Corn kernel nutrient abundance and deficiency state evaluation method based on spectral analysis

Abstract

The invention relates to the technical field of infrared spectroscopic analysis, in particular to a spectroscopic analysis-based corn kernel nutrient abundance and deficiency state evaluation method, which comprises the following steps of: irradiating corn kernels by using near-infrared and mid-infrared light sources, collecting reflected spectral signals, and recording spectral signal data; and based on the spectral signal data, extracting key wavelength region data to obtain key spectral data. According to the invention, the near-infrared light source and the mid-infrared light source are combined, so that the capability of acquiring nutrient information of the corn kernels is enhanced, the reflection characteristic of spectral signals is utilized, the interference of environmental factors on data acquisition is reduced, and the stability of spectral data is improved. The screening of key wavelength regions reduces irrelevant information, so that the spectral data is more concentrated on the characteristic wave bands influencing the nutrient state. The spectral signals are processed through wavelet transform, the resolution ratio of the characteristic peak value is improved, the local change trend of the spectral signals is enhanced, and characteristic extraction is more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of infrared spectroscopy analysis, and particularly relates to a method for evaluating the nutrient abundance and deficiency status of corn kernels based on spectral analysis. Background Art

[0002] Infrared spectroscopy analysis is a spectral technique for detecting the composition of substances and analyzing their structures based on the vibration characteristics of molecules. It identifies and quantitatively analyzes chemical components by detecting the absorption characteristics of substances to infrared light. This technique utilizes the characteristic absorption peaks generated by molecules under the irradiation of infrared light to reflect the stretching and bending vibrations of chemical bonds inside the molecules.

[0003] However, in the prior art, during the acquisition process of infrared spectral data, it is affected by the stability of the light source, environmental light interference, and the surface characteristics of the sample, resulting in large fluctuations in some spectral signals and affecting data consistency. Moreover, the resolution ability of characteristic peaks is affected by signal noise, and some weak nutrient-related peaks are easily covered by background noise, affecting the accuracy of subsequent analysis. Therefore, improvements are needed. Summary of the Invention

[0004] The purpose of the present invention is to solve the deficiencies existing in the prior art, and to propose a method for evaluating the nutrient abundance and deficiency status of corn kernels based on spectral analysis.

[0005] To achieve the above purpose, the present invention adopts the following technical solutions. A method for evaluating the nutrient abundance and deficiency status of corn kernels based on spectral analysis includes the following steps: Irradiate corn kernels with near-infrared and mid-infrared light sources, collect the reflected spectral signals, and record the spectral signal data; based on the spectral signal data, extract the data in the key wavelength region to obtain the key spectral data; Based on the key spectral data, apply wavelet transform to analyze the characteristic peaks in the data, enhance the identifiability of spectral peaks, and generate enhanced spectral data; based on the enhanced spectral data, calibrate the spectral peaks associated with nutrients, record the position and intensity of each peak, and obtain the spectral peak calibration result; Based on the spectral peak calibration result, conduct a comparative analysis with the standard nutrient spectra in the database, extract the characteristics of the data, generate a data feature set; based on the data feature set, use a support vector machine to classify the nutrient status, distinguish each nutrient deficiency status, and obtain the nutrient status classification result; Based on the nutrient status classification result, conduct a quantitative evaluation, set the threshold standards for different nutrients, analyze the abundance and deficiency of each nutrient element, and describe the deficiency status for nutrients below the threshold standard to obtain the nutrient abundance and deficiency evaluation result.

[0006] Preferably, the step of obtaining the spectral signal data is as follows: Set near-infrared and mid-infrared light sources to irradiate corn kernels, and simultaneously capture the spectral signals reflected from the corn kernels, convert the spectral signals into digital format, and obtain complete spectral signal data; Record and store the complete spectral signal data to obtain spectral signal data.

[0007] Preferably, the steps for obtaining the key spectral data are as follows: Screen the wavelength regions sensitive to the nutrient evaluation of corn kernels from the spectral signal data, and determine the key wavelengths affecting the nutrient analysis by comparing the absorption intensities and peak shape characteristics of each wavelength, so as to obtain the preliminarily screened wavelength region data; Based on the preliminarily screened wavelength region data, eliminate the non-key wavelength points to obtain the key wavelength region data; Based on the key wavelength region data, perform spectral feature extraction, including calculating the peak position, peak width, and integral peak area, to obtain the key spectral data.

[0008] Preferably, the steps for obtaining the enhanced spectral data are as follows: Based on the key spectral data, analyze the characteristic peaks through wavelet transform to obtain the wavelet transform result. The formula is: Wherein, represents the transformed peak intensity at the frequency , is the wavelet coefficient at the time and the frequency points, is the spectral intensity at the time , represents the total number of time points; Based on the wavelet transform result, identify the spectral peaks through peak comparison and intensity analysis to obtain the enhanced spectral data.

[0009] Preferably, the steps for obtaining the spectral peak calibration result are as follows: Based on the enhanced spectral data, extract all spectral peaks, analyze the local extreme points of the spectral curve in different wavelength ranges, and record the wavelength and spectral intensity to generate preliminary spectral peak data; According to the preliminary spectral peak data, calculate the nutrient correlation degree corresponding to the peak. The calculation formula is: Wherein, is the correlation degree between the spectral peak and the th nutrient, is the spectral intensity at the wavelength , is the standard spectral response curve of the th nutrient, and are the starting and ending wavelengths of the analysis wavelength range; Based on the nutrient correlation degree corresponding to the peak value, spectral peaks highly correlated with the target nutrient are screened, the wavelength positions and spectral intensities of the peak values are sorted out, and a spectral peak calibration result is generated.

[0010] Preferably, the steps for obtaining the data feature set are as follows: Based on the spectral peak calibration result, the standard nutrient spectral data in the database is retrieved, the changing trends of the spectral curves of the two groups of spectra within the same wavelength range are compared, and the changes in the positions, widths, and intensities of the spectral absorption peaks are analyzed to obtain a standard nutrient spectral matching result; According to the standard nutrient spectral matching result, the principal component contribution degree is calculated, and the calculation formula is: where, represents the principal component contribution degree, represents the spectral intensity of the th wavelength in the spectral peak calibration result, represents the spectral intensity of the corresponding wavelength in the standard nutrient spectral database, represents the spectral absorption difference corresponding to the th wavelength, represents the comparison matching error of the th wavelength, represents the total number of wavelength points; Based on the principal component contribution degree, principal component analysis is performed, spectral feature variables are extracted, a feature vector space is constructed, the spectral data is dimensionally reduced, noise and redundant variables are removed, and a data feature set is generated.

[0011] Preferably, the steps for obtaining the nutrient status classification result are as follows: Based on the data feature set, the kernel function type and kernel parameters of the support vector machine classifier are configured to obtain a configured SVM model; According to the configured SVM model, a classification result function is obtained, and the expression is: where, is the classification result function, is the sign function, is the Lagrange multiplier corresponding to the support vector, is the th class label of the data point, is the similarity between the th support vector calculated by the kernel function and the input , is the bias term, is the number of support vectors; Based on the classification result function, classify the nutrient status of each data point, distinguish between sufficient and deficient nutrient states, evaluate the degree of deficiency of each nutrient, and generate the classification result of the nutrient status.

