Rice taste detection method, system and equipment, storage medium and program product

Abstract

The invention provides a rice taste detection method, system and device, a storage medium and a program product, and relates to the technical field of agricultural product quality analysis and detection. The method comprises the following steps: acquiring target spectral data corresponding to rice to be detected; according to the sample type of the to-be-detected rice, determining a corresponding rice taste near-infrared prediction regression model, the rice taste near-infrared prediction regression model is constructed based on a corresponding relation between sample spectrum data of sample rice of different sample types and sample taste meter correction values corresponding to the sample spectrum data; and inputting the target spectrum data into the rice taste near-infrared prediction regression model to obtain a taste prediction value corresponding to the to-be-detected rice output by the rice taste near-infrared prediction regression model. According to the method, the rice taste value can be quickly and nondestructively detected, and the taste value can be quickly and efficiently predicted only by performing spectral data acquisition on a small amount of rice samples during detection.

Description

Technical Field

[0001] The present invention relates to the technical field of agricultural product quality analysis and detection, and in particular to a rice taste detection method, system, equipment, storage medium and program product. Background Art

[0002] Improving the taste quality of rice is an important goal of rice breeding. How to objectively evaluate the quality of rice taste has always been a hot topic for breeders. For a long time, the evaluation of rice taste quality has been mainly based on sensory evaluation. The evaluation method is to organize a certain number of people who have received tasting training to taste the rice and assign points to each variety. Since sensory tasting requires a lot of time, manpower and material resources, it is difficult to quickly and effectively identify the taste value of a large number of samples, and it is not suitable for large-scale screening of rice taste breeding materials. At the same time, due to the existence and interference of various factors (preference, difference), the error of human taste tasting is large, and there is subjectivity and instability. The use of instrument detection can eliminate human interference and objectively give the taste value of the variety, and can improve the detection efficiency.

[0003] The existing rice taste meter is used for taste detection, which requires less sample (30 grams of polished rice). It is faster, easier to operate, and more objective than the sensory method, but it needs to be steamed. It is suitable for breeding intermediate materials, variety comparison test materials, regional test varieties, etc., and can reflect the taste quality of rice to a certain extent. However, the disadvantage is that the rice taste meter test model is based on japonica rice varieties, while the main cultivated varieties are indica rice varieties, so it is different from the current rice raw material quality model basis. In addition, rice, especially indica rice, has many types and types, and the rice taste meter has not established a relationship with the sensory tasting of indica rice, so it is still imperfect to evaluate the taste quality of indica rice by the existing rice taste meter alone. In addition, the use of the rice taste meter for taste detection requires steaming of rice, which makes the detection speed slow, the pre-treatment process cumbersome, the sample dosage large, and the shortcomings of damaging the rice grains cannot be completely solved. At the same time, rice cannot be planted after it has been processed into polished rice, which makes it impossible for breeders to screen the taste quality of the middle and lower generations, resulting in longer breeding time and lower breeding efficiency.

[0004] Therefore, there is an urgent need for a rice taste detection method, system, equipment, storage medium and program product to solve the above problems. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a rice taste detection method, system, equipment, storage medium and program product.

[0006] The present invention provides a rice taste detection method, comprising: Obtain target spectrum data corresponding to the rice to be detected; According to the sample type of the rice to be tested, a corresponding near-infrared prediction regression model of rice taste is determined, wherein the near-infrared prediction regression model of rice taste is constructed based on the correspondence between sample spectral data of sample rice of different sample types and sample taste meter correction values ​​corresponding to the sample spectral data; The target spectrum data is input into the rice taste near-infrared prediction regression model to obtain the taste prediction value corresponding to the rice to be tested output by the rice taste near-infrared prediction regression model.

