A rapid detection method for the mildew ratio of corn based on spectral perspective technology

Abstract

The present application discloses a rapid detection method for the mildew ratio of corn based on spectral perspective technology, comprising the following steps: collecting the spectra of mildewed corn samples; performing smoothing and denoising processing on the collected spectral data; resampling the obtained series of perspective curves by means of the discrete wavelet algorithm to obtain lower perspective curves with spectral resolutions of 2nm, 4nm, 8nm, 16nm, 32nm, and 64nm, and extracting sensitive features within the relative perspective rate, absolute perspective rate, and other perspective index curves in combination with the correlation analysis algorithm; constructing a mildew ratio estimation model for corn based on the sensitive features by using the partial least squares algorithm, and evaluating the detection accuracy by using evaluation indexes; adopting the estimation model constructed based on TC and TR, and taking the average value of the estimation results of the two as the final diagnosis result.

Description

Technical Field

[0001] The present invention relates to the field of food security, and particularly to a rapid detection method for the mildew ratio of corn based on spectral perspective technology. Background Art

[0002] As one of the important food sources in our life, corn is also an important component of artificial feed for animal husbandry and industrial processing. Corn is susceptible to mildew during transportation and storage due to environmental influences. Consuming related foods produced from mildewed corn seriously endangers the health of humans and livestock and also affects industrial processing. Therefore, there is a strong practical need to quickly and accurately identify the mildew ratio of corn for the processing of related foods.

[0003] To ensure the quality of food production, the state has formulated a series of relevant standards, laws, and regulations, which are strictly implemented by the food quality supervision department. Food enterprises are in urgent need of detection technologies for raw materials and food processing processes to meet the food production requirements. Currently, the monitoring methods for the mildew ratio of corn mainly include two types: traditional chemical analysis methods and spectral detection methods. Although traditional chemical analysis methods have high-precision detection, they have deficiencies such as long time consumption and high costs and cannot meet the application requirements of food enterprises. The spectral detection method mainly relies on the surface reflection spectrum of corn kernels, but the detection accuracy of this reflection spectrum for the mildew ratio of corn with a relatively low mildew ratio is low, and the production cost of the instrument is high, unable to meet the actual needs of related enterprises. Up to now, although there have been many related studies, the related technical methods cannot meet the batch production requirements due to low universality, high cost, and low accuracy. Summary of the Invention

[0004] The technical problem to be solved by the present invention is to provide a rapid detection method for the mildew ratio of corn based on spectral perspective technology in view of the above deficiencies. Starting from the perspective of the transparency of mildewed corn, the corn perspective energy is obtained by means of hyperspectral technology, and the relative perspective rate, absolute perspective rate, and perspective index are obtained by using the developed perspective rate calculation method. Then, the indication effects of each index are quantitatively analyzed by using the correlation analysis algorithm, and the detection accuracy of each perspective index is evaluated by using the evaluation index, which can realize rapid, accurate, and non-destructive detection of the mildew ratio of corn and can achieve high-precision detection of the mildew ratio of corn.

[0005] To solve the above technical problems, the present invention adopts the following technical solutions:

[0006] A rapid detection method for the mildew ratio of corn based on spectral perspective technology, comprising the following steps:

[0007] Step 1, collect the spectrum of the mildewed corn sample;

[0008] Step 2: Smooth and denoise the spectral data collected in Step 1, and use the transmittance calculation method to obtain the relative transmittance curve of the corn sample denoted as TC(x)i, the absolute transmittance curve denoted as TA(x)i, and the transmittance ratio curve denoted as TR(x)i, which are characteristic parameters for detecting the mildew ratio of corn;

[0009] Step 3: Resample the series of transmittance curves obtained in Step 2 by means of the discrete wavelet algorithm to obtain lower transmittance curves with spectral resolutions of 2nm, 4nm, 8nm, 16nm, 32nm, and 64nm, and extract sensitive features within the relative transmittance, absolute transmittance, and other transmittance index curves in combination with the correlation analysis algorithm;

[0010] Step 4: Based on the sensitive features obtained in Step 3, construct a mildew ratio estimation model for corn using the partial least squares algorithm, that is, input the characteristic parameters and the measured results of the corn mildew ratio into the partial least squares regression algorithm simultaneously to obtain the parameters required for model construction, and evaluate the detection accuracy using evaluation indicators;

[0011] Step 5: Adopt the estimation model constructed based on TC and TR in Step 4, and take the average of the estimation results of the two as the final diagnosis result.