[0012] Preferably, the steps for obtaining the nutrient abundance and deficiency evaluation result are as follows: According to the classification result of the nutrient status, calculate the nutrient abundance and deficiency metric, and the calculation formula is: Wherein, represents the nutrient abundance and deficiency metric, is the measured nutrient concentration, is the threshold standard of this nutrient, is the maximum response concentration of this nutrient in the standard data, is the minimum response concentration of this nutrient, is the nutrient concentration change rate; Based on the nutrient abundance and deficiency metric, analyze the abundance and deficiency of each nutrient element, and generate the nutrient abundance and deficiency evaluation result.

[0013] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In the present invention, the combination of near-infrared and mid-infrared light sources enhances the ability to obtain maize grain nutrient information. Utilizing the reflection characteristics of spectral signals, it reduces the interference of environmental factors on data collection and improves the stability of spectral data. The screening of the key wavelength region reduces irrelevant information, making the spectral data more concentrated on the characteristic bands affecting the nutrient status. Wavelet transform processes the spectral signal, improves the resolution of the characteristic peaks, strengthens the local change trend of the spectral signal, and makes the feature extraction more accurate. The calibration process of the spectral peak correlates the characteristic peak with the specific nutrient content. Through data structured storage, it improves the operability of subsequent analysis. When comparing with the database, the feature extraction method is used to reduce the data dimension, improve the calculation efficiency of spectral data, and at the same time reduce the influence of redundant information. The support vector machine classification method optimizes the discrimination ability of the nutrient status, improves the generalization of the model, and still maintains stable classification performance under different spectral data. The quantitative evaluation stage combines the nutrient threshold standard to clearly divide the nutrient abundance and deficiency status, provides a numerical description of the nutrient status, and improves the interpretability of the data. Description of the Drawings

[0014] Figure 1 It is a step schematic diagram of the present invention. Detailed Embodiment

[0015] In order to make the objectives, technical solutions and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0016] Please refer to Figure 1 , the present invention provides a technical solution, a method for evaluating the nutrient deficiency status of corn kernels based on spectral analysis, comprising the following steps: Irradiate corn kernels with near-infrared and mid-infrared light sources, collect the reflected spectral signals, and record the spectral signal data; based on the spectral signal data, extract the data in the key wavelength region to obtain the key spectral data; Based on the key spectral data, apply wavelet transform to analyze the characteristic peaks in the data, enhance the recognizability of the spectral peaks, and generate enhanced spectral data; based on the enhanced spectral data, calibrate the spectral peaks associated with nutrients, record the position and intensity of each peak, and obtain the spectral peak calibration result; Based on the spectral peak calibration result, compare and analyze with the standard nutrient spectra in the database, extract the characteristics of the data, and generate a data feature set; based on the data feature set, use a support vector machine to classify the nutrient status, distinguish each nutrient deficiency status, and obtain the nutrient status classification result; Based on the nutrient status classification result, conduct a quantitative evaluation, set the threshold standards for different nutrients, analyze the abundance and deficiency of each nutrient element, and describe the deficiency status of the nutrients below the threshold standard to obtain the nutrient abundance and deficiency evaluation result.

[0017] The steps for obtaining the spectral signal data are as follows: Set the near-infrared and mid-infrared light sources to irradiate the corn kernels, simultaneously capture the spectral signals reflected from the corn kernels, convert the spectral signals into digital format, and obtain the complete spectral signal data; Record and store the complete spectral signal data to obtain the spectral signal data.

[0018] Specifically, based on the set near-infrared and mid-infrared light source ranges, first determine in the experimental environment to use a near-infrared light source with a wavelength between 780 nm and 2500 nm and a mid-infrared light source between 2500 nm and 25000 nm to irradiate the surface of the corn kernels. Refer to a previously established document on light source application parameters, which was compiled by conducting optical reflection measurements on 100 batches of corn kernel samples and statistically analyzing their reflection peak intervals. Then, use an optical signal acquisition device to detect the reflection situation on the surface of the kernels under the light source irradiation. The device digitizes the analog optical signal with a 16-bit analog-to-digital converter, and the sampling frequency is set to 100 times per second. This setting is obtained after multiple actual measurements of the reflection waveform stability. Then, calibrate the intensity of the collected original waveform between 0 and 1 and generate a preliminary data set. If it is found that some peaks exceed the pre-set reflection intensity threshold , where 0.85 is obtained by taking the average value of multiple measurements on 300 random grains and adding 0.05. Then, this part of the peak information is regarded as the noise area and marked. Next, the remaining reflected signals are separated from the background noise. By comparing the spectral curves of different acquisition batches, it is verified whether there are abnormal fluctuations. For example, if a single acquisition curve lasts for more than 2 seconds above 0.9 or below 0.02, it is regarded as abnormal. Finally, the digital data within the threshold range is uniformly integrated and recorded in the memory structure to obtain the complete spectral signal data.

[0019] When recording and storing the complete spectral signal data, first allocate storage space and index tables according to a parameter configuration description document for the storage structure. This document is compiled by combining the maximum capacity threshold of 500,000 entries deduced from previous spectral data scales and evaluating the index conflict situation. Then, arrange each digital spectral record in chronological order and assign a unique index number. Before writing, perform error checking on each record to determine whether there are sudden jumps or extreme outliers. For example, if a value of the reflection intensity between 0 and 1 suddenly exceeds 1.2 or is lower than -0.1, it is regarded as invalid and marked for archiving. If it is detected multiple times that the number of records approaches 500,000 for the written data, a threshold warning will be triggered. The threshold is statistically calculated at a frequency of adding 20,000 new data each time, and exceeding this range is the trigger criterion. Then, continue to supplement entries according to the index table during storage. If it is confirmed that the record meets the currently set index section, continue to write to the memory and keep the entry numbers in order. Finally, after completing the above operations, the spectral signal data is obtained.

[0020] The steps for obtaining key spectral data are as follows: Screen the wavelength regions sensitive to the evaluation of maize grain nutrients from the spectral signal data. By comparing the absorption intensities and peak shape characteristics of each wavelength, determine the key wavelengths affecting nutrient analysis to obtain the initially screened wavelength region data; Based on the initially screened wavelength region data, eliminate non-key wavelength points to obtain the key wavelength region data; Based on the key wavelength region data, perform spectral feature extraction, including calculating the peak position, peak width, and integral peak area, to obtain the key spectral data.