[0007] According to a rice taste detection method provided by the present invention, the rice taste near-infrared prediction regression model is constructed by the following steps: Acquire sample spectral data corresponding to first sample rice of different sample types, wherein the sample types include rice type, brown rice type and polished rice type; the sample spectral data is constructed by spectrally preprocessing the original spectral data corresponding to the first sample rice; Obtaining a sample rice taste meter measurement value corresponding to the first sample rice; Based on the taste meter correction regression model, the sample rice taste meter measurement value is corrected to obtain the sample taste meter correction value; wherein the taste meter correction regression model is constructed based on the corresponding relationship between the sensory evaluation value and the rice taste meter measurement value; Based on partial least squares regression, the sample spectral data is associated with the sample taste meter correction value, and the rice taste near-infrared prediction regression model corresponding to the first sample rice of different sample types is constructed.

[0008] According to a rice taste detection method provided by the present invention, the taste meter correction regression model is constructed by the following steps: Obtaining a sensory evaluation value and a rice taste meter measurement value corresponding to a second rice sample, wherein the second rice sample and the first rice sample are rice of the same variety, mature season and location; The rice taste meter measurement value corresponding to the second rice sample is associated with the sensory evaluation value to establish the taste meter correction regression model, and the taste meter correction regression model is used to correct the rice taste meter measurement value.

[0009] According to a rice taste detection method provided by the present invention, the method further comprises: The near-infrared diffuse reflectance spectrum of the first sample rice is scanned to obtain the original spectrum data corresponding to the first sample rice.

[0010] According to a rice taste detection method provided by the present invention, the spectral preprocessing includes: removing abnormal spectra in the original spectral data, and performing spectral data preprocessing on the original spectral data after the abnormal spectra are removed based on derivative transformation and standard normal variable transformation.

[0011] According to a rice taste detection method provided by the present invention, after inputting the target spectrum data into the rice taste near-infrared prediction regression model to obtain the taste prediction value corresponding to the rice to be detected output by the rice taste near-infrared prediction regression model, the method further includes: According to the taste prediction value corresponding to the rice to be detected and the target spectral data, the rice taste near-infrared prediction regression model is optimized and updated to obtain an optimized and updated rice taste near-infrared prediction regression model.

[0012] The present invention also provides a rice taste detection system, comprising: A spectral data acquisition module is used to obtain target spectral data corresponding to the rice to be tested; A model determination module, for determining a corresponding near-infrared prediction regression model for rice taste according to the sample type of the rice to be detected, wherein the near-infrared prediction regression model for rice taste is constructed based on the correspondence between sample spectral data of sample rice of different sample types and sample taste meter correction values ​​corresponding to the sample spectral data; The taste detection module is used to input the target spectrum data into the rice taste near-infrared prediction regression model to obtain the taste prediction value corresponding to the rice to be detected output by the rice taste near-infrared prediction regression model.

[0013] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any of the above-mentioned rice taste detection methods is implemented.

[0014] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the rice taste detection method as described in any one of the above is implemented.

[0015] The present invention also provides a computer program product, comprising a computer program, wherein when the computer program is executed by a processor, the rice taste detection method as described above is implemented.

[0016] The rice taste detection method, system, device, storage medium and program product provided by the present invention acquire target spectral data corresponding to rice, and based on partial least squares regression, associate the target spectral data with the taste meter correction value to establish a corresponding rice taste near-infrared prediction regression model. During detection, only a small amount of rice samples need to be used for spectral data collection, so that the taste value can be predicted quickly and efficiently. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic diagram of correlation analysis between the rice taste meter measurement value and the sensory evaluation value provided by the present invention; Figure 2 A distribution histogram of taste correction values ​​based on the first sample rice provided by the present invention; Figure 3 A graph showing the regression model for predicting food taste by near infrared provided by the present invention; Figure 4 A schematic diagram of the process of the rice taste detection method provided by the present invention; Figure 5 This is one of the schematic diagrams of the model verification results based on the taste meter correction values ​​of 92 samples and the predicted values ​​of the rice taste near-infrared prediction regression model provided by the present invention; Figure 6 The second schematic diagram of the model verification result based on the taste meter correction value of 92 samples and the predicted value of the rice taste near infrared prediction regression model provided by the present invention; Figure 7 The third schematic diagram of the model verification result based on the taste meter correction value of 92 samples and the predicted value of the rice taste near infrared prediction regression model provided by the present invention; Figure 8 A schematic diagram of the overall process of establishing, verifying and applying the method for rapid rice taste detection based on near-infrared high-throughput detection provided by the present invention; Fig. 9 A schematic diagram of the structure of the rice taste detection system provided by the present invention; Fig.10 This is a schematic structural diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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 creative work are within the scope of protection of the present invention.

