Tobacco leaf maturity identification method and system based on chromatic value
By using colorimeter measurements and model analysis, the problem of subjective bias in traditional manual judgment of tobacco leaf maturity has been solved, enabling accurate identification of tobacco leaf maturity and improving the overall efficiency of tobacco production.
Patent Information
- Application Number
- CN202510944446.8
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-07-09
- Publication Date
- 2025-10-17
AI Technical Summary
The traditional manual method of judging tobacco leaf maturity is subject to subjective bias and external environmental influences, resulting in inaccurate maturity judgment and affecting the quality of tobacco leaf harvesting and processing.
The colorimetric values of tobacco leaves were measured using a colorimeter. A functional relationship between tobacco leaf maturity and maturity was established through Fisher discriminant analysis and RF random forest model. A maturity discrimination model was constructed, and accurate identification was achieved by combining saturation and total color difference.
It improves the accuracy and consistency of tobacco leaf maturity identification, increases identification efficiency, provides precise guidance for tobacco leaf harvesting, and optimizes tobacco leaf quality and economic benefits.
Smart Images
Figure CN120801210A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of tobacco detection, in particular to a method and system for identifying the maturity of tobacco leaves. BACKGROUND
[0002] Tobacco leaf maturity is a key factor in determining the quality and availability of tobacco leaves, and its accurate judgment is of great significance to tobacco planting, harvesting and subsequent processing. In the long-term practice of the tobacco industry, the traditional method of judging tobacco leaf maturity mainly relies on human experience. Tobacco farmers or technical personnel observe the color, leaf shape, hair condition and vein characteristics of tobacco leaves, and determine the maturity of tobacco leaves according to their own accumulated experience. However, this subjective experience-based judgment method has defects. Due to differences in experience level and cognition, different personnel have different standards for judging the maturity of tobacco leaves, resulting in poor accuracy and consistency of the judgment results. In addition, the judgment process is also affected by different weather and different light. Influenced by subjective factors and external factors, during the tobacco harvesting process, tobacco leaves of various maturity levels are mixed together, resulting in unstable tobacco baking quality in the later stage, and ultimately affecting the processing and production of cigarettes.
[0003] With the wide application and in-depth development of colorimetry in various fields, its theory and technology have gradually penetrated into the research of the tobacco industry. Previous studies have shown that the maturity of tobacco leaves is closely related to the internal chemical composition. During the growth, maturity and conditioning of tobacco leaves, the internal pigment content and chemical composition ratio will change dynamically, which is directly reflected in the color of the tobacco leaves. The colorimeter, as a professional instrument for accurately measuring the color parameters of objects, can obtain the color values of tobacco leaves, including L* (lightness value, reflecting the change from black to white), a* (redness value, reflecting the change from green to red), and b* (yellowness value, reflecting the difference from blue to yellow). Further research has found that these color values are related to the maturity of tobacco leaves.
[0004] However, the method of using a colorimeter to determine and identify the maturity of field tobacco leaves based on color values has not been widely applied and thoroughly researched. Most studies use a colorimeter to detect the color of cured tobacco leaves, thereby classifying them reasonably. There is a lack of systematic methods that directly correlate color values with the maturity of fresh tobacco leaves and establish maturity analysis models. Therefore, a method is needed to objectively, accurately and efficiently identify the maturity of field tobacco leaves before harvesting based on color values, in order to meet the demand of the tobacco industry for precise control of tobacco leaf quality and improve the overall efficiency of tobacco production. SUMMARY
[0005] Therefore, the application provides a tobacco maturity identification method and system based on chroma values, so as to realize objective and accurate judgment of different tobacco maturity, solve the problem of inaccurate maturity caused by subjective deviation and external influence in the traditional manual judgment method, and provide a scientific basis for timely picking and subsequent processing of tobacco.