[0021] Specifically, to screen the wavelength regions sensitive to the evaluation of maize grain nutrients from the spectral signal data, first, conduct a preliminary review of the collected spectral curve to find the bands where significant absorption peaks may exist. Then, refer to a record of the band ranges pre-statistically obtained from 200 batches of maize grain samples, which lists the common positions of absorption peaks and their peak height distributions in each frequency band within the range of 300 nm to 2500 nm. Then, gradually check the correspondence between the wavelength and intensity on the actual spectral curve to determine whether the peak height in each local wavelength segment exceeds the intensity threshold , for example, dividing the aforementioned wavelength band into several 5-nm sub-intervals and calculating the difference between the highest absorption peak and the lowest absorption valley therein. If the difference is greater than 0.12, it is included in the subsequent comparison. This value of 0.12 is obtained by adding 0.02 to the average peak-valley difference of the measurement results of different corn kernels. On this basis, the peak shape characteristics of the included sub-intervals are further evaluated. By calculating the slope change from the peak top to the peak shoulder and the ratio of the peak width to the length of the sub-interval, if the slope continuously changes by more than 0.03 set by prior experience and the peak width ratio exceeds 0.2, then this sub-interval is regarded as a potential sensitive area. After summarizing the wavelength information of all potential sensitive areas, it is then checked whether there are locally overlapping or overly close sections. If the wavelength distance between two adjacent sections is less than 2 nm, they are merged and the average value of the peak shape characteristics after merging is calculated to determine whether it still meets the aforementioned criteria. Finally, all sections that meet the requirements are retained to form a relatively complete sensitive wavelength range, and the wavelength region data of the preliminary screening is obtained.

[0022] Based on the wavelength region data of the preliminary screening, first check the average absorption intensity and the peak position for each divided wavelength segment. If the situation where the value is lower than 40% of the segment mean and lasts for two wavelength sampling points occurs within the segment, it is judged as a noise point and excluded. This ratio of 40% is obtained by statistically analyzing the extreme low values of 50 measured corn kernel spectral samples and upwardly correcting the mean by 0.05. Then, by comparing the change rate of each wavelength point between two adjacent points, if the change rate exceeds 0.08, it is regarded as an unstable point and excluded. This value of 0.08 is obtained by adding 0.01 to the average fluctuation range selected after multiple rounds of regression analysis of the peak volatility under the same measurement conditions. Then, the positions of all excluded points are saved in the memory structure for subsequent abnormal distribution inspection. If it is found that more than 3 consecutive wavelength points are excluded, an additional scan of 1 nm above and below is performed to exclude potential edge noise points. Finally, when merging the remaining wavelength segments, if the wavelength difference between adjacent segments is within 1 nm, they are combined into the same segment. The total length after merging is compared with the pre-set lower limit of the effective length of 5 nm. If it is greater than 5 nm, it is regarded as a retained segment; otherwise, it is merged into the adjacent larger segment. After multiple comparison and merging operations, a robust wavelength range is obtained, and the key wavelength region data is obtained.

[0023] Based on the key wavelength region data, when locating the peak position within each retained wavelength segment, first detect it at a step of 0.5 nm, find the point with the highest absorption intensity and record it as the peak top. Then, expand from the peak top to both sides until the absorption intensity is lower than half of the peak top to measure the peak width. If the sum of the wavelength distances between the left and right edges and the peak top exceeds 3 nm, this distance is recorded as the effective peak width; otherwise, it is regarded as a too-narrow peak and not retained. Next, when calculating the integrated peak area of each effective peak segment, let the wavelength resolution be , for the absorption intensity of each sampling point Perform discrete and computational approximate integration , where is the total number of sampling points after segmentation, is the average wavelength interval between each sampling point. At this time, if the peak area calculated for a certain segment exceeds twice the average level of this segment, it will be registered as a suspicious peak and the peak intensity will be re-compared to exclude abnormal jumps in the data record. Once it is confirmed that no jump has occurred, this peak will be retained; otherwise, an error mark will be made and it will not be included in the statistical scope for the time being. Finally, after calculating the peak position, peak width, and integrated peak area for each wavelength segment, key spectral data is obtained.

[0024] The steps for obtaining enhanced spectral data are as follows: Based on the key spectral data, analyze the characteristic peaks through wavelet transform to obtain the wavelet transform result. The formula is: where, represents the transformed peak intensity at frequency , is the wavelet coefficient at time and frequency points, is the spectral intensity at time , represents the total number of time points; Based on the wavelet transform result, identify the spectral peaks through peak comparison and intensity analysis to obtain the enhanced spectral data.

[0025] Specifically, the advantage of the formula lies in combining the time dimension and the frequency dimension, and using the product of the wavelet coefficient and the spectral intensity to achieve the energy aggregation at specific frequencies for different time points. Through this idea, the peak intensity at a certain frequency can be more directly quantified, and the frequency position where the characteristic peak is located can be more clearly located when analyzing the key spectral data.

[0026] The steps for obtaining For example, if the wavelet decomposition is divided into 5 layers, each layer contains 64 sampling points, then the frequency range of the single layer transitions from higher frequency band to lower frequency band according to the configuration of the actual measuring instrument. For example, the center frequency corresponding to the highest frequency layer is about 500 Hz, and the center frequency corresponding to the lowest frequency layer is about 31.25 Hz. The first frequency in the layer will be divided into The wavelet amplitude value corresponding to each position is normalized and finally recorded in the coefficient table to obtain the Data, if the total number of time points is 512, then Sampling can be completed from 0 to 511.

[0027] The steps to obtain are: at each time point The corresponding spectral intensity comes from the real-time detection of the reflected signal on the surface of the corn kernels, and is discretized by a 16-bit analog-to-digital converter to obtain an integer value between 0 and 65535. The integer value is then mapped to a value between 0.0 and 1.0 through a calibration coefficient. The calibration coefficient is determined based on a comparison table obtained by performing multiple rounds of spectral tests on samples from the same batch. The specific calculation method is: , where "mean dark current" is the base noise collected during the device self-test (obtained by collecting 1000 data points per second and averaging them, for example, the average dark current of the device is 200), and the maximum signal is the saturation value under the peak illumination of the light source (for example, 65535 corresponds to the maximum reflection that the device can measure). If the collected value is is 5000, the average dark current is 200, and the maximum signal is 65535. Approx. This mapping method can be multiplied with the wavelet coefficients in subsequent processing without additional dimensional conversion. The same formula can be applied to each time point to form a length of of Sequence and perform subsequent operations.

[0028] The acquisition step is to obtain the total number of time points after forming a unified index count from the start to the end of the spectrum data acquisition. For example, in the experiment, a sampling frequency of 500 times per second and continuous acquisition for 30 seconds can be used to obtain 15,000 valid time points. If there is an interruption in the acquisition process, the data index of the interruption range is removed and renumbered continuously, and finally a time index from 0 to 1 is obtained. A sequence of The specific value depends on the actual experimental conditions. The sampling time and sampling frequency can be flexibly adjusted by setting the sampling time and sampling frequency of the upper control of the acquisition device. The size. If it is found that higher precision is still required in the subsequent analysis stage, the sampling frequency can be appropriately increased or the acquisition time can be extended to expand Once the sampling is completed this time, it will be fixed for calculation.