[0012] Further, the specific process of collecting the spectrum of the mildewed corn sample in Step 1 is as follows:

[0013] Lay the i - type corn containing mildewed corn in a single layer on a transparent and colorless glass, then place a white board under the glass, collect the corn spectrum using a ground object spectrometer, and take the average value of the collected spectrum as the final result, denoted as f(x); i白 Then remove the white board and place a black board under the glass, and again collect the corn spectrum using a ground object spectrometer, and take the average value of the collected spectrum as the final result, denoted as f(x), i黑 Repeat this process to collect the spectra of 6 types of corn samples in total;

[0014] Select a type of corn sample and lay it in a single layer, then use a ground object spectrometer to collect the spectrum of the laid - out corn successively from the right side to the left side until the collection of the spectra of all corn samples is completed, and then collect the spectra of the corn samples from the left side to the right side in turn. The above steps can be repeated to collect the spectra of the corn samples, and repeat the above steps at least 1 time. Then, take the average value of the spectral data collected when the white board or black board is placed under the glass as the final spectrum of the corn sample with the white board or black board as the background. The spectra of 6 types of corn samples are laid out and the spectra are collected in turn to complete the collection of all corn spectra. Among them, the calculation formula for the corn mildew ratio is:

[0015]

[0016] In the formula, GM is the weight of the mildewed corn, G H is the weight of the non-mildewed corn.

[0017] Further, the second step specifically includes the following steps:

[0018] Step 2.1, Spectral smoothing and denoising processing

[0019] For the acquisition of spectral data, a low-pass filter is used to smooth the spectrum. The specific calculation formula is as follows:

[0020] Assume that the spectrum of the corn sample is a function f(x) of the wavelength, and f s (x) is the smoothed spectrum, and i is the wavelength, then:

[0021]

[0022] Further, the second step specifically further includes the following steps:

[0023] Step 2.2, Perspective calculation

[0024] The blackboard has a strong absorption effect on light, almost full absorption. Assume its reflectivity is 0 and the perspective is 0; the whiteboard has a strong reflection effect on light, almost full reflection. Assume its reflectivity is 1 and the perspective is 0; therefore, the spectrum of the corn sample collected based on the blackboard background is the sum of reflection R and volume scattering S; the spectrum of the corn sample collected based on the whiteboard background is the sum of reflection R, volume scattering S, secondary perspective T, and secondary volume scattering S caused by perspective, denoted as RT. Assume the input energy is E, and the energy received by the optical signal sensor is E 传 :

[0025] E = R + T + A + S Formula 3;

[0026] When the background is the blackboard:

[0027] E 传_黑 = R + S Formula 4;

[0028] When the background is the whiteboard:

[0029] E 传_白 = R + S + T^2 + S*T Formula 5;

[0030] Perspective rate inference:

[0031] E 传_白 -E 传_黑 = T*(T + S) Formula 6;

[0032] Assume α = S / T, then S = T*α;

[0033] Then E 传_白 -E传_黑 = T^2 * α;

[0034]

[0035] Furthermore, the specific steps of Step 2 further include the following steps:

[0036] Step 2.3, Calculation of the perspective curve:

[0037]

[0038]

[0039]

[0040] In the formula, f(x) i白 is the spectrum of the i-th type of corn sample collected based on the whiteboard background, and f(x) i黑 is the spectrum of the i-th type of corn sample collected based on the blackboard background.

[0041] Furthermore, the specific steps of Step 3 include the following steps:

[0042] Step 3.1, Discrete wavelet algorithm

[0043] Adopt the low-frequency information of the discrete wavelet. The formula of the discrete wavelet algorithm is as follows:

[0044]

[0045] In the formula, the parameter a represents the scale coefficient, which is the reciprocal of the frequency; the parameter b represents the time shift (or translation), λ is the wavelength, is the mother wavelet.

[0046] Furthermore, the specific steps of Step 3 further include the following steps:

[0047] The calculation formula of the correlation analysis algorithm in Step 3.2 is as follows:

[0048]

[0049] In the formula, r j is the correlation coefficient, X i is the perspective rate of the i-th sample at the j-th band, Y i is the mildew ratio of the sample, n is the total number of samples; X A is the average value of the perspective rates of all samples at the j-th band, and Y A is the average value of the mildew rates of all samples at the j-th band.