[0006] The application solves the above technical problems by adopting the technical scheme of The tobacco maturity identification method comprises S1 invites experienced tobacco farmers and experts of tobacco companies to judge the maturity of the middle tobacco leaves in the tobacco field in the mature harvesting period, mark and record the maturity of the tobacco leaves, and the number of each maturity sample is not less than 100 pieces to ensure the universality and representativeness of the samples; The maturity of the marked tobacco leaves includes maturity, full maturity, still mature, under mature and false mature; S2, the color difference instrument is preheated to ensure the stability of the instrument performance, and the color difference instrument is calibrated by using a standard chroma plate, the calibration process comprises calibration of brightness, chroma coordinates and other parameters, the color difference value of the instrument measured on the standard chroma plate is ensured to be within the allowable range, and the accuracy and reliability of the measurement data are ensured; S3, the measurement plate is placed at the lower end of the marked leaf sample to facilitate measurement by the color difference instrument and avoid wrinkles or overlapping of the leaves during measurement, and 5 measurement points are selected at the leaf tip, middle and base of each tobacco leaf, the measurement points are distributed at different parts of the leaves to fully reflect the chroma characteristics of the tobacco leaves; Three chroma values are measured at each measurement point by the color difference instrument, and an average value is automatically calculated once for every 3 measurements, a total of 7500 chroma values are obtained to ensure sufficient data; S4, the measurement probe of the color difference instrument is vertically aligned with the measurement point to measure the chroma values L*, a* and b*, wherein L* represents the brightness value, reflecting the change from black to white; a* represents the redness value, reflecting the change from green to red; and b* represents the yellowness value, representing the change from blue to yellow, and the chroma values of each tobacco leaf are carefully and in detail recorded; S5, according to the data measured by the color difference instrument, the chroma formula
[0007]
[0008]
[0009]
[0010] The saturation (C*), color ratio (H) and hue angle (H 0 , so as to calculate the total chroma value (ΔE). wherein L0, a0, b0 are color parameters of the white board, wherein L0 = 91.97, a0 = -2.53, b0 = -5.68; S6, sequentially number the tobacco leaf samples of each maturity as 1-5, perform SPSS statistical analysis and Fisher linear discriminant analysis on the corresponding colorimetric value data, perform one-way analysis of variance (One-way Anova) for pairwise comparison between means, perform two-way ANOVA using Prism to draw a box plot to compare differences between different maturity groups, and find that there are significant differences between data of different maturity, so as to ensure the usability of the data; S7, taking different maturity of tobacco leaves as the independent variable, and taking colorimetric values L*, a*, b*, C* and ΔE as the dependent variable, respectively establishing the functional relationship between the colorimetric values L*, a*, b*, C* and ΔE and the maturity of tobacco leaves, and obtaining the Fisher discriminant function of the maturity of tobacco leaves; The colorimetric value data of the tobacco leaves to be identified are measured by a color difference meter, and other colorimetric values are calculated, and the Fisher discriminant function of the maturity of tobacco leaves is obtained by bringing the function value into the Fisher discriminant function of the maturity of tobacco leaves, and the maturity of the tobacco leaves to be identified is determined according to the size of the function value.
[0011] As an improvement of the above technical solution, Y1-Y5 represent mature, perfect, still mature, under mature, and false mature in the maturity of tobacco leaves, and the Fisher discriminant function of the maturity of tobacco leaves comprises Y1 = 13752.093L*- 3364.141a*- 1432.407b*- 10149.307C* + 16519.082ΔE- 675832.511; Y2 = 13741.407L*- 3112.255a*- 1736.42b*- 9801.936C* + 16499.745ΔE- 677337.005; Y3 = 13787.554L*- 3228.687a*- 1572.521b*- 10005.09C* + 16569.603ΔE- 677368.739; Y4 = 13715.413L*- 3213.784a*- 1557.2b*- 9956.065C* + 16475.53ΔE- 669948.526; Y5 = 13753.89L*- 3363.517a*- 1432.246b*- 10149.448C* + 16522.321ΔE- 675889.752; The chroma values L*, a*, b*, C*, ΔE are brought into each discriminant function to calculate five function values respectively, and the maturity classification corresponding to the largest function value is the maturity of the tobacco leaf.
[0012] As an improvement of the above technical solution, S1 invites experienced tobacco farmers and experts of tobacco companies to judge the maturity of the middle tobacco leaves in the tobacco field during the mature harvesting period, including The tobacco farmers and experts are divided into two rounds to identify the maturity of the tobacco leaves, the first round is to identify the maturity of the tobacco leaves by the farmers according to years of experience, and record the maturity of the tobacco leaves, then the experts of the tobacco company identify, only the same identification result of the same piece of tobacco leaf can be confirmed and finally recorded the maturity of the tobacco leaf; If the identification results are different, the farmers and experts jointly identify and confirm the final result; Ensure that each piece of tobacco leaf is identified by at least one farmer and one expert.
[0013] As an improvement of the above technical solution, S3 selects five measurement points at the tip, middle and base of each tobacco leaf, including Select one point, three points and one point at the tip, middle and base of each tobacco leaf respectively.
[0014] As an improvement of the above technical solution, the identification method further includes Based on S4, the color value L*, a*, b* data of tobacco leaves with different maturity is obtained, and the maturity of the tobacco leaves is identified by constructing an RF random forest model: The color value data is used as the feature data of the model input, the digital coding of the maturity grade of the tobacco leaf is used as the model identification label, and the RF random forest algorithm is used to model the feature set; the color value L*, a*, b* data and the corresponding digital coding of the maturity grade of the tobacco leaf are divided into a training data set and a test data set according to a ratio of 7:3, and the training data set is used for learning and training to obtain a trained RF random forest model.