[0029] The acquisition steps of are as follows: Divide the specific frequency values according to the center frequency corresponding to each frequency band in the wavelet analysis process. For example, if the device hardware frequency detection range is from 1 Hz to 1000 Hz, it can be divided into several sub-bands by logarithmic intervals, and a central value is determined for each sub-band as Then, when performing wavelet decomposition on the collected time series, locate the center frequency of each layer, extract the corresponding and perform a multiplication operation with If it is necessary to refine the interval, interpolation can be performed on the basis of the original sub-bands and the list of center frequencies can be updated, and finally a set of discrete frequency values is formed for subsequent calculation. For example, one of the center frequencies represents the component corresponding to 12 Hz, and the coefficient sequence at this frequency can be queried in the wavelet coefficient table and the above steps can be executed to obtain

[0030] The acquisition steps of are as follows: Represent the sampling point sequence of the passage of time with a sequence of consecutive integers starting from 0. Each corresponds to the spectral intensity value obtained by one device sampling. If the device is set to sample at a frequency of 500 times per second for 30 seconds, a total of 15000 sampling moments can be obtained. After numbering them the value range of is from 0 to 14999. Each can find the corresponding item in the previously calculated

[0031] Calculation process: Taking a specific situation as an example, select , the time points range from to , and set the target frequency to Hz. The wavelet coefficients corresponding to 5 moments at this frequency and the spectral intensity have the following values:

[0032]

[0033] Substitute into the formula as follows:

[0034] The result shows that at the frequency of 12 Hz, the comprehensive peak intensity calculated based on the wavelet coefficients and spectral intensity at 5 sampling moments is 0.623. The larger the value, the higher the overall energy level detected at this frequency, which has reference significance for subsequent judgment of whether there are significant peaks in this frequency region.

[0035] Based on the wavelet transform results, through peak comparison and intensity analysis, first find the calculated value corresponding to each frequency in the obtained frequency-intensity mapping table , and locate the regions where local protrusions may occur by comparing the numerical ranges at different frequencies. If it is found that the values in certain frequency bands significantly exceed the set intensity threshold, for example, the threshold is set to 0.4 and appropriately increased by 0.05 after reading the difference between the highest peak and the average peak in the test data of 200 batches of samples is about 0.35, and then check the adjacent frequencies to see if there is a continuous upward trend. If the continuous upward amplitude exceeds 0.05, mark this frequency as the candidate peak position, and check the peak distribution of each sampling point one by one within the adjacent frequency range according to the spectral peak search strategy. If the intensity in the adjacent frequency range remains above 0.35, it is determined that there is a steady-state peak here, and then record the maximum value corresponding to this steady-state peak for classification as an item to be summarized. If multiple local peaks are detected at the same time at certain frequencies, compare the differences between these peak points and select the point with the maximum value as the main peak, and classify the other peaks as secondary features. After completing the above search and comparison operations, the peak information of all potential peaks can be obtained, and enhanced spectral data can be obtained.

[0036] The steps to obtain the spectral peak calibration result are as follows: Based on the enhanced spectral data, extract all spectral peaks, analyze the local extreme points of the spectral curve in different wavelength ranges, and record the wavelength and spectral intensity to generate preliminary spectral peak data; According to the preliminary spectral peak data, calculate the nutrient correlation degree corresponding to the peak. The calculation formula is: where, is the correlation degree between the spectral peak and the th nutrient, is the spectral intensity at wavelength , is the th nutrient standard spectral response curve, and are the starting and ending wavelengths of the analysis wavelength range; Based on the nutrient correlation degree corresponding to the peak value, spectral peaks highly correlated with the target nutrient are screened, the wavelength positions and spectral intensities of the peak values are sorted out, and a spectral peak calibration result is generated.

[0037] Specifically, when extracting all spectral peak values based on the enhanced spectral data, first refer to a record containing a description of the wavelength range division, which lists several peak segments and peak shape information that may appear in the range from 450 nanometers to 2400 nanometers. Then, scan the corresponding interval segment by segment in the enhanced spectral data. Whenever the local wavelength continuously shows a relatively high intensity and the peak shape is significantly convex, record this segment as a potential peak region. To ensure the reliability of peak region identification, a peak height threshold of 0.15 is set and measured from the bottom to the peak with a discrete step size of 1 nanometer. This 0.15 is obtained by subtracting the standard deviation of 0.05 from the average peak value selected after statistically analyzing the spectra of 100 batches of corn kernels. When it is detected that the peak intensity exceeds 0.15 at two consecutive sampling points and the adjacent wavelength difference is less than 3 nanometers, it is regarded as a coherent peak region. Subsequently, the exact wavelength of the peak vertex is determined by calculating the extreme value position of the curve within this segment, and a certain interval is extended on both the left and right sides of the peak vertex to distinguish the main peak from the interference of smaller surrounding peaks. If the intensity within this interval is lower than 40% of the peak top intensity, this position is regarded as the peak shoulder and the extension stops. This 40% value is obtained by upwardly correcting the average value of 35% after multiple rounds of observation of the same type of peak shape structure. Then, calculate the central wavelength of each peak region and synchronously record the spectral intensity at this central wavelength. If there are more than two peak vertices in the same wavelength range, select the highest one as the main peak point and mark the remaining peak points as secondary peaks for comparison with the peaks in other bands during subsequent determination. After all peak regions are identified, they are sorted and associated with their wavelength ranges and intensity values, and integrated into a preliminary spectral peak value data.

[0038] The advantage of the formula is that by integrating and matching the spectral peak intensity with the standard spectral response curve within the specified wavelength interval, the similarity between the current peak value of the corn kernel and the known nutrient characteristics can be quantified as a numerical value , which has the advantages of intuitiveness and operability when determining the correlation degree of different nutrient contents.

[0039] is obtained by retrieving the spectral intensity sequence of this peak within the range from to from the preliminary spectral peak value data generated in the previous stage. During the specific operation, first obtain several discrete measurement values with a wavelength step size of 1 nanometer or a smaller interval, and then use these measurement values as ([ , sorted in the form of (wavelength, intensity). If the intensity of some discrete points is missing, it is complemented by local interpolation of adjacent points, and it is ensured that the interpolation does not produce jumps. The weight of the interpolation is determined by the wavelength spacing of this section and the actual values of adjacent points. For example, linear interpolation can be used. , after continuously measuring 100 batches of corn kernels, the intensity means in some wavelength ranges are approximately between 0.1 and 0.7, and the corresponding extreme values can reach approximately 1.0 and will not exceed this range significantly.

[0040] The acquisition steps of are as follows. For the th nutrient, check a standard spectral response library, which stores the absorption characteristic curves of each nutrient at different wavelengths in advance. Each curve is measured by a professional testing institution under laboratory conditions using high-purity samples. For the convenience of numerical operations, the original curve is sampled at equal intervals and a ( , response value) list is obtained. Taking nitrogen as an example, a response value is obtained every 1 nanometer from 1200 nanometers to 1400 nanometers and recorded as , for example, is measured at 1250 nanometers , is measured at 1251 nanometers etc. The complete curve data can span multiple bands, or it can be intercepted according to the previously determined key wavelength region and aligned with . Whenever it is necessary to calculate , the corresponding can be called from the library.

[0041] The acquisition steps of are as follows. Referring to the database and the enhanced spectral peak positioning results, the starting wavelength position where the peak is most likely to correspond to the nutrient information is screened out. Taking nitrogen as an example, based on a large number of experiments, it is found that a strong absorption characterization appears around 1200 nanometers. Therefore, is set as the starting point of the analysis section.

[0042] The acquisition steps of are as follows. Corresponding to , the termination wavelength position within the same peak segment is selected, and the sampled peak segment is completely included in . Still taking nitrogen as an example, if the detected peak shape covers around 1390 nanometers, then can be taken.

