stage, the spectral reflectance shows a significant increase, especially near the green light reflectance peak, indicating a characteristic band of crop physiological changes. In the maturity stage, the increase in reflectance is further amplified, especially near the red edge inflection point, and the reflectance of the near-infrared plateau also shows a continuous increasing trend. The positions of the green light reflectance peak, the red edge inflection point, and the near-infrared plateau are also marked in the figure to show the spectral characteristics of the crop at each stage. These bands are of great significance for monitoring the physiological state and health assessment of crops and can effectively support the application of hyperspectral remote sensing technology in agricultural remote sensing.

[0036] like Figure 3As shown, firstly, in the initial stage, the system acquires continuous hyperspectral data of the crop area through step a, using the acquired continuous spectral data as the basis for subsequent processing. Next, step b involves extracting spectral features and calibrating the feature centroids. By extracting feature bands such as red edge inflection points, green light reflection peaks, and near-infrared plateaus, the spectral feature centroids of the crop are calibrated. In step c, the drift of the spectral feature centroids per unit time is calculated in real time, and the dynamic change of the centroids is calculated. The stability of the spectral data is monitored in real time. The process then enters the decision-making part: step d determines whether the drift exceeds the physiological threshold. If it does not exceed the physiological change value (determined as no), then the drift is considered to be due to crop... The changes are caused by normal physiological changes in the crop, and are judged as physiological changes in step e. The latest calibration benchmark is updated, the change data of the spectral centroid is recorded, and new health status parameters are updated. If the drift exceeds the physiological threshold (judged as yes), the process proceeds to step f to determine if it is caused by external disturbances, such as changes in light intensity or posture. The system generates a correction factor to correct the data and continues to perform subsequent analysis operations. Finally, in the early stress diagnosis module, combined with steps g1 to g3, the micro-oscillation energy is calculated through high-frequency sampling to identify possible crop stresses. The entire process achieves efficient and real-time monitoring and early warning of crop physiological status through dynamic monitoring and analysis of spectral data.

[0037] Example 4: This example provides an integrated explanation of the entire measurement system's offline calibration and online operation cycle. Before deploying drones for field operations, a rigorous offline modeling and parameter calibration phase is completed. The output of this phase provides a deterministic quantitative basis for subsequent online attribution and calibration. This calibration process uses rice as the target crop and is conducted during its jointing stage. First, for the physiological change threshold, which is the core judgment criterion, the calibration follows a standardized statistical analysis process. A healthy sample group of the same variety and growth stage as the target crop is selected, and continuous hyperspectral imaging monitoring is conducted for 72 hours in a growth chamber with controlled environmental parameters. During this period, by gradually reducing the water supply, a slow physiological change process driven by drought stress is induced in the crop. The processing system collects hyperspectral data at a frequency of once per minute and calculates the spectral feature centroid based on the red edge inflection point wavelength, green light reflection peak wavelength, and near-infrared platform starting point wavelength defined in this invention. It is precisely defined as an equivalent wavelength after weighted fusion, and its calculation follows the formula ,in These represent the real-time positions of the three characteristic wavelengths mentioned above, and their weighting coefficients are respectively... satisfy Constraints were imposed and determined based on prior agronomic knowledge to reflect the sensitivity of each characteristic to changes in overall physiological state. For drought stress, the near-infrared platform initiation wavelength, which is most sensitive to changes in canopy moisture, was assigned the highest weight. This was achieved by analyzing the centroid drift over all consecutive one-minute time intervals throughout the entire monitoring period. Calculations and statistics were performed to obtain a drift distribution containing thousands of sample points. This distribution approximates a normal distribution centered at zero, and its standard deviation characterizes the amplitude of natural fluctuations in the spectral centroid caused by purely physiological changes in crops.

[0038] Secondly, the identification and calibration of crop lodging, a specific physical scenario, relies on the construction of a lodging attitude-spectral morphology response model, which is also completed offline. Using a mechanical platform with precisely controllable tilt angles, under the same controlled lighting conditions as described above, a single healthy crop sample is subjected to attitude simulations ranging from zero degrees vertical to a completely flat 90 degrees, in five-degree increments. At each tilt angle, a UAV hyperspectral imager hovers stably above, recording the corresponding spectral feature centroid position. By fitting the centroid position data at all angles, a lodging attitude-spectral morphology response model can be established. A low-order polynomial function accurately maps the deterministic relationship between the physical tilt angle of the crop canopy and the drift of the spectral characteristic centroid. This model is stored in the onboard processing system. Its function is to calculate and eliminate spectral artifacts introduced by physical morphological changes based on the UAV attitude angle provided by the inertial measurement unit once lodging is detected online. Simultaneously, the physiological resting baseline for early stress warning is established by analyzing the time-series signal of the spectral characteristic centroid of completely healthy, stress-free crop samples under constant conditions for 24 hours, using a variance calculation formula. The variance of the centroid time-series signal of the healthy sample is calculated, and this value is defined as the baseline of micro-oscillation energy representing physiological homeostasis.