[0015] To solve the above technical problems, the following technical solutions are also used: An identification system for the maturity of tobacco leaves, comprising A tobacco leaf sample maturity marking unit for judging the maturity of the middle tobacco leaves in the tobacco field during the mature harvesting period, marking and recording the maturity of the tobacco leaves; wherein the maturity of the marked tobacco leaves includes maturity, full maturity, still mature, under mature and false mature; A tobacco leaf sample chroma value collection unit for selecting multiple measurement points at the tip, middle and base of each tobacco leaf, measuring the brightness value, redness value and yellowness value at each measurement point by a color difference meter, and recording the original data of the chroma value of each tobacco leaf in detail; a chroma value data processing unit configured to obtain data of saturation, color ratio, hue angle and total chroma value of the tobacco leaves according to the chroma value raw data and the chroma formula, and further construct a standardized chroma value matrix capable of quantitatively representing the chroma value difference of the tobacco leaves with different maturity; a chroma value data analysis unit configured to sequentially number the tobacco leaf samples with different maturity, perform SPSS statistical analysis and Fisher linear discriminant analysis on the corresponding chroma value matrix data, take the different maturity of the tobacco leaves as independent variables, take the brightness value, redness value, yellowness value, saturation and total chroma value in the chroma value as dependent variables, establish a functional relationship between the chroma value and the maturity of the tobacco leaves, and obtain a Fisher discriminant function of the maturity of the tobacco leaves; a tobacco leaf maturity identification unit configured to determine the chroma value data of the tobacco leaves to be identified by using a color difference meter, obtain other chroma values through data processing, input the Fisher discriminant function of the maturity of the tobacco leaves to obtain a function value, and determine the maturity of the tobacco leaves to be identified according to the size of the function value.
[0016] The present application has the following beneficial effects: The tobacco leaf maturity identification method based on the chroma value provided by the present application uses a color difference meter for measurement, avoids subjective differences and influences of external environment in manual judgment, and improves the accuracy and consistency of identification of different maturity of the tobacco leaves. By constructing a maturity discriminant model, the maturity of a large number of tobacco leaf samples can be quickly and accurately identified, and the identification efficiency is improved. The present application provides more accurate guidance for tobacco picking, which helps to optimize the quality of the tobacco leaves and improve the economic benefits of the tobacco industry. BRIEF DESCRIPTION OF DRAWINGS
[0017] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without any creative labor.
[0018] Figure 1 is a method flowchart provided by the embodiments of the present application; Figure 2 is a system framework diagram provided by the embodiments of the present application; Figure 3 is a fitting function effect diagram of the chroma value and the maturity of the tobacco leaves provided by the embodiments of the present application; Figure 4 is a random forest model error curve diagram provided by the embodiments of the present application; Figure 5 is an importance analysis diagram of the color feature in the random forest model provided by the embodiments of the present application. DETAILED DESCRIPTION
[0019] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.
[0020] Furthermore, the following description is for purposes of illustration and not limitation, and specific details, such as particular system structures and techniques, are provided to facilitate a thorough understanding of the embodiments of the present invention. However, it will be apparent to those skilled in the art that the present invention may be practiced in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted so as not to obscure the description of the present invention with unnecessary detail.
[0021] Reference Figure 1 The first embodiment of the present invention relates to a method for identifying tobacco leaf maturity based on chromaticity value, comprising the following steps: S1 In tobacco fields in different growing areas, experienced tobacco farmers and experts from tobacco companies are invited to judge the maturity of tobacco leaves in the middle of the fields that are at the mature harvest period, mark and record the maturity of the tobacco leaves, and the number of samples for each maturity level is no less than 100 to ensure the breadth and representativeness of the samples.
[0022] Specifically, tobacco farmers and experts are divided into two rounds to identify the maturity of tobacco leaves. In the first round, tobacco farmers identify the maturity of tobacco leaves based on their years of experience and record the maturity of the tobacco leaves. Next, experts from the tobacco company identify the leaves. Only when the identification results are the same for the same tobacco leaf can the maturity of the tobacco leaf be confirmed and finally recorded. If the identification results are different, the tobacco farmers and experts will jointly identify and confirm the final results; Ensure that each tobacco leaf is authenticated by at least one tobacco farmer and one expert.
[0023] Among them, the maturity of the marked tobacco leaves includes five types: mature, fully ripe, moderately ripe, underripe, and falsely ripe.
[0024] S2. Turn on the colorimeter and preheat it to ensure stable instrument performance. Use a standard colorimetric plate to calibrate the colorimeter. The calibration process includes calibration of parameters such as brightness and chromaticity coordinates to ensure that the error of the colorimetric value of the standard colorimetric plate measured by the instrument is within the allowable range, ensuring the accuracy and reliability of the measurement data.
[0025] S3, place the measuring plate under the lower end of the marked leaf sample, facilitate the measurement with the colorimeter while avoiding the folding or overlapping of the leaf during measurement, select 1, 3, and 1 measurement points at the tip, middle, and base of each tobacco leaf, respectively, and these measurement points are distributed at different parts of the leaf to comprehensively reflect the colorimetric characteristics of the tobacco leaf; At each measurement point, 3 colorimetric values are measured by the colorimeter, and an average value is automatically calculated for each determination of 3 times, a total of 7500 colorimetric values are obtained to ensure sufficient data.