[0043] Calculation process: In the example, take nanometers, nanometers. Assume that the measurement step size of a certain batch of corn kernels in this interval is 1 nanometer, and a total of 201 sampling points are obtained, recorded as ; Selected nutrient (corresponding to nitrogen element), extract from the preliminary spectral peak data and retrieve from the standard response library , set at nanometers , nanometers , and so on for the entire interval, perform discrete integration point by point, and denote the discrete step size , then the numerator can be approximated as:

[0044] The denominator can be approximated as:

[0045] The specific numerical operation example is as follows, select four local points for illustration:

[0046]

[0047]

[0048]

[0049] The sum of these four points in the numerator part:

[0050] The sum of these four points in the denominator part:

[0051] After accumulating all 201 sampling points, the numerator term , and the denominator term are obtained, and thus:

[0052] This result shows that the correlation between the corn kernels of this batch and the standard response curve of nitrogen element in the range of 1200 nanometers to 1400 nanometers is about 0.732. When exceeds 0.70, it can be regarded as highly correlated, meaning that this peak segment is more likely to correspond to the absorption characteristics of nitrogen element. If is lower than 0.40, it indicates that the matching degree of this section with the corresponding nutrient is low, and it can be further investigated through subsequent steps.

[0053] When screening spectral peaks that are highly correlated with the target nutrient based on the nutrient correlation degree corresponding to the peak value, first retrieve the numerical list calculated in the previous step, and compare it with a pre-set correlation degree threshold, which can be determined according to statistical methods. For example, for 100 batches of corn kernels, the average correlation degree measured in the same wavelength range is 0.55 and combined with a standard deviation of 0.1. Thus, the threshold is set to 0.75 by adding twice the standard deviation of 0.2 to 0.55. When the numerical value of a certain peak is greater than or equal to 0.75, it is classified into the category of "highly correlated". If there are a few peak segments between 0.5 and 0.75, it is marked as "moderately correlated" for subsequent repeated measurements or further verification. During the process, the central wavelength and the maximum spectral intensity corresponding to this peak segment are recorded respectively. If it is found that the central wavelength is within the key nutrient section selected above and significantly higher than the correlation degrees of other nutrients, then this peak segment is preferentially identified as the characteristic peak of the target nutrient. Finally, after sorting out all the information of highly correlated peak values, a list containing the wavelength position and spectral intensity is formed to obtain the spectral peak calibration result.

[0054] The steps for obtaining the data feature set are as follows: Based on the spectral peak calibration result, retrieve the standard nutrient spectral data in the database, compare the spectral curve change trends of the two spectra within the same wavelength range, analyze the changes in the position, width, and intensity of the spectral absorption peaks, and obtain the standard nutrient spectral matching result; According to the standard nutrient spectral matching result, calculate the principal component contribution degree. The calculation formula is: Among them, represents the principal component contribution degree, represents the spectral intensity of the th wavelength in the spectral peak calibration result, represents the spectral intensity corresponding to the same wavelength in the standard nutrient spectral database, represents the spectral absorption difference corresponding to the th wavelength, represents the comparison matching error of the th wavelength, represents the total number of wavelength points; Based on the principal component contribution degree, perform principal component analysis, extract spectral feature variables, construct a feature vector space, reduce the dimension of the spectral data, remove noise and redundant variables, and generate a data feature set.

[0055] Specifically, when retrieving the standard nutrient spectral data in the database based on the spectral peak calibration results, first locate the reference number corresponding to the current corn kernel in the database, then read the sampling information of each wavelength range in the standard spectrum, including its characteristic wavelength points and the absorption intensity at each point. Then, compare the characteristic wavelength points with the spectral peak calibration results recorded previously one by one in terms of wavelength. If it is found that the wavelength difference between the two is within 3 nanometers and the difference in absorption intensity is less than 0.05, it is determined that this wavelength point has a comparable significance, and the corresponding absorption peak shape is extended for inspection, that is, continuously compare the intensity change of the original spectrum within 1 nanometer forward and backward from the wavelength point, and focus on recording the sections where the intensity continuously rises or falls. If it is found that the peak width differs from the peak width information in the database by less than 0.2 nanometers or the peak height difference is less than 0.03, it is considered that the absorption peak has the same height as the reference peak. At the same time, in order to exclude local interference, it is also necessary to compare the slope change of the peak shape within 5 nanometers above and below. If the intensity slope fluctuates significantly by more than 0.08 within this range, mark this as an abnormal peak section and temporarily suspend the comparison. Next, sort the wavelength points that have been successfully compared in sequence and view their position distribution in the spectral curve. If the wavelength difference between two adjacent effective comparison points does not exceed 2 nanometers, merge them into the same reference section, and determine whether it meets the existing standard waveform definition in the database according to the average absorption intensity and the peak position within this section. If all are satisfied, summarize them into a standardized peak section. Finally, after confirming the completion of the comparison operation for all reference peak sections, conduct an overall evaluation of the differences in the overall wavelength range, peak width, peak height, etc., to obtain the standard nutrient spectral matching result.

[0056] The benefit of the formula is that by introducing the spectral difference in the numerator part and the correction term , it can more precisely measure the difference degree between the measured spectrum and the standard spectrum at different wavelengths, and perform normalization processing on the overall intensity value in the denominator part, so as to unify various amplitudes and errors into a comparable main component contribution degree index .

[0057] : represents the spectral intensity of the th wavelength in the spectral peak calibration result. All wavelengths involved in the comparison and their intensities have been recorded during the previous spectral peak calibration, and these values can be directly called and arranged in order. Acquisition example: If the intensity measured at a wavelength of 620 nanometers is 0.35 and the intensity measured at 621 nanometers is 0.37, then , , etc. can be stored in an array in sequence for subsequent calculations.

[0058] : represents the spectral intensity of the corresponding wavelength in the standard nutrient spectrum database, which is obtained by one-to-one correspondence with the wavelength of the standard spectrum in the previous section and combined with the curve data pre-registered in the database. If the standard intensity corresponding to the same wavelength of 620 nanometers is 0.40 in the database, then , and 0.38 at 621 nm. , and so on to extract the standard values ​​of all matching wavelength points in sequence.

[0059] :Represents The spectral absorption difference corresponding to each wavelength needs to be calculated by differentially calculating the actual measured absorption intensity and the absorption characteristics in the standard spectrum curve. Example: If the actual measured absorption value at 620 nanometers is 0.35 and the standard absorption value is 0.40, the basic difference of 0.05 can be calculated first, and then the correction factor (for example, 0.1) pre-set for this band can be added to obtain .

[0060] :Represents The contrast matching error of wavelengths is used to adjust The influence of the item on the overall contribution is obtained by collecting the deviation data of the wavelength on multiple batches of different samples and performing error estimation. And the variance , then let .

[0061] : represents the total number of wavelength points, that is, the number of wavelengths involved in the actual comparison. Each wavelength point corresponds to a group If the sample is taken every 1 nanometer from 600 nanometers to 1400 nanometers and the same number of standard spectral points are paired, 801 wavelength points can be obtained, corresponding to Example: If researchers only compare the 100 key wavelength points, then .