[0050] Furthermore, in Step 4, the evaluation indicators are selected as the coefficient of determination and the root mean square error as the evaluation indicators, and their calculation formulas are as follows:

[0051]

[0052]

[0053] Wherein, YM i is the measured mildew ratio of corn, and YMP i is the predicted value based on the mildew ratio estimation model of corn, YM A is the average value of the measured soil organic matter content;

[0054] The coefficient of determination can be used to show the correlation between the estimation result and the measured result. The higher the correlation, the higher the credibility; the root mean square error can be used to represent the offset between the estimation result and the measured result.

[0055] Further, the calculation formula of the final diagnosis result in step five is as follows:

[0056]

[0057] Wherein, BL TC is the prediction result of the estimation model constructed based on TC, and BL TR is the prediction result of the estimation model constructed based on TR; MBL finai is the final result of the result to be detected; Δ is the correction coefficient.

[0058] The present invention adopts the above technical solutions, and compared with the prior art, has the following technical effects:

[0059] The present invention takes hyperspectral technology as the main means, can realize accurate, real-time and non-destructive detection of the mildew ratio of corn, can provide basic technical support for the detection of the mildew ratio of corn by food-related enterprises, and the detection accuracy of the present invention is higher, and the robustness and universality are better, which can meet the requirements of the detection accuracy of the mildew ratio of corn. BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for the description of the specific embodiments or the prior art. In all the drawings, similar elements or parts are generally identified by similar reference numerals. In the drawings, the elements or parts do not necessarily draw according to the actual scale.

[0061] Figure 1 is the flowchart of the method of the present invention;

[0062] Figure 2 is the schematic diagram of the perspective effect of normal corn and mildewed corn in the present invention;

[0063] Figure 3The background in the present invention is the spectral curve diagram of a blackboard and a whiteboard;

[0064] Figure 4 The perspective curve diagrams in the present invention;

[0065] Figure 5 The perspective curve correlation effect diagram in the present invention;

[0066] Figure 6 The corn mildew detection model constructed based on TA, TC, and TR in the present invention. Detailed implementation manners

[0067] Example 1, as Figure 1 shown, a rapid detection method for the mildew ratio of corn based on spectral perspective technology includes the following steps:

[0068] Step 1, spread the first type of corn containing mildewed corn in a single layer on a transparent and colorless glass, then place a whiteboard under the glass, use a ground object spectrometer to collect the corn spectrum, and take the average value of the collected spectrum as the final result, denoted as f(x) i白 ; then remove the whiteboard and place the blackboard under the glass, and again use a ground object spectrometer to collect the corn spectrum, and take the average value of the collected spectrum as the final result, denoted as f(x) i黑 , and the spectral curve is as Figure 3 shown. Repeat this process to collect the spectra of 6 types of corn samples in total.

[0069] As Figure 2 shown, the mildew ratios of the 6 types of corn samples are 0, 0.01, 0.02, 0.03, 0.04, and 0.05 in sequence; select one type of corn sample and spread it in a single layer (spread the corn in a rectangle, with a width of 5 cm and the length adjustable according to the number or weight of the samples, and the superposition of corn kernels is prohibited), then use a ground object spectrometer to collect the spectra of the spread corn successively from the right side to the left side of the spread corn until the collection of all the corn sample spectra is completed, and then collect the corn sample spectra from the left side to the right side of the spread corn in sequence. The above steps for collecting the corn sample spectra can be repeated, and at least repeat the above steps once. Then, take the average value of the spectral data collected when the whiteboard or the blackboard is placed under the glass as the final spectrum of the corn sample when the whiteboard or the blackboard is the background. The 6 types of corn samples are spread and the spectra are collected in sequence to complete the collection of all the corn spectra. Among them, the calculation formula for the corn mildew ratio is:

[0070]

[0071] In the formula, G M is the weight of the mildewed corn, and G H is the weight of the non-mildewed corn.

[0072] The glass used to place the corn samples should be transparent, colorless, and 1 mm thick. The width of the whiteboard background or blackboard background should be greater than 8 cm. When measuring the sample spectrum, the whiteboard or blackboard should be placed closely against the glass. The reflectivity of the whiteboard in the visible-near infrared region should be 1, and the reflectivity of the blackboard in the visible-near infrared region should be 0.