[0020] In view of the problems existing in the prior art, the present invention provides a method for rapid detection of rice taste based on near-infrared high-throughput. The present invention adopts the Botong DA7250 grain near-infrared analyzer to collect spectral data of rice, brown rice and polished rice, and then establishes a regression model with the spectral data and the corresponding taste values, so that regardless of the type of rice sample, the taste value can be predicted quickly and with high throughput; in addition, the rice taste detection method provided by the present invention is simple to operate, and the entire testing process only takes 15 seconds, which can greatly improve the rice taste screening throughput and solve the problem that breeders cannot screen rice taste quality indicators for breeding materials in the middle and low generations (with less seeds).

[0021] In the present invention, a near infrared prediction regression model of rice taste is first constructed, and the steps are as follows: Acquire sample spectral data corresponding to first sample rice of different sample types, wherein the sample types include rice type, brown rice type and polished rice type; the sample spectral data is constructed by spectrally preprocessing the original spectral data corresponding to the first sample rice; Obtaining a sample rice taste meter measurement value corresponding to the first sample rice; Based on the taste meter correction regression model, the sample rice taste meter measurement value is corrected to obtain the sample taste meter correction value; wherein the taste meter correction regression model is constructed based on the corresponding relationship between the sensory evaluation value and the rice taste meter measurement value; Based on partial least squares regression, the sample spectral data is associated with the sample taste meter correction value, and the rice taste near-infrared prediction regression model corresponding to the first sample rice of different sample types is constructed.

[0022] Specifically, the present invention first collects a first group of rice samples (this group of rice samples can be recorded as the first sample rice) to establish a rice taste near-infrared prediction regression model. The first sample rice is a rice sample that was harvested in 2024 and has different component contents. Each sample is not less than 120 grams (clean, free of impurities and mixed samples), and the moisture is balanced. In the present invention, the number of the first sample rice is 2163.

[0023] Then, 2163 portions of the first sample rice were prepared into three categories: the first category, 15 grams of rice; the second category, 15 grams of brown rice; the third category, 45 grams of polished rice; the rice, brown rice and polished rice were respectively used for near-infrared scanning to obtain the corresponding near-infrared raw spectral data.

[0024] In the present invention, the spectral data is collected using a DA7250 near-infrared analyzer, and the collection process is as follows: First, turn on the DA7250 near-infrared analyzer and preheat it for 30 minutes. Refer to the near-infrared analyzer operating instructions to complete the basic settings of the instrument (no need to change after setting it once, just use it directly next time). Take about 10 grams of the first sample rice and pour it into the sample tray. The sample is required to fall naturally and be scraped flat with a ruler.

[0025] Then, the plate with the sample is placed at a designated position on the magnetic disk, and a continuous wavelength near-infrared scanning is used to collect the spectrum. The scanning spectrum range is 950nm-1650nm, the scanning mode is continuous wavelength scanning, and the spectrum data is collected once every 5nm. In order to overcome the spectral drift caused by sample differences and reduce errors, each sample is repeated 3 times, and the scan is repeated 3 times each time, and the average value of the sample spectrum absorbance is taken as the spectrum data. The present invention scans different types of rice, brown rice, and polished rice corresponding to 2163 portions of the first sample rice, respectively, to obtain corresponding sample spectrum data.