[0062] Calculation process: by As an example, take four wavelength points ,make:

[0063]

[0064]

[0065]

[0066] Substitute these values into the numerator part of the formula:

[0067] First, calculate ;

[0068]

[0069] So, this term is approximately ; Calculate the other three terms in the same way and accumulate the results as the numerator part. For example, successively obtain equal values. Here, they are not expanded item by item. Finally, the sum of the numerator part is denoted as ; Denominator part:

[0070] Substitute the values:

[0071] Then:

[0072] This result shows that under this example, the contribution degrees of the principal components at 4 wavelength points are 0.112. If this value is on the low side, it indicates that there are relatively large differences between the measured spectrum and the standard spectrum at these wavelength points. The researcher can further check in combination with the differences and matching errors in the previous process or repeat the measurement at more wavelength points. If in actual operation is greater than 0.5, it is usually regarded as a high matching degree; if it is less than 0.2, it indicates an obvious difference degree and the source of the deviation needs to be re-compared and analyzed.

[0073] When performing principal component analysis based on the contribution degree of principal components, first refer to a pre-organized mapping table of wavelength and absorption intensity. This table lists several wavelength points that can be used for principal component extraction and the intensity values measured in different corn kernel samples. Subsequently, organize these wavelength points and intensity values in matrix form and perform matrix decomposition steps to analyze the proportion of each wavelength in the variance distribution. Wavelengths with significantly too small variance contribution are marked as redundant variables and excluded from subsequent calculations. If the contribution degree of some of these wavelength points is lower than 0.02, they are classified into the noise interval together. The value of 0.02 is obtained by statistically analyzing the measurement data of 200 batches of corn kernels, selecting the mean value of 0.015 and correcting it upward by 0.005. After decomposition, map the remaining principal component wavelengths back to the spectral intensity sequence to form a set of eigenvectors, and then perform a similarity determination on this set of eigenvectors to exclude possible multicollinearity relationships. If the similarity exceeds 0.9, it means that the two vectors are close in the data space. Here, the threshold of 0.9 is obtained by statistically analyzing the probability of highly similar scenarios in multiple measurements of the same type of spectral vectors. Finally, retain vectors with sufficiently obvious differences to cover more information dimensions, and combine these vectors into a new coordinate system for storage and summarize them into a data feature set.

[0074] The steps to obtain the classification results of nutrient status are as follows: Based on the data feature set, configure the kernel function type and kernel parameters of the support vector machine classifier to obtain the configured SVM model; According to the configured SVM model, obtain the classification result function, and the expression is: Among them, is the classification result function, is the sign function, is the Lagrange multiplier corresponding to the support vector, is the th class label of the data point, is the similarity between the kernel function calculating the th support vector and the input , is the bias term, is the number of support vectors; Based on the classification result function, classify the nutrient status of each data point, distinguish between sufficient and lacking nutrient states, evaluate the lack degree of each nutrient, and generate the classification results of nutrient status.

[0075] Specifically, when configuring the kernel function type and kernel parameters of the support vector machine classifier based on the data feature set, first refer to a record of the characteristics of common kernel functions, which details the performance of linear kernel, polynomial kernel, and radial basis kernel under different data scales, as well as their corresponding bias and soft margin setting ranges. If overfitting is likely to occur in the case of small-scale data, the linear kernel can be considered and the upper limit of the soft margin can be set to a value between 0 and 5. After experimenting with hundreds of sets of corn kernel characteristics, the most suitable value is determined to be approximately 2.0. At this time, 2.0 is obtained based on the balance point between the input dimension and the classification error rate. When checking the polynomial kernel, attention should be paid to its order setting. For example, the order can be limited between 2 and 4, and the error rate can be compared for each order. If the error rate decreases significantly as the order increases, the higher order can be retained, but it is necessary to ensure that the final classification speed is not lower than expected. Whenever the kernel function is changed or the kernel parameters are adjusted, the cross-validation process should be repeated, and the accuracy and recall rate on the validation set should be compared. If the average accuracy in 10-fold cross-validation remains in the range of 85% to 90% and the recall rate remains in the range of 80% to 85%, it is determined to be acceptable. After completing this evaluation, the final kernel function type and the corresponding kernel parameters are registered to obtain the configured SVM model.

[0076] The advantage of the formula is that it combines the Lagrange multipliers corresponding to the support vectors with their class labels and measures the similarity between the input and each support vector through the kernel function , thereby effectively discriminating samples in the high-dimensional feature space.

[0077] : Represents the weighted sum of all support vectors. This term incorporates the Lagrange multipliers , labels and the kernel function into the calculation. Before obtaining the result, the researcher needs to extract the corresponding and values for each support vector from the trained SVM model. At the same time, test samples are extracted from the spectral feature set and the kernel function is calculated with the support vectors . Then, the obtained results are multiplied and summed in order. For example, if , , , , after the kernel function calculation, , the weighted cumulative value is 。

[0078] The acquisition steps of are as follows. When training the SVM, the Lagrange multiplier value corresponding to each support vector is determined by solving the Lagrange dual problem. Generally, the objective function is first established and iteratively solved in a maximization manner. The current solution is updated in each iteration until convergence. If the final result of is greater than 0, it indicates that belongs to the support vector. Researchers will also record the convergence threshold during training and judge whether the stopping condition is met. For example, the threshold is set to , if the update amplitude of before and after iteration is less than this value, the acquisition stops. Example: In a certain training, 200 corn kernel data are iterated, and finally 15 samples correspond to , and their average value is about 0.45, up to 1.2, meeting the actual classification requirements.

[0079] The acquisition steps of are as follows. In the training samples, each piece of data corresponds to a known class label. Through manual annotation or automatic annotation in the early stage, two categories are formed. One category is defined as the nutrient - sufficient label 1, and the other is the nutrient - deficient label - 1. When the sample collection is completed, researchers conduct physical and chemical tests on the corn kernels and obtain the measured nutrient content values, and then determine the category according to a certain content threshold. If the content is higher than the threshold, it is recorded as 1, and if it is lower than the threshold, it is recorded as - 1. Example: In the range of nitrogen element concentration, with 2% as the benchmark, greater than or equal to 2% is labeled 1, and less than 2% is labeled - 1. Each sample is obtained through the same process 。

[0080] The acquisition steps of are as follows. By selecting the type of kernel function and substituting the input and the support vector into the kernel function to calculate the similarity. If the linear kernel is used, then ; if the radial basis kernel is used, then , where can be determined by grid search in the training stage. Researchers will detect the classification accuracy and generalization performance within a group of candidate values and select the better value. Example: Take Substitute it into the radial basis kernel and for the same spectral feature vector and the support vector calculate the Euclidean distance to obtain , then .

[0081] The acquisition steps of are as follows. After solving the dual problem of SVM training, the bias is obtained from the KKT conditions. Generally, after knowing the support vector and , substitute the samples that satisfy for solution. The calculation results of multiple known support vector pairs can be averaged to reduce noise. Example: When substituting a certain support vector , the solution is , and when substituting another support vector , the solution is . The average of the two can be obtained as for final prediction.