[0073] Step 2: Smooth and denoise the spectral data collected in Step 1, and use the transmittance calculation method to obtain the relative transmittance curves of 6 types of corn as Figure 4 shown, denoted as TC(x) i , the absolute transmittance curve denoted as TA(x) i and the transmittance ratio curve denoted as TR(x) i , which are characteristic parameters for detecting the mildew ratio of corn, specifically including the following steps:

[0074] 2.1. Spectral smoothing and denoising

[0075] The spectral data is collected and smoothed using a low-pass filter to remove the influence of noise information in the spectrum and improve the signal-to-noise ratio of the spectral data. The coefficients of the low-pass filter are as follows:

[0076] SM = [0.0800, 0.2147, 0.5400, 0.8653, 1.0000, 0.8653, 0.5400, 0.2147, 0.0800].

[0077] The specific calculation formula is as follows:

[0078] Let the spectrum of the corn sample be a function of wavelength f(x), and f s (x) be the smoothed spectrum, and i be the wavelength, then:

[0079]

[0080] 2.2. Transmittance calculation principle

[0081] The blackboard has a strong absorption effect on light, almost full absorption. Let its reflectivity be 0 and its transmittance be 0; the whiteboard has a strong reflection effect on light, almost full reflection. Let its reflectivity be 1 and its transmittance be 0. Therefore, the spectrum of the corn sample collected based on the blackboard background is the sum of reflection R and body scattering S; the spectrum of the corn sample collected based on the whiteboard background is the sum of reflection R, body scattering S, secondary transmittance T, and secondary body scattering S caused by transmittance, denoted as RT. Let the input energy be E, and the energy received by the optical signal sensor be E 传 :

[0082] E = R + T + A + S Formula 3.

[0083] When the background is a blackboard:

[0084] E 传_黑 = R + S, Equation 4.

[0085] When the background is a whiteboard:

[0086] E 传_白 = R + S + T^2 + S*T, Equation 5.

[0087] Transmittance inference:

[0088] E 传_白 -E 传_黑 = T*(T + S), Equation 6.

[0089] Let α = S / T, then S = T*α;

[0090] Then E 传_白 -E 传_黑 = T^2*α;

[0091]

[0092] 2.3. Calculation formula of the perspective curve:

[0093]

[0094]

[0095]

[0096] In the formula, f(x) i白 is the spectrum of the i-th type of corn sample collected based on the whiteboard background, and f(x) i黑 is the spectrum of the i-th type of corn sample collected based on the blackboard background.

[0097] Step 3: Resample the series of perspective curves obtained in Step 2 with the discrete wavelet algorithm. The discrete wavelet algorithm can make the resampling effect of the spectrum better, and obtain perspective curves with lower spectral resolutions of 2nm, 4nm, 8nm, 16nm, 32nm, and 64nm. Then, combined with the correlation analysis algorithm, extract the sensitive features in the relative transmittance, absolute transmittance, and other transmittance index curves. The sensitive features refer to the characteristic bands that are sensitive to the mildew ratio of corn. The correlation effect diagram of the perspective curve is as Figure 5 shown, and specifically includes the following steps:

[0098] 3.1. Discrete wavelet algorithm

[0099] The discrete wavelet algorithm is a new technology for signal processing, with multi-scale analysis and singularity detection functions. It can decompose a signal into a series of high-frequency and low-frequency signals using low-pass and high-pass filters. Among them, the spectral resolution of the low-frequency information decreases exponentially with the increase in the number of decompositions. This invention mainly uses the low-frequency information of discrete wavelets. The formula of the discrete wavelet algorithm is as follows:

[0100]

[0101] In the formula, the parameter a represents the scale coefficient, which is the reciprocal of the frequency; the parameter b represents the time shift (or translation), λ is the wavelength, is the mother wavelet.

[0102] 3.2. The calculation formula of the correlation analysis algorithm is as follows:

[0103]

[0104] In the formula, r j is the correlation coefficient, X i is the transmittance of the i-th sample in the j-th band, Y i is the mildew ratio of the sample, n is the total number of samples; X A is the average value of the transmittances of all samples in the j-th band, Y A is the average value of the mildew rates of all samples in the j-th band.

[0105] The correlation coefficient r j varies in the range of [-1, 1]. The larger its absolute value, the stronger the correlation. The correlation of characteristic parameters can be used to extract sensitive features.