[0026] Furthermore, 30 grams of the third type of polished rice was weighed and subjected to relevant detection of the rice taste meter to obtain the rice taste meter measurement value. The taste meter measurement range is shown in Table 1: Table 1 Statistical results of the taste value of the first sample rice

[0027] It can be concluded from Table 1 that the first sample rice taste value selected by the present invention has a wide range from 48.8 to 95.6 and is highly representative.

[0028] In the present invention, the sample rice taste meter measurement values ​​are obtained by processing the first sample rice into polished rice and measuring the rice taste value. The taste value measurement adopts STA1B rice taste meter. The specific process of the taste value measurement is as follows: Washing and weighing rice: Pour 30 grams of polished rice weighed in the above embodiment into a stainless steel tank, add appropriate amount of water to soak, connect a rice washing device, rinse with water for about 20 seconds, wipe the surface of the stainless steel tank dry, and then add water according to the total weight (rice weight + tank weight + water weight) to weigh (the amount of water added for indica rice is generally 1.4 times the weight of rice, and 1.35 times for japonica rice), then cover the stainless steel tank with filter paper and seal it with a rubber sleeve. It should be noted that the above-mentioned soaking, rice washing and water adding operations must be completed within 30 minutes.

[0029] Steaming: Put the stainless steel pot covered with filter paper into the steamer and cover the pot. Turn on the power of the electric rice cooker and set the steaming time. In the present invention, after soaking for 30 minutes, steaming for 40 minutes is started, and the rice is stewed for 10 minutes after stopping heating.

[0030] Cooling: After the rice is cooked, take out the stainless steel pot from the steamer, remove the filter paper, and gently stir the rice in the pot (stir in one direction to keep the rice grains intact). After stirring, cover the filter paper again and put the pot in a special cooler. Turn on the switch to start cooling for 20 minutes, then turn off the cooler, take out the pot from the cooler, remove the filter paper, and cover the pot. At this time, it needs to be cooled at room temperature for 1.5 hours and wait for subsequent testing.

[0031] Rice taste value measurement: Prepare the rice to be tested into a rice ball measurement circle, put the measurement circle into the instrument, and measure its taste value. Each measurement circle needs to be measured once in both directions, and each rice sample needs to be measured three times. The instrument automatically displays the comprehensive taste measurement value of each rice sample, records the corresponding data, and obtains the rice taste meter measurement value corresponding to the first sample of rice.

[0032] Based on the above embodiment, the taste meter correction regression model is constructed by the following steps: Obtaining a sensory evaluation value and a rice taste meter measurement value corresponding to a second rice sample, wherein the second rice sample and the first rice sample are rice of the same variety, mature season and location; The rice taste meter measurement value corresponding to the second rice sample is associated with the rice sensory evaluation value to establish the taste meter correction regression model, and the taste meter correction regression model is used to correct the rice taste meter measurement value.

[0033] Due to the differences in measurement methods and evaluation criteria, the taste meter measurement value obtained by using a taste meter may have certain errors or inconsistencies. In the present invention, the sample taste correction value is a taste meter correction regression model established by the sample sensory evaluation value and the rice taste meter measurement value, and the result obtained after adjusting the sample rice taste meter measurement value. This correction value can not only avoid the subjectivity and complexity brought by the sensory evaluation value, but also solve the imperfection of evaluating the taste quality of rice by a single taste meter, so as to be closer to the real taste quality of the sample, because it has removed the errors and inconsistencies in the original measurement, and is also convenient for the verification and optimization of the later data model.

[0034] Specifically, the present invention first collects sample rice (i.e., second sample rice) of the same variety, mature season and location as the first sample rice, constructs a rice group for artificial sensory evaluation, and uses the sensory evaluation data to correct the rice taste meter measurement values ​​to form rice taste meter correction values, wherein each portion of the second sample rice will be processed into 500 grams of polished rice.