[0082] The acquisition steps of are as follows. The number of support vectors, usually the number of samples that can be counted after SVM training because only are regarded as support vectors, and the total number is . In a series of training experiments, the value of can also be changed by adjusting the penalty factor or kernel function parameters . Example: In a classification task, there are 100 samples. After training, about 10 samples meet the conditions, then .

[0083] Calculation process: In a simplified calculation, select support vectors and assign their , , the bias . Denote the input spectral vector as . The support vectors are respectively and . If the kernel function obtains after sampling, then the internal summation term:

[0084] Add the bias , and the result is . Then:

[0085] This result shows that in this example, the input sample It is classified into the category corresponding to label 1, i.e., the nutrient - sufficient category.

[0086] When classifying the nutrient status of each data point based on the classification result function, first apply the previously obtained SVM model to all spectral data points to be measured and calculate one by one and the offset The total value of their sum is calculated, and then the sign function is called to judge its positive or negative nature. If the result is positive, it is classified into the nutrient - sufficient category; if it is negative, it is classified into the nutrient - deficient category. To quantify the degree of deficiency of different nutrients, the intensity of negative values that exceed a certain threshold is additionally recorded to measure the magnitude of the deficiency. This threshold is calculated by adding 0.05 to the average deficiency determination score measured by the device for 200 batches of corn kernels. If it is detected that the negative - value intensity of a certain batch of corn kernels in the same nutrient dimension has reached this threshold, it is regarded as significantly deficient; if it is lower than the threshold, it is regarded as slightly deficient. To more comprehensively verify the classification result, the negative - value intensity is compared with the actual nutrient content of on - site chemical analysis. If the two maintain the same trend in multiple measurements, the classification accuracy is confirmed. Finally, the classification information of all kernels is summarized, including the nutrient category corresponding to each kernel and its degree of deficiency, and this summarized information is recorded as the nutrient - status classification result.

[0087] The steps to obtain the nutrient abundance - deficiency assessment result are as follows: According to the nutrient - status classification result, calculate the nutrient abundance - deficiency metric. The calculation formula is: where represents the nutrient abundance - deficiency metric, is the measured nutrient concentration, is the threshold standard of this nutrient, is the maximum response concentration of this nutrient in the standard data, is the minimum response concentration of this nutrient, is the nutrient - concentration change rate; Based on the nutrient abundance - deficiency metric, analyze the abundance - deficiency situation of each nutrient element to generate the nutrient abundance - deficiency assessment result.

[0088] Specifically, the advantage of the formula is that it comprehensively incorporates the deviation degree of the nutrient concentration from the maximum and minimum response concentrations and in the standard data into the numerator term, and combines the logarithmic function in the denominator to quantify the influence of the concentration change rate so that the abundance - deficiency degree of the current nutrient can be reflected in a unified index .

[0089] : \(q\) is the measured nutrient concentration, obtained through actual sampling and analysis. Researchers first collect a certain number of corn kernels in the field and send them to a laboratory or on-site testing instrument for precise measurement, forming multiple concentration data sequences. Then, all entries are combined and the mean or median value is extracted as the representative concentration. Example: In the detection of nitrogen, the concentrations of several samples are measured to be 2.2%, 2.3%, 2.5%, etc. 2.3% can be selected as the representative value for this , or the samples can be repeatedly measured by a higher-precision method to improve accuracy, and finally \(q = 2.3\%\) is obtained.

[0090] : \(T\) is the threshold standard for this nutrient. A certain concentration limit needs to be determined to distinguish whether the nutrient is sufficient. This value is generally obtained from agricultural science institutions or long-term experimental data statistics. Researchers establish a suitable based on the average demand of similar crops in similar growth environments. Example: When detecting nitrogen, the threshold standard is selected as 2.0%. This is the conclusion obtained from multiple observations of nitrogen concentration and plant health status in a 300-hour field experiment. After performing a correlation analysis on all concentration data and the final growth performance of the plants, 2.0% is taken as .

[0091] : \(C_{max}\) is the maximum response concentration of this nutrient in the standard data. When establishing the standard data, researchers detect the grains in multiple high-concentration environments, obtain the maximum actual detection value of the nutrient in the corn grains, and obtain a high-confidence upper limit concentration after screening out abnormal high values based on multiple batches of experiments. Example: When detecting nitrogen, the measured limit concentrations in multiple batches are approximately 3.5%. After considering some random fluctuations, 3.4% is recorded as for actual calculation. This 3.4% is the highest credible concentration obtained by combining samples from different periods and different fields.

[0092] : \(C_{min}\) is the minimum response concentration of this nutrient in the standard data, generally measured through long-term tracking of corn grains in a large range of low-input fields or barren soil environments. Researchers sample and detect situations far below the normal demand of the crop and screen out the most frequently occurring extremely low concentration values. Example: In the detection of nitrogen, the detection results of the grains collected from severely nitrogen-deficient plots multiple times are about 1.2%, 1.1%, 1.0%. Finally, is set to represent the most common extremely low value.

[0093] : is the change rate of this nutrient concentration, which is used to quantify the amplitude of concentration increase or decrease at different observation points or different time periods. Researchers calculate a relative change rate by recording the concentration changes of the same batch of corn kernels in multiple detections. Example: 20 samples are collected, and the concentrations are obtained in the first and second measurements respectively , and can be calculated in the form of difference ratios such as . After dozens of repeated observations, take the average value or a more suitable comprehensive index. If is greater than 0.2, it means a 20% increase compared to before. If it is -0.1, it means a 10% decrease compared to before.

[0094] Calculation process: The following is an example of numerical values Let , the concentration change rate , replace all numerical values in the same form (such as decimal expression): .

[0095] First calculate the numerator part :

[0096]

[0097]

[0098] Then calculate the denominator :

[0099]

[0100]

[0101]

[0102] Substitute it into the formula:

[0103] This result shows that the measure of nutrient abundance or deficiency at this time . If it is considered that under the same set of standards, that is, there are obvious differences in abundance or deficiency, the nutrient status of the kernels can be paid special attention to; if is close to 0.5 or lower, it means that the overall concentration deviation is not large. Researchers can judge whether to add fertilizers or carry out other nutrient supplements based on this in actual planting management.

[0104] When analyzing the abundance and deficiency of each nutrient element based on the nutrient abundance and deficiency measurement, first cite a record of the determination instructions for the concentration ranges of common elements, which sets several reference ranges for nitrogen, phosphorus, potassium and trace elements. The ranges list the corresponding actual concentrations and growth performances in sequence according to the ranges of 0.5 to 1.0, 1.0 to 2.0, 2.0 to 3.0, etc. When the value exceeds 2.0, it is listed as a type that requires urgent attention and query how much higher or lower its concentration is than the reference value. If more than half of the samples in 30 samples meet the urgent attention condition, an additional investigation is conducted on the entire batch of samples. During this period, the concentration change process is checked from the earliest to the latest according to the measurement date of each sample, and it is compared with the on-site environmental information, including the soil organic matter content, recent fertilizer application records, etc. If within two consecutive sampling periods, it always remains in the range of 2.0 to 3.0, it is recorded that this batch of samples is very likely to be in an unbalanced state. The same method of comparison is also carried out for other elements. For the phosphorus element, the range of 2.5 to 3.5 can be regarded as a high-risk area, and the phosphorus concentration recorded previously is also matched item by item. The increase and decrease trends of the corresponding values are marked in the test results of each record, and then compared with the pre-established observation dimensions. For example, in the case of nitrogen, the range of 0.6 to 1.5 is regarded as mild deficiency, and in the case of the corresponding phosphorus element, the range of 1.0 to 1.8 is regarded as mild deficiency. If there are obvious deviations in multiple elements at the same time, they are included in the subsequent analysis objects. Finally, all the comparison results are summarized to form the nutrient abundance and deficiency assessment results.