[0106] Step 4: Based on the sensitive features obtained in Step 3, use the partial least squares algorithm to construct a corn mildew ratio estimation model, that is, input the characteristic parameters and the measured results of the corn mildew ratio into the partial least squares regression algorithm at the same time to obtain the parameters required for model construction, and use evaluation indicators to evaluate the detection accuracy.

[0107] Among them, the determination coefficient and the root mean square error are selected as evaluation indicators. Their calculation formulas are as follows:

[0108]

[0109]

[0110] In the formula, YM i is the measured corn mildew ratio, YMP i is the predicted value based on the corn mildew ratio estimation model, YM A is the mean value of the measured soil organic matter content.

[0111] The coefficient of determination can be used to show the correlation between the estimated result and the measured result. The higher the correlation, the higher the credibility. The root mean square error can be used to represent the deviation between the estimated result and the measured result.

[0112] Step Five: Adopt the estimation model constructed based on TC and TR in Step Four, as Figure 6 shown, and take the average of the estimation results of the two as the final diagnosis result. The calculation formula is as follows:

[0113]

[0114] In the formula, BL TC is the prediction result of the estimation model constructed based on TC, and BL TR is the prediction result of the estimation model constructed based on TR; MBL finai is the final result of the result to be detected; Δ is the correction coefficient, which needs to be adjusted according to the corn variety and origin, and the value range is [0.525, 2.86].

[0115] The description of the present invention is given for purposes of illustration and description, and is not intended to be exhaustive or to limit the invention to the disclosed form. Many modifications and variations are obvious to those of ordinary skill in the art. The embodiments are chosen and described in order to best explain the principles of the invention and its practical application, and to enable those of ordinary skill in the art to understand the invention and design various embodiments with various modifications suitable for a particular purpose.

Claims

1. A rapid detection method for the mildew ratio of corn based on spectral perspective technology, characterized in that: It includes the following steps: Step 1: Collect the spectra of mildewed corn samples; Step 2: Perform smoothing and denoising processing on the spectral data collected in Step 1, and use the transmittance calculation method to obtain the relative transmittance curve of the corn sample denoted as TC(x)i, the absolute transmittance curve denoted as TA(x)i, and the transmittance ratio curve denoted as TR(x)i, which are characteristic parameters for detecting the mildew ratio of corn; Step 3: Resample the series of transmittance curves obtained in Step 2 by means of the discrete wavelet algorithm to obtain lower transmittance curves with spectral resolutions of 2nm, 4nm, 8nm, 16nm, 32nm, and 64nm, and extract sensitive features within the relative transmittance, absolute transmittance, and other transmittance index curves in combination with the correlation analysis algorithm; Step 4: Construct an estimation model for the mildew ratio of corn based on the sensitive features obtained in Step 3 using the partial least squares algorithm, and evaluate the detection accuracy using evaluation indicators; Step 5: Adopt the estimation model constructed based on TC and TR in Step 4, and take the average of the estimation results of the two as the final diagnosis result; In Step 4, the determination coefficient and root mean square error are selected as the evaluation indicators, and their calculation formulas are as follows: Formula 12; Formula 13; In the formula, YMi is the measured mildew ratio of corn, YMPi is the predicted value based on the estimation model of the mildew ratio of corn, and YMA is the average value of the measured soil organic matter content; The determination coefficient is used to show the correlation between the estimation result and the measured result, and the higher the correlation, the higher the credibility; the root mean square error is used to represent the offset between the estimation result and the measured result; Step 2 specifically includes the following steps: Step 2.1: Smoothing and denoising processing of the spectrum The collection of spectral data uses a low-pass filter to smooth the spectrum, and the specific calculation formula is as follows: Let the spectrum of the corn sample be a function of wavelength f(x), fs(x) be the smoothed spectrum, and i be the wavelength, then: Formula 2; Step 2 specifically further includes the following steps: Step 2.2: Transmittance calculation The blackboard has a strong absorption effect on light and is almost completely absorbed, assuming its reflectivity is 0 and its transmittance is 0; the whiteboard has a strong reflection effect on light and is almost completely reflected, assuming its reflectivity is 1 and its transmittance is 0; therefore, the spectrum of the corn sample collected based on the blackboard background is the sum of reflection R and volume scattering S; the spectrum of the corn sample collected based on the whiteboard background is the sum of reflection R, volume scattering S, secondary transmittance T, and secondary volume scattering S caused by transmittance, denoted as RT. Let the input energy be E and the energy received by the optical signal sensor be E_trans: E = R + T + A + S Formula 3; When the background is the blackboard: E_trans_black = R + S Formula 4; When the background is the whiteboard: E_trans_white = R + S + T^2 + S*T Formula 5; Transmittance reasoning: E_trans_white - E_trans_black = T * (T + S) Formula 6; Let α = S / T, then S = T * α; Then E_trans_white - E_trans_black = T^2 * α; Formula 7; Step 2.3: Calculation of the transmittance curve: Formula 8; Formula 8; Formula 9; Within the formula, is the spectrum of the i-th type of corn sample collected based on the whiteboard background, is the spectrum of the i-th type of corn sample collected based on the blackboard background.