[0035] In the present invention, the second sample rice is a rice sample harvested in the mature season of 2024, with different component contents, each sample is not less than 2 kg, and the moisture is balanced, totaling 203 samples. Before the sensory evaluation of the second sample rice, it is first necessary to steam it, weigh 500 grams of polished rice, use an electromagnetic heating (Induction Heat, referred to as IH) rice cooker with a capacity of 4 liters, use a rice-water ratio of 1:1.4, soak for 30 minutes, steam for 40 minutes, and simmer for 10 minutes.

[0036] Then, professionally trained tasters compared and evaluated the smell, appearance structure, taste, viscosity, hardness and comprehensive score of the test sample rice with the reference sample one by one. According to the degree of goodness or badness, the scale evaluation was carried out to obtain the sensory evaluation value of the sample rice. For example, after tasting once, it was confirmed that there was a significant difference with the reference sample. This difference was defined as the best (plus 3 points) and the worst (minus 3 points); after tasting once, it was not confirmed that there was a significant difference, but a certain degree of difference was felt. This difference was defined as better (plus 2 points) and worse (minus 2 points); after tasting once, it was still unclear, and the difference was felt after tasting the second time. This difference was defined as slightly better (plus 1 point) and slightly worse (minus 1 point); after tasting the second time, it was still impossible to judge whether there was a difference. This difference indicated that the test sample rice was the same as the reference sample, and the score remained unchanged.

[0037] After the evaluation, data statistics were performed to determine the relative average score of each second rice sample. The score was then quantified based on the score of the reference sample (Huang Huazhan: sensory evaluation score of 75 points) to obtain the sensory evaluation taste value of each second rice sample.

[0038] Then, a mathematical model related to the taste sensory tasting value (i.e., the sample rice sensory evaluation value) and the rice taste meter (i.e., the sample rice taste meter measurement value) is established. Specifically, the rice taste meter measurement process in the above embodiment can be referred to, and the second sample rice is measured by the rice taste meter. Then, based on the obtained sample rice taste meter measurement value and the sample rice sensory evaluation value, a mathematical model related to the taste value is established. Figure 1 Schematic diagram of correlation analysis between the rice taste meter measurement value and the sensory evaluation value provided by the present invention, the horizontal axis X represents the taste value measured by the taste meter, the vertical axis Y represents the sensory evaluation value, and the established regression equation is: y=1.0369x+3.1532, the coefficient R 2 It is 0.8568.

[0039] Through analysis, it can be seen that there is a high correlation between the taste value measured by the rice taste meter and the sensory evaluation value. The taste meter correction regression model established in the present invention can be used to correct the taste value of the rice taste meter, and then obtain the taste correction value of the first sample rice. Figure 2 The taste correction value distribution histogram based on the first sample rice provided by the present invention is as follows: Figure 2 As shown, the taste values ​​of the samples basically conform to the normal distribution after correction, indicating that the selected samples are very representative.

[0040] In the above embodiment, the rice taste meter used is STA1B, and the rice taste meter is correlated with the sensory evaluation of rice in different planting areas and types, and then the regression prediction equation is established using the corrected rice taste meter value and near-infrared spectrum data. On the one hand, the difference problem of evaluating the taste quality of rice by a single rice taste meter is solved. On the other hand, whether it is rice, brown rice or polished rice samples, rapid taste detection can be achieved, there are no restrictions on samples, and the scope of application is wide. In addition, brown rice can be returned to breeders for sowing. At the same time, in the actual rice taste detection process, there is no need to pre-treat the rice sample, which realizes non-destructive detection, and the dosage is small, and the screening of rice taste quality can be achieved in the low generation.

[0041] Based on the above embodiment, the method further includes: The near-infrared diffuse reflectance spectrum of the first sample rice is scanned to obtain the original spectrum data corresponding to the first sample rice.