[0026] S4, use the measurement probe of the colorimeter to vertically align the measurement point and measure the colorimetric values L*, a*, and b*, wherein L* represents the lightness value, reflecting the change from black to white; a* represents the redness value, reflecting the change from green to red; and b* represents the yellowness value, representing the change from blue to yellow, and carefully and in detail record the colorimetric values of each tobacco leaf.
[0027] S5, according to the data measured by the colorimeter, according to the colorimetric formula
[0028]
[0029]
[0030]
[0031] Calculate the saturation (C*), the color ratio (H), and the hue angle (H 0 , so as to calculate the total colorimetric value (ΔE); In the formula, L0, a0, and b0 are the color parameters of the white plate, wherein L0 = 91.97, a0 = -2.53, and b0 = -5.68.
[0032] S6, sequentially number the tobacco leaf samples of each maturity as 1-5, perform SPSS statistical analysis and Fisher linear discriminant analysis on the corresponding colorimetric value data, use one-way ANOVA for pairwise comparison between means, use Prism's two-way ANOVA to draw a box plot to compare the differences between different maturity groups, and find that there are obvious differences between the data of different maturity, so as to ensure the usability of the data.
[0033] S7, take the different maturity of the tobacco leaf as the independent variable, and take the colorimetric values L*, a*, b*, C*, and ΔE as the dependent variable, respectively, to establish the functional relationship between the colorimetric values L*, a*, b*, C*, and ΔE and the maturity of the tobacco leaf, and obtain the Fisher discriminant function of the maturity of the tobacco leaf; When matching the maturity function, the matching relationship between the lightness value, the redness value, the yellowness value, the saturation, and the total color difference needs to be fully considered.
[0034] where the coefficients reflect the degree and direction of contribution of each parameter to the maturity discrimination.
[0035] The high weight of lightness value is the most important parameter in the discriminant function, indicating that the light and dark degree of tobacco plays a leading role in maturity distinction; The negative weight of redness value represents the green direction, and the negative weight indicates that the decrease of redness (more green) is negatively correlated with maturity; The negative weight of yellowness value represents the blue direction, and the negative weight indicates that the decrease of yellowness (more blue) is negatively correlated with maturity; And the negative weight of saturation reflects the purity of color, and the negative weight indicates that the low saturation (more gray) is negatively correlated with maturity; The introduction of positive weight of total color difference represents the overall difference between the color of tobacco and the standard white board, and the positive weight indicates that the greater the color difference (deviation from white) is, the closer to maturity; And the setting of the constant term is used to adjust the discrimination threshold of different categories to maximize the separation of the centroids of each tobacco in the multi-dimensional color space.
[0036] In this embodiment, Y1-Y5 represent mature, perfect, still mature, under mature, and false mature, respectively, in the maturity of tobacco, and the Fisher discriminant function of the maturity of tobacco includes Y1=13752.093L*-3364.141a*-1432.407b*-10149.307C*+16519.082ΔE-675832.511; Y2=13741.407L*-3112.255a*-1736.42b*-9801.936C*+16499.745ΔE-677337.005; Y3=13787.554L*-3228.687a*-1572.521b*-10005.09C*+16569.603ΔE-677368.739; Y4=13715.413L*-3213.784a*-1557.2b*-9956.065C*+16475.53ΔE-669948.526; Y5=13753.89L*-3363.517a*-1432.246b*-10149.448C*+16522.321ΔE-675889.752。
[0037] The chromaticity data of the tobacco leaves to be identified are measured by a colorimeter and other chromaticity values are obtained by calculation. The chromaticity values L*, a*, b*, C*, and ΔE are substituted into the Fisher discriminant function of tobacco leaf maturity to obtain five function values. The function values are compared, and the maturity classification corresponding to the one with the largest value is the maturity of the tobacco leaf.
[0038] Specifically, this embodiment introduces saturation and total color difference as color dimension features for judging the maturity of tobacco leaves based on the collected brightness value, redness value and yellowness value.
[0039] Saturation reflects color vividness, breaking the "green-yellow-brown-dark" turbidity. For example, when identifying underripe leaves, they exhibit a dark, turbid green color. The lightness of underripe leaves easily overlaps with that of immature leaves, while the redness of underripe leaves easily overlaps with that of mature leaves. With the introduction of saturation, significant differences in saturation are observed between underripe leaves, immature leaves, and mature leaves. A lack of vividness directly indicates immaturity.