[0105] The above is only the preferred embodiment of the present invention, and it does not limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes and apply them to other fields. However, as long as it does not depart from the technical solution content of the present invention, any simple modification, equivalent change and modification made to the above embodiments based on the technical essence of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. A method for assessing the nutrient abundance and deficiency status of corn kernels based on spectral analysis, characterized in that: The following steps are involved: Using near-infrared and mid-infrared light sources to illuminate corn seeds, collecting reflected spectral signals, and recording spectral signal data; Based on the spectral signal data, extract key wavelength region data to obtain key spectral data; Based on the key spectral data, applying wavelet transform to analyze characteristic peaks in the data, enhancing the identifiability of the spectral peaks, and generating enhanced spectral data; Based on the enhanced spectral data, calibrate the spectral peaks associated with the nutrients, record the position and intensity of each peak, and obtain a spectral peak calibration result; Based on the spectral peak calibration result, a comparative analysis is performed with the standard nutrient spectrum in the database to extract the characteristics of the data and generate a data feature set; Based on the data feature set, a support vector machine is used to classify the nutrient status, distinguish each nutrient deficiency state, and obtain a nutrient status classification result; Based on the nutrient status classification results, a quantitative assessment is performed, threshold standards for different nutrients are set, the abundance and deficiency of each nutrient element is analyzed, and the deficiency status of nutrients below the threshold standard is described to obtain nutrient abundance and deficiency assessment results.

2. The method for assessing the nutrient abundance and deficiency status of corn kernels based on spectral analysis according to claim 1, characterized in that: The steps for acquiring the spectral signal data are as follows: Setting near-infrared and mid-infrared light sources to illuminate corn kernels, capturing spectral signals reflected from the corn kernels, converting the spectral signals into digital formats, and obtaining complete spectral signal data; The complete spectrum signal data is recorded and stored to obtain spectrum signal data.

3. The method for assessing the nutrient abundance and deficiency status of corn kernels based on spectral analysis according to claim 1, characterized in that: The steps for obtaining the key spectral data are as follows: Screening wavelength regions that are sensitive to nutrient evaluation of corn kernels from the spectral signal data, determining key wavelengths that affect nutrient analysis by comparing the absorption intensity and peak shape characteristics of each wavelength, and obtaining preliminary screened wavelength region data; Based on the initially screened wavelength region data, non-critical wavelength points are eliminated to obtain critical wavelength region data; Based on the key wavelength region data, spectral feature extraction is performed, including calculating the peak position, peak width and integrated peak area to obtain key spectral data.

4. The method for assessing the nutrient abundance and deficiency status of corn kernels based on spectral analysis according to claim 1, characterized in that: The steps for acquiring the enhanced spectral data are: Based on the key spectral data, the characteristic peaks are analyzed by wavelet transform to obtain the wavelet transform result, and the formula is: in, Represents the frequency The transformed peak intensity at For in time and frequency The wavelet coefficients of the point, For time The spectral intensity, Represents the total number of time points; Based on the wavelet transform result, the spectrum peak is identified through peak value comparison and intensity analysis to obtain enhanced spectrum data.

5. The method for assessing the nutrient abundance and deficiency status of corn kernels based on spectral analysis according to claim 1, characterized in that: The steps for obtaining the spectrum peak calibration result are: Based on the enhanced spectral data, all spectral peaks are extracted, local extreme points of the spectral curve within different wavelength ranges are analyzed, and the wavelength and spectral intensity are recorded to generate preliminary spectral peak data; According to the preliminary spectral peak data, the nutrient correlation corresponding to the peak is calculated using the following formula: in, The spectral peak and The correlation between nutrients, The wavelength The spectral intensity at For the Standard spectral response curves for various nutrients. and To analyze the starting and ending wavelengths of the wavelength range; Based on the nutrient correlation corresponding to the peak, the spectral peaks that are highly correlated with the target nutrients are screened, the wavelength position and spectral intensity of the peaks are sorted out, and the spectral peak calibration results are generated.

6. The method for assessing the nutrient abundance and deficiency status of corn kernels based on spectral analysis according to claim 1, characterized in that: The steps of obtaining the data feature set are: Based on the spectral peak calibration result, the standard nutrient spectrum data in the database is retrieved, the spectral curve change trends of the two groups of spectra in the same wavelength range are compared, the position, width and intensity changes of the spectral absorption peak are analyzed, and the standard nutrient spectrum matching result is obtained; According to the standard nutrient spectrum matching results, the main component contribution is calculated using the following formula: in, represents the principal component contribution, Represents the peak value of the spectrum. The spectral intensity of the wavelength, Represents the spectral intensity of the corresponding wavelength in the standard nutrient spectrum database, Representative The spectral absorption difference corresponding to the wavelength is Representative The contrast matching error of wavelengths, Represents the total number of wavelength points; Based on the principal component contribution, principal component analysis is performed to extract spectral characteristic variables, construct a characteristic vector space, reduce the dimension of the spectral data, remove noise and redundant variables, and generate a data feature set.

7. The method for assessing the nutrient abundance and deficiency status of corn kernels based on spectral analysis according to claim 1, characterized in that: The steps for obtaining the nutrient status classification result are: Based on the data feature set, configuring the kernel function type and kernel parameters of the support vector machine classifier to obtain a configured SVM model; According to the configured SVM model, the classification result function is obtained, and the expression is: in, is the classification result function, is the symbolic function, is the Lagrange multiplier corresponding to the support vector, For the The class labels of the data points, Calculate the kernel function Support vectors and input The similarity between is the bias term, is the number of support vectors; Based on the classification result function, the nutrient status of each data point is classified, the nutrient sufficiency and nutrient deficiency status are distinguished, the deficiency degree of each nutrient is evaluated, and the nutrient status classification result is generated.

8. The method for assessing the nutrient abundance and deficiency status of corn kernels based on spectral analysis according to claim 1, characterized in that: The steps for obtaining the nutrient abundance and deficiency assessment results are as follows: According to the nutrient status classification results, the nutrient abundance and deficiency metrics are calculated using the following formula: in, Indicates the nutrient abundance and deficiency. is the measured nutrient concentration, is the threshold standard for this nutrient, is the maximum response concentration of the nutrient in the standard data, is the minimum response concentration of the nutrient, is the rate of change of nutrient concentration; Based on the nutrient abundance and deficiency measurement, the abundance and deficiency of each nutrient element is analyzed to generate a nutrient abundance and deficiency assessment result.

Images