[0042] On the basis of the above embodiment, the spectrum preprocessing includes: removing abnormal spectra in the original spectrum data, and performing spectrum data preprocessing on the original spectrum data after the abnormal spectra are removed based on derivative transformation and standard normal variable transformation.

[0043] In the present invention, the software Unscramble X10.4 is used to export the near-infrared raw spectral data of the first sample rice, and the derivative transformation (Derivative) and the standard normal variable transformation (Standard NormalVariate, referred to as SNV) are used to perform second-order derivative processing, so as to pre-process the raw spectral data to improve the quality of the spectral data and make it more suitable for subsequent analysis and modeling. It should be noted that in the present invention, before the taste detection of the rice to be detected, the raw spectral data of the rice to be detected also needs to undergo the above-mentioned spectral data pre-processing process.

[0044] Then, the near-infrared raw spectral data (i.e., sample spectral data) after spectral data preprocessing was associated with the corresponding taste meter correction value using chemometric software. The three near-infrared spectral data of paddy rice, brown rice, and polished rice of the first sample rice were respectively derived; the sample spectral data were associated with the taste meter correction value using software Unscramble X10.4.

[0045] Furthermore, a partial least squares (PLS) model was established using chemometric software. That is, the partial least squares method in Unscramble X10.4 software was used to establish near infrared prediction regression models for the taste of rice, brown rice, and polished rice. The correlation between the predicted values ​​of the model and the measured values ​​of the sample rice can be found in Table 2: Table 2 Correlation results between predicted values ​​and measured values ​​(rice, brown rice, polished rice)

[0046] Figure 3 This is a graph of the near infrared prediction regression model of taste provided by the present invention. Figure 3 (A) is the near infrared prediction regression model of rice taste. Figure 3 (B) is the near infrared prediction regression model of the taste of brown rice. Figure 3 (C) is the near infrared prediction regression model of taste corresponding to polished rice; Figure 3 The horizontal axis is the measured value, and the vertical axis is the predicted value of the taste near infrared prediction regression model, that is, the taste prediction value output by the model. It can be seen that the rice taste near infrared prediction regression model constructed for different rice types in the present invention has a more significant prediction effect.

[0047] In the present invention, the collected spectral data and the corresponding taste meter correction value are input into the regression model and trained by a machine learning algorithm (such as linear regression, nonlinear regression, support vector regression, etc.). During the training process, the algorithm will continuously adjust the model parameters to minimize the difference between the predicted value and the actual value. In addition, after the training is completed, the model needs to be verified using an independent validation set to evaluate its prediction performance. If the prediction performance is not ideal, the model parameters can be further adjusted or the sample size can be increased to optimize the model.

[0048] Figure 4 The schematic diagram of the process of the rice taste detection method provided by the present invention is as follows: Figure 4 As shown, the present invention provides a rice taste detection method, comprising: Step 401, obtaining target spectrum data corresponding to the rice to be detected; Step 402, determining a corresponding near-infrared prediction regression model for rice taste according to the sample type of the rice to be tested, wherein the near-infrared prediction regression model for rice taste is constructed based on the correspondence between sample spectral data of sample rice of different sample types and sample taste meter correction values ​​corresponding to the sample spectral data; Step 403: input the target spectrum data into the rice taste near-infrared prediction regression model to obtain a taste prediction value corresponding to the rice to be tested output by the rice taste near-infrared prediction regression model.

[0049] In the present invention, when the rice taste near-infrared prediction regression model receives the target spectral data as input, the model will calculate according to the learned pattern, and then output a prediction value, which represents the taste prediction value of the rice to be tested, thereby realizing a non-destructive, fast and relatively accurate rice taste value prediction method, and the specific process includes: 1. Collect a group of samples (i.e. the third sample rice) to verify the accuracy of the near-infrared prediction regression model of rice taste. Collect 92 rice samples that were harvested in 2024, had different component contents, and each sample was no less than 120 grams (clean without impurities and mixed samples), and the moisture content was balanced.