[0040] For example, when identifying partially ripe leaves, they exhibit a turbid, yellowish-green color. The lightness of partially ripe leaves easily overlaps with that of mature leaves, and the yellowness of partially ripe leaves also easily overlaps with that of mature leaves. However, after introducing saturation, the saturation of partially ripe leaves is significantly lower than that of mature leaves, making saturation also useful for distinguishing the turbid, yellowish-green color of underripe leaves.
[0041] For example, when identifying mature leaves, it's also possible to distinguish overripe leaves from mature ones. Mature leaves exhibit vibrant yellow / orange color characteristics. The lightness of mature leaves easily overlaps with that of fully mature leaves, and the redness of mature leaves easily overlaps with that of overripe leaves. After introducing saturation, the saturation of mature leaves is significantly higher than that of other grades. Therefore, the inclusion of saturation directly indicates that bright colors indicate physiological maturity.
[0042] For example, when identifying fully ripe leaves, they exhibit a turbid brownish-yellow / brown-red color. The lightness of fully ripe leaves tends to overlap with that of underripe leaves, and the yellowness of fully ripe leaves tends to overlap with that of still-ripe leaves. However, after introducing saturation, the saturation of fully ripe leaves is significantly lower than that of mature leaves, making it possible to accurately distinguish overripe, turbid colors.
[0043] Furthermore, by introducing the total color difference, the overall color matching degree can be quantified to solve the pseudo-maturity problem of "a single indicator meets the standard but the overall deviation".
[0044] A single indicator can only judge a certain dimension of color (such as redness), but cannot guarantee that other dimensions (such as brightness and yellowness) meet the standards at the same time. The introduction of total color difference can comprehensively evaluate whether it is close to the target color.
[0045] In the judgment of under-ripe leaves, the introduction of total color difference can avoid misjudgment of "near maturity" state due to a single indicator.
[0046] In the judgment of still-ripe leaves, the introduction of total color difference can reflect that part meets the standard but the whole does not approach, and distinguish still-ripe as not fully meeting the standard.
[0047] In the judgment of mature leaves, the introduction of total color difference can verify that each dimension cooperates to meet the standard, and ensure that mature leaves are fully in line with the standard.
[0048] In the judgment of over-ripe leaves, the introduction of total color difference can mark the overall deviation caused by browning, and avoid misjudgment of maturity due to a single indicator.
[0049] The most critical introduction of total color difference is in the judgment of false-ripe leaves. The brightness value, redness value and yellowness value of false leaves have a large overlap with under-ripe leaves. Moreover, when a single saturation is introduced, the saturation of false leaves is also relatively close to that of under-ripe leaves. However, by introducing total color difference combined with saturation for judgment, false-ripe can be further distinguished as neither bright nor close to the standard.
[0050] The introduction of total color difference solves the problem of pseudo-maturity that a single indicator meets the standard but the whole is not coordinated through the overall matching degree, a comprehensive indicator, and ensures that the classification is closer to the target mature color.
[0051] When only using brightness value, redness value and yellowness value, there may be an "indicator overlap area" between different grades, making it difficult for the discrimination model to distinguish. After introducing saturation and total color difference, through the dual-dimension constraint of "brightness + overall matching degree", the discrimination specificity (the ability to correctly identify the grade) and sensitivity (the sensitivity of detecting grade differences) of each grade can be significantly improved.
[0052] Therefore, if only brightness value, redness value and yellowness value are introduced for judgment, it is easy to make identification errors. After introducing saturation and total color difference at the same time, the judgment of tobacco leaf maturity can be more accurate.
[0053] Reference Figure 2 The second embodiment of the present application relates to a tobacco leaf maturity identification system, comprising A tobacco leaf sample maturity marking unit is used for judging the maturity of middle tobacco leaves in the tobacco field during the mature harvesting period, marking and recording the maturity of the tobacco leaves; wherein the maturity of the marked tobacco leaves includes five kinds of maturity, over-ripe, still-ripe, under-ripe, false-ripe and mature; A tobacco leaf sample color value collection unit is used for selecting multiple measurement points at the tip, middle and base positions of each tobacco leaf, measuring the brightness value, redness value and yellowness value at each measurement point through a color difference meter, and recording the original data of the color value of each tobacco leaf in detail; a chroma value data processing unit configured to obtain data of saturation, color ratio, hue angle and total chroma value of the tobacco leaves according to the chroma value raw data and the chroma formula, and further construct a standardized chroma value matrix capable of quantitatively representing chroma value differences of different maturity of the tobacco leaves; a chroma value data analysis unit configured to sequentially number the tobacco leaf samples of different maturity, perform SPSS statistical analysis and Fisher linear discriminant analysis on the corresponding chroma value matrix data, take different maturity of the tobacco leaves as independent variables, take brightness value, redness value, yellowness value, saturation, total chroma value in the chroma value as dependent variables, establish a functional relationship between the chroma value and the maturity of the tobacco leaves, and obtain a Fisher discriminant function of the maturity of the tobacco leaves; a tobacco leaf maturity identification unit configured to determine chroma value data of the tobacco leaves to be identified by the color difference meter, obtain other chroma values through data processing, input the Fisher discriminant function of the maturity of the tobacco leaves to obtain a function value, and determine the maturity of the tobacco leaves to be identified according to the size of the function value.