[0039] When the drone, carrying the thresholds, models, and baselines established through the aforementioned offline calibration, enters the actual field operation scenario, the entire online calibration and diagnostic process operates efficiently and reliably. The drone flies along the flight path, continuously acquiring hyperspectral data cubes, and the onboard processing system calculates the position of the spectral feature centroids in real time. and its drift amount per unit time The system will The absolute value and the calibrated physiological change threshold Perform real-time comparison, if The system determines that this change originates from the crop's intrinsic physiological activities. At this point, the calibration benchmark is updated to the current centroid position to ensure that the measurement system can flexibly follow the crop's natural growth rhythm. Simultaneously, the system caches the centroid position at high frequencies and calculates its temporal micro-oscillation energy. And compare it with the physiological resting baseline. Once the energy value remains above the baseline, an early stress warning is triggered; conversely, if... The system determines that the drift is caused by external interference. At this moment, the system immediately queries the attitude angle change rate of the airborne inertial measurement unit. If the attitude angle change rate also shows a drastic fluctuation exceeding the threshold, the system definitively attributes the event to the observation of lodged crops and immediately calls the preset lodging attitude-spectral morphology response model. Based on the current attitude angle, the spectral feature centroid is corrected, so that the true physiological state of the crop can still be calculated under the interference of drastic changes in physical morphology. If the spectral drift exceeds the threshold but the attitude is stable, it is determined to be a sudden change in light intensity. At this time, the calibration benchmark remains unchanged, and a temporary correction factor is generated based on the magnitude and direction of the drift to perform reverse compensation on the subsequent data stream until the light intensity returns to stability. This offline calibration and online decision-making process together constitute a measurement closed loop that does not rely on external references and can adapt to complex farmland environments.

[0040] It will be apparent to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential characteristics of the present invention.

[0041] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention.

Claims

1. A method for online hyperspectral calibration measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing, characterized in that, The method includes the following steps: Step a: Obtain continuous hyperspectral data of the crop area; Step b: Extract multiple spectral feature wavelengths from continuous hyperspectral data in real time, based on crop type and growth stage, and construct spectral feature centroids that characterize the current hyperspectral morphology based on the multiple spectral feature wavelengths. Step c: Calculate the drift of the spectral feature centroid per unit time in real time; Step d: Compare the drift of the spectral feature centroid with the physiological change threshold determined based on historical data statistical analysis and modeling, and perform attribution determination of the spectral centroid drift based on the comparison results; Step e: When the drift of the spectral feature centroid does not exceed the physiological change threshold, it is determined that the drift of the spectral feature centroid is caused by the change in the physiological state of the crop, and the current spectral feature centroid is updated to the new calibration benchmark. Step f: When the drift of the spectral feature centroid exceeds the physiological change threshold, it is determined that the drift of the spectral feature centroid is caused by external illumination or changes in the attitude of the UAV. At this time, the calibration benchmark is not updated, and a correction factor is generated to perform real-time calibration on the subsequently acquired continuous hyperspectral data.

2. The method for online hyperspectral calibration and measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing according to claim 1, characterized in that, The multiple spectral feature wavelengths used to construct the spectral feature centroid in step b include: the red edge inflection point wavelength, the green light reflection peak wavelength, and the near-infrared plateau initiation point wavelength.

3. The method for online hyperspectral calibration and measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing according to claim 1, characterized in that, In step d, the physiological change threshold is determined based on statistical analysis and modeling of historical spectral data of the target crop type at different growth stages.

4. The method for online hyperspectral calibration and measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing according to claim 1, characterized in that, In step e, when it is determined that the centroid drift of the spectral features is caused by changes in the physiological state of the crop, the method further includes: using the amplitude and direction of the centroid drift as an indicator of the degree and type of physiological stress on the crop.

5. The method for online hyperspectral calibration and measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing according to claim 1, characterized in that, In step f, when the drift of the spectral feature centroid exceeds the physiological change threshold, the method further includes the following sub-steps: Step f1, obtain the current flight attitude data from the UAV's onboard inertial measurement unit; Step f2, perform crop lodging observation scenario determination based on the flight attitude data; Step f3, call the pre-established lodging attitude-spectral morphology response model, and correct the current spectral feature centroid according to the model.

6. The method for online hyperspectral calibration and measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing according to claim 5, characterized in that, In step f2, the determination of the crop lodging observation scenario is performed by comparing the instantaneous rate of change of the flight attitude angle with a predetermined maneuvering flight angle threshold. If the rate of change exceeds the maneuvering flight angle threshold and is synchronized with the drift of the spectral feature centroid in time, it is determined that the crop lodging observation scenario is in progress.

7. The method for online hyperspectral calibration and measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing according to claim 5, characterized in that, The lodging posture-spectral morphology response model is a calibrated and stored lookup table or low-order polynomial function, which is established by data acquisition and model training of simulated lodged crops at different tilt angles.

8. The method for online hyperspectral calibration and measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing according to claim 1, characterized in that, The method also includes temporal micro-oscillation energy analysis of the spectral feature centroid for early warning of early physiological stress in crops. The analysis includes the following steps: Step g1, buffering the position data of the spectral feature centroid over a continuous time period at a high sampling frequency to form a centroid time-series signal; Step g2, processing the centroid time-series signal to calculate the total energy index of micro-oscillations, which is obtained by calculating the variance of the centroid time-series signal; Step g3, comparing the total energy index of micro-oscillations with the physiological resting baseline in real time. The physiological resting baseline represents the normal energy level of crop micro-oscillations. When the index is consistently higher than the physiological resting baseline, early stress warning information is output.

9. A method for online hyperspectral calibration and measurement of unmanned aerial vehicles (UAVs) for agricultural remote sensing according to claim 8, characterized in that, In step g2, the variance of the centroid time-series signal The calculation method is as follows: in, Indicates the first The centroids of spectral features collected at each time point This represents the total number of data points in the buffered centroid timing signal. This represents the average value of the centroid timing signal of the cache.

Images