2. The rapid detection method for the mildew ratio of corn based on spectral perspective technology according to claim 1, characterized in that: The specific process of collecting the spectra of mildewed corn samples in Step 1 is as follows: Spread the Class I corn containing mildewed corn in a single layer on a transparent and colorless glass plate, then place a whiteboard under the glass plate, use a ground object spectrometer to collect the corn spectrum, and take the average value of the collected spectrum as the final result, denoted as f(x). i白 ; Then remove the whiteboard and place a blackboard under the glass plate. Again, use a ground object spectrometer to collect the corn spectrum, and take the average value of the collected spectrum as the final result, denoted as f(x). i黑 , and repeat this process to collect the spectra of 6 types of corn samples in total. Select a type of corn sample and lay it flat in a single layer. Then, use a ground object spectrometer to sequentially collect the spectra of the flat corn from the right side to the left side until the spectra of all corn samples are collected. Then, collect the spectra of the corn samples from the left side to the right side in sequence. Repeat the above steps to collect the spectra of the corn samples, and repeat the above steps at least 1 time. Then, take the average value of the spectral data collected when the whiteboard or blackboard is placed under the glass as the final spectrum of the corn samples with the whiteboard or blackboard as the background. The 6 types of corn samples are laid flat and their spectra are collected in sequence to complete the collection of all corn spectra. Among them, the calculation formula for the corn mildew ratio is: Formula 1; Where G M is the weight of the mildewed corn, and G H is the weight of the non-mildewed corn.

3. A rapid detection method for the mildew ratio of corn based on spectral perspective technology according to claim 1, characterized in that: Step 3 specifically includes the following steps: Step 3.1, Discrete wavelet algorithm Adopt the low-frequency information of the discrete wavelet. The formula of the discrete wavelet algorithm is specifically as follows: Formula 10; Wherein, parameter a represents the scale coefficient, which is the reciprocal of the frequency; parameter b represents the time shift or translation, is the wavelength, and is the mother wavelet.

4. A rapid detection method for the mildew ratio of corn based on spectral perspective technology according to claim 1, characterized in that: Step 3 specifically further includes the following steps: The calculation formula of the correlation analysis algorithm in Step 3.2 is as follows: Formula 11; Wherein, is the correlation coefficient, X i is the transmittance of the i-th sample in the j band, Y i is the mildew ratio of the sample, and n is the total number of samples; is the average value of the transmittances of all samples in the j band, is the average value of the mildew rates of all samples in the j band.

5. A rapid detection method for the mildew ratio of corn based on spectral perspective technology according to claim 1, characterized in that: In Step 4, the determination coefficient and root mean square error are selected as the evaluation indicators. Their calculation formulas are as follows: Formula 12; Formula 13; In the formula, YMi is the measured corn mildew ratio, YMPi is the predicted value based on the corn mildew ratio estimation model, and YMA is the average value of the measured soil organic matter content; The determination coefficient can be used to show the correlation between the estimation result and the measured result. The higher the correlation, the higher the credibility; the root mean square error can be used to represent the offset between the estimation result and the measured result.

6. A rapid detection method for the mildew ratio of corn based on spectral perspective technology according to claim 1, characterized in that: The calculation formula for the final diagnosis result in Step 5 is as follows: Formula 14; In the formula, is the prediction result of the estimation model constructed based on TC, is the prediction result of the estimation model constructed based on TR; is the final result of the result to be detected; is the correction coefficient.

Images