[0054] In the above embodiments, the description of each embodiment has its own focus. The parts not described or recorded in detail in a certain embodiment can be referred to the related description of other embodiments.
[0055] The application will be described in detail below by using specific implementation scenarios: In the test examples of the present application, the middle tobacco leaves of Yunyan 87 variety were selected.
[0056] The test was carried out in a certain tobacco growing area from May 17, 2024 to September 30, 2024.
[0057] The tobacco fields of Yunyan 87 variety were selected. During the mature picking period of the tobacco fields, five local experienced tobacco farmers and five experts from Kunming Tobacco Company were invited to judge the maturity of the middle tobacco leaves in the tobacco fields, mark and record the maturity of the tobacco leaves. The tobacco farmers and experts were divided into two rounds to identify the maturity of the tobacco leaves. First, the tobacco farmers identified the maturity of the tobacco leaves according to their years of experience and recorded the maturity of the tobacco leaves. Then, the experts from Kunming Tobacco Company identified the maturity of the tobacco leaves. Only when the identification results of the same piece of tobacco leaf were the same, the maturity of the tobacco leaf could be confirmed and finally recorded. If the identification results were different, the tobacco farmers and experts would jointly identify and confirm the final result. It was ensured that each piece of tobacco leaf was identified by at least one tobacco farmer and one expert. 100 pieces of tobacco leaves of each maturity were selected as samples, and a total of 500 pieces of tobacco leaf samples were obtained. The maturity, planting area, variety, etc. were recorded in detail.
[0058] Select 1 point, 3 points, 1 point respectively at the tip of each tobacco leaf, the middle of the leaf, the base of the leaf flat and wrinkle-free surface, and use the colorimeter to measure the CIE-L* a* b* color space value at each point, wherein the colorimeter automatically calculates the average value once for every 3 measurements. Record the L*, a*, b* values of each point of each tobacco leaf. Calculate the saturation and total color value according to the color formula.
[0059] The average L, a, b, C, ΔE values of the colorimetric values of 100 pieces of tobacco leaves of each maturity of Yunyan 87 are shown in Table 1. It can be seen that for Yun 87 tobacco leaves, the average L, a, b, C, ΔE values of maturity, full maturity, maturity, under maturity and false maturity are different. The L, a, b, C values of full maturity tobacco leaves are significantly higher than those of other maturity; the L, a, b, C values of maturity tobacco leaves are the lowest; in addition, the five values of maturity and false maturity are relatively close, but each color value is different.
[0060] Table 1 Average colorimetric values of different maturity of Yun 87 tobacco leaves
[0061] To further clarify the relationship between colorimetric values L*, a*, b*, C* and ΔE and tobacco leaf maturity, the fitting function relationship between colorimetric values L*, a*, b*, C* and ΔE and the maturity of Yun 87 tobacco leaves was established with the different maturity of Yun 87 tobacco leaves as the independent variable and the colorimetric values L*, a*, b*, C* and ΔE as the dependent variable, and the results are shown in Figure 3 It can be seen that for Yun 87 tobacco leaves of different maturity, maturity, full maturity, maturity, under maturity and false maturity, the group centroids are not completely overlapped, which shows that the Fisher discrimination effect of the colorimetric values of tobacco leaves of different maturity in the field is good, and the five maturity degrees are effectively distinguished.
[0062] The middle field tobacco leaves of Yun 87 were used as the analysis sample, and the tobacco leaf samples of each maturity were numbered in sequence as 1-5, and the corresponding colorimetric values were entered into SPSS to obtain the canonical discriminant function and its eigenvalues. As shown in Table 2, the eigenvalues of functions 3 and 4 are significantly smaller, and the cumulative total variation coefficient is only 1%, and the cumulative percentage of the first two numbers reaches 99%, and the cumulative total variation coefficient also reaches 99%, indicating that the functions have discriminant significance.
[0063] Table 2 Canonical discriminant function and its eigenvalues
[0064] Table 3 shows the Fisher discriminant results of different maturity of the middle part of field tobacco leaves. As can be seen from the results, the recognition rate of the training set is 94.4%, and the recognition rate of the cross-validation is 94.4%. Overall, the effect is good, and the discriminant model has certain applicability.
[0065] Table 3 Fisher discriminant results of tobacco leaf maturity
[0066] For the tobacco leaves of Yun 87 which were not identified by Tanaka, the color value data of the tobacco leaves were obtained according to the same color value determination method, and the function values of the sample were calculated by substituting the data into the established Fisher discriminant function formula. According to the size of the function value, the corresponding maturity was selected, and the identification results were evaluated by inviting tobacco farmers and experts, thereby further verifying the effectiveness and universality of the method.