[0050] 2. Each rice sample is prepared into three categories: the first category, 15 grams of rice; the second category, 15 grams of brown rice; the third category, 45 grams of polished rice; then, the rice, brown rice and polished rice are subjected to near-infrared scanning to obtain corresponding near-infrared original spectral data, 30 grams of the third category of polished rice is weighed for relevant detection of the rice taste meter to obtain the taste measurement value of the taste meter, and the taste value is corrected according to the taste meter correction regression model of the above embodiment.

[0051] 3. The 92 rice materials collected above were used as validation samples for the rice taste near-infrared prediction regression model to evaluate the prediction effect of the method of the present invention on the taste quality indexes of paddy, brown rice and polished rice. First, the taste prediction value of the validation sample was obtained. The specific operation steps were as follows: export the spectrum file, open the software Unscramble X10.4, and then open the corresponding model (there is one model for each of paddy, brown rice and polished rice), and then import the scanned spectrum file to predict the near-infrared taste index respectively.

[0052] Then, it is compared and analyzed with the corresponding taste meter correction value (laboratory value). The determination method can refer to the rice taste meter determination process in the above embodiment. After the verification sample is measured by the rice taste meter, the taste meter correction regression model is used for correction processing. The statistical results are shown in Table 3: Table 3 Statistical results of taste correction values ​​of validation samples

[0053] Figure 5 This is one of the schematic diagrams of the model verification results based on the taste meter correction values ​​of 92 samples and the predicted values ​​of the rice taste near infrared prediction regression model provided by the present invention, Figure 6 The second schematic diagram of the model verification result based on the taste meter correction value of 92 samples and the predicted value of the rice taste near infrared prediction regression model provided by the present invention, Figure 7 The third schematic diagram of the model verification result based on the taste meter correction value of 92 samples and the predicted value of the rice taste near infrared prediction regression model provided by the present invention can be referred to Figure 5 , Figure 6 and Figure 7 As shown, the model validation results of three sample types, including paddy, brown rice and polished rice, are included. The present invention uses the estimated standard error (Standard Error of Prediction, referred to as SEP) as an indicator to measure the size of the model prediction error. SEP is usually used to evaluate the predictive ability of the regression model, and can be used to compare the prediction effects of different models. The smaller the SEP, the higher the accuracy of the model prediction. In the present invention, the SEP of the validation results of the rice taste near-infrared prediction regression model corresponding to paddy, brown rice and polished rice are 2.580, 2.312, and 2.172, respectively, which are all in line with the model prediction error, and the validation effect is good. It can be seen that the predicted values ​​of the constructed rice taste near-infrared prediction regression model corresponding to paddy, brown rice and polished rice are highly consistent with the laboratory measurement values, indicating that the rice taste near-infrared prediction regression model established by the present invention has good accuracy, and it can be widely used in the field of screening high-quality taste new varieties.

[0054] Figure 8 The present invention provides a schematic diagram of the overall process of establishing, verifying and applying the method for rapid detection of rice taste based on near-infrared high-throughput detection. The specific steps are as follows: First, a certain number of representative samples with different taste values ​​are collected, namely, the second sample rice. These samples should be able to reflect the overall level of the taste value of the rice to be tested.

[0055] Next, professionals were organized to conduct manual sensory evaluation of the taste of the samples. At the same time, a rice taste meter was used to measure the taste of the samples. A rice taste meter is an instrument specifically used to measure the taste of rice. It evaluates the comprehensive taste value of rice by measuring indicators such as the hardness and viscosity of the rice.

[0056] Then, according to the artificial sensory evaluation values ​​and the rice taste meter measurement values, the taste value correlation analysis equation was established to construct the taste meter correction regression model.