[0067] The present application also includes a third embodiment as follows, which is different from the first embodiment in that: The color values L*, a* and b* of tobacco leaves of different maturity are obtained to establish a data set, and an RF random forest model is constructed for tobacco leaf maturity identification. The specific steps include: The color value data are taken as the characteristic data of the model input, and the digital coding of the tobacco leaf maturity grade is taken as the model identification label, and the RF random forest algorithm is used to model the characteristic set. The color values L*, a* and b* and the corresponding digital coding of the tobacco leaf maturity grade are divided into a training data set and a test data set according to a ratio of 7:3, and the training data set is used for learning and training to obtain the trained RF random forest model.
[0068] Among them, the number of decision trees is 200, and the number of leaf nodes is 1. The number of decision trees is sufficient, which can make the diversity of the model higher, thereby reducing the risk of overfitting. However, too many trees can lead to diminishing returns, that is, after increasing the number of trees to a certain extent, the performance of the model is no longer obviously improved. According to the error curve Figure 4 It can be seen that when the number of decision trees in the model is 200, the error is close to 0, and the error value tends to be stable, so the number of decision trees is selected as 200. A smaller number of leaf nodes can make the tree simpler, thereby making the calculation speed of the model faster.
[0069] An importance analysis graph of the color characteristics in the RF random forest model is established, and the results are shown in Figure 5 It can be seen that the lightness value (L*) has the greatest influence on the model, which means that the lightness value (L*) has the greatest influence or importance on the identification of the tobacco leaf maturity grade; the importance of the yellow value (b*) is second, which means that the yellow value (b*) has a certain contribution to the identification of the tobacco leaf maturity grade; the importance of the red value (a*) is the lowest, which means that the red value has a weak influence on the identification of the tobacco leaf maturity grade.
[0070] The trained RF random forest model is identified by the test set data and output, and the accuracy is 98%.
[0071] For the cloud 87 tobacco leaves of which the maturity is not identified by Tianzhong, the color value data of the tobacco leaves is obtained according to the same color value determination method, the data is substituted into the trained RF random forest model, and the maturity identification result of the tobacco leaves is output. The maturity identification result is highly consistent with the maturity grade result identified by experts and tobacco farmers, and the good applicability of the method is further verified.
[0072] The above examples are only used to illustrate the technical solutions of the present application, but not to limit it; although the present application has been described in detail with reference to the foregoing examples, those skilled in the art should understand that the technical solutions recorded in the foregoing examples can be modified, or some technical features can be replaced by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application, and should be included in the protection scope of the present application.
Claims
1. A method for identifying tobacco leaf maturity, characterized in that: The following steps are included S1. In tobacco fields in different growing regions, invite experienced tobacco farmers and tobacco company experts to determine the maturity of tobacco leaves in the middle of the fields that are ripe for harvest. Mark and record the maturity of the leaves. The number of samples for each maturity level should be no less than 100 to ensure the breadth and representativeness of the samples. The maturity of the marked tobacco leaves includes five levels: mature, fully mature, moderately mature, under-ripe, and falsely mature. S2. Preheat and calibrate the colorimeter to ensure that the colorimetric error of the standard colorimetric plate measured by the instrument is within the allowable range; S3. Place the measuring plate under the marked leaf sample to facilitate colorimeter measurement while avoiding wrinkles or overlaps on the leaf. Select five measurement points at the tip, middle, and base of each leaf. At each measuring point, measure at least 3 chromaticity values using a colorimeter, and automatically calculate the average value every 3 measurements; S4. Use the colorimeter's measuring probe to align vertically with the measuring point and measure its chromaticity values L*, a*, and b*. L* represents the lightness value, reflecting the change from black to white; a* represents the redness value, reflecting the change from green to red; and b* represents the yellowness value, representing the change from blue to yellow. Record the chromaticity value of each tobacco leaf in detail. S5. Based on the data measured by the colorimeter, calculate the saturation C*, color ratio H, and hue angle H according to the chromaticity formula. 0 , and thus calculate the total chromaticity value ΔE; S6. Tobacco leaf samples of different maturity levels were numbered 1-5, and the corresponding color value data were statistically analyzed by SPSS and Fisher linear discriminant analysis. One-way analysis of variance was used to compare the means between the two groups, and box plots were drawn using Prism's two-way analysis of variance to compare the differences between the groups of different maturity levels. Significant differences were found between the data of different maturity levels to ensure the usability of the data. S7. Using different maturity of tobacco leaves as independent variables and chromaticity values L*, a*, b*, C*, and ΔE as dependent variables, establish functional relationships between chromaticity values L*, a*, b*, C*, and ΔE and tobacco leaf maturity, and obtain the Fisher discriminant function of tobacco leaf maturity; The colorimetric data of the tobacco leaves to be identified are measured by a colorimeter, and other colorimetric values are obtained by calculation. The function value is substituted into the Fisher discriminant function of tobacco leaf maturity to obtain the function value, and the maturity of the tobacco leaves to be identified is determined according to the size of the function value.
2. The method for identifying tobacco leaf maturity according to claim 1, wherein: Y1-Y5 represent the maturity of tobacco leaves, namely, mature, fully mature, moderately mature, under-mature, and falsely mature. The Fisher discriminant function of tobacco leaf maturity includes Y1=13752.093L*-3364.141a*-1432.407b*-10149.307C*+16519.082ΔE-675832.511; Y2=13741.407L*-3112.255a*-1736.42b*-9801.936C*+16499.745ΔE-677337.005; Y3=13787.554L*-3228.687a*-1572.521b*-10005.09C*+16569.603ΔE-677368.739; Y4=13715.413L*-3213.784a*-1557.2b*-9956.065C*+16475.53ΔE-669948.526; Y5=13753.89L*-3363.517a*-1432.246b*-10149.448C*+16522.321ΔE-675889.
752.
3. The method for identifying tobacco leaf maturity according to claim 1, wherein: S1 invites experienced tobacco farmers and tobacco company experts to judge the maturity of the tobacco leaves in the middle of the tobacco field that are at the mature harvest period, including Tobacco farmers and experts are divided into two rounds to identify the maturity of tobacco leaves. In the first round, tobacco farmers identify the maturity of tobacco leaves based on their years of experience and record the maturity of the tobacco leaves. Then, experts from the tobacco company identify the leaves. Only when the identification results are the same for the same tobacco leaf can the maturity of the tobacco leaf be confirmed and finally recorded. If the identification results are different, the tobacco farmers and experts will jointly identify and confirm the final results; Ensure that each tobacco leaf is authenticated by at least one tobacco farmer and one expert.
4. The method for identifying tobacco leaf maturity according to claim 1, wherein: S3 selects 5 measurement points at the tip, middle and base of each tobacco leaf, including Select 1 point, 3 points, and 1 point from the tip, middle, and base of each tobacco leaf on the smooth, wrinkle-free surface.
5. The method for identifying tobacco leaf maturity according to claim 1, wherein: The color formula of S5 includes Where L0, a0, and b0 are the color parameters of the whiteboard, L0=91.97, a0=-2.53, and b0=-5.
68.
6. The method for identifying tobacco leaf maturity according to claim 1, wherein: The identification method further includes The chromaticity values L*, a*, and b* data of tobacco leaves of different maturity are obtained based on S4, and the maturity of tobacco leaves is identified by building an RF random forest model: The chromaticity value data is used as the feature data of the model input, and the digital code of the tobacco leaf maturity grade is used as the model identification label. The RF random forest algorithm is used to model the feature set. The chromaticity value L*, a*, b* data and the corresponding tobacco leaf maturity grade digital code are divided into training data set and test data set in a ratio of 7:
3. The training data set is used for learning and training to obtain the trained RF random forest model.
7. A tobacco leaf maturity identification system, characterized by: The system is used to implement the tobacco leaf maturity identification method according to any one of claims 1 to 5, comprising: The tobacco leaf sample maturity marking unit is used to judge the maturity of tobacco leaves in the middle of the tobacco field that are at the mature harvest period, mark and record the maturity of the tobacco leaves; the maturity of the marked tobacco leaves includes five types: mature, fully mature, moderately mature, under-mature, and falsely mature; The tobacco leaf sample color value acquisition unit is used to select multiple measurement points at the tip, middle, and base of each tobacco leaf, measure the lightness, redness, and yellowness values at each measurement point using a colorimeter, and record the original color value data of each tobacco leaf in detail; A chromaticity value data processing unit is used to obtain data such as saturation, color ratio, hue angle and total chromaticity value of tobacco leaves based on the original chromaticity value data and the chromaticity formula, and further construct a standardized chromaticity value matrix that can quantitatively characterize the chromaticity value differences of tobacco leaves at different maturity levels; The chromaticity value data analysis unit is used to sequentially number tobacco leaf samples of different maturity levels, perform SPSS statistical analysis and Fisher linear discriminant analysis on the corresponding chromaticity value matrix data, use the different maturity levels of tobacco leaves as independent variables, and use the lightness value, redness value, yellowness value, saturation, and total chromaticity value in the chromaticity value as dependent variables to establish a functional relationship between the chromaticity value and the maturity level of tobacco leaves, thereby obtaining the Fisher discriminant function of the maturity level of tobacco leaves; The tobacco leaf maturity identification unit is used to measure the color value data of the tobacco leaf to be identified by using a colorimeter and obtain other color values through data processing, and then bring the color value into the Fisher discriminant function of the tobacco leaf maturity to obtain the function value, and determine the maturity of the tobacco leaf to be identified according to the size of the function value.