[0057] While collecting the second sample of rice, more representative rice samples (i.e., the first sample of rice) are collected. These representative samples will be used for subsequent taste value determination and spectrum collection to ensure the accuracy and reliability of the analysis results. The collected representative samples are processed as necessary, such as shelling, to prepare three types of rice (i.e., rice grains, brown rice, and polished rice) for subsequent taste value determination and spectrum collection.

[0058] Furthermore, a near infrared spectrometer is used to collect spectra of representative samples to obtain spectral data of the samples, and a rice taste meter is also used to measure the taste value of the representative samples.

[0059] Furthermore, based on the taste meter correction regression model, the taste values ​​of representative samples are corrected to obtain more accurate and reliable sample taste correction values. The purpose of this step is to eliminate errors and interferences in the measurement process and improve the accuracy of the analysis results.

[0060] Furthermore, based on the correlation analysis results of near-infrared spectral data and sample taste correction values, a near-infrared prediction regression model of rice taste (abbreviated as: PLS model) was established and verified to ensure the accuracy and reliability of the model and provide a basis for subsequent daily analysis.

[0061] According to a rice taste detection method provided by the present invention, after inputting the target spectrum data into the rice taste near-infrared prediction regression model to obtain the taste prediction value corresponding to the rice to be detected output by the rice taste near-infrared prediction regression model, the method further includes: According to the taste prediction value corresponding to the rice to be detected and the target spectral data, the rice taste near-infrared prediction regression model is optimized and updated to obtain an optimized and updated rice taste near-infrared prediction regression model.

[0062] In the present invention, samples (such as samples with low taste scores ≤ 60 points and high taste scores ≥ 90 points, that is, samples at the two ends of the model taste value, this part of the sample collection is highly representative and can enrich the model data) are added to the PLS model for optimization or regular verification of the spectrum and taste value of the sample to be tested, so as to further optimize the model, improve its predictive ability, ensure the long-term effectiveness and accuracy of the model, and provide support for the breeding of high-quality rice taste.

[0063] Finally, the optimized PLS model was applied to daily analysis to quickly detect the taste value of new rice samples. At the same time, outliers (values ​​outside the model range) were marked, and the reasons were analyzed or retested.

[0064] The present invention adopts Unscrambler X10.4 software modeling to build three types of near-infrared high-throughput taste value rapid detection systems for paddy rice, brown rice and polished rice, and is widely used in the taste value detection and analysis of breeding materials of different generations.

[0065] The sensory evaluation of traditional rice requires the participation of 15-20 qualified tasters, and 500 to 700 grams of polished rice need to be prepared. The sample consumption is large, the pre-treatment process is complicated, the evaluation procedure is time-consuming, and it is difficult to measure in batches. It is only suitable for a few high-generation materials, regional trials and field production varieties. The detection process of the rice taste meter also requires the preparation of 30 grams of polished rice, and then weighing, soaking, steaming, cooling, and measuring steps. One process takes at least 3 to 4 hours. Whether it is sensory evaluation or taste meter instrument measurement, the existing rice taste detection takes a lot of time, and it is difficult to quickly and effectively identify the taste value of a large number of samples. It is not suitable for large-scale screening of rice taste breeding materials.

[0066] Under the same conditions, the rice taste near-infrared prediction regression model established by the present invention is used to detect the rice taste, which greatly reduces the labor cost, and the entire test process only takes 15 seconds. At the same time, the present invention overcomes the shortcomings of the traditional taste detection method that requires a sufficient amount of sample for detection, and the existing methods are mostly destructive sampling (the sample needs to be ground into polished rice). The present invention greatly shortens the breeding time and has obvious economic value.

[0067] The near-infrared high-throughput rice taste rapid detection system established by the present invention has the remarkable characteristics of small sample quantity, fast detection, no damage, low cost and high throughput compared with other taste detection methods. It has a good effect in practical application, can realize high-throughput rapid screening of rice taste value, and verification through a large amount of data shows that the correlation coefficients R of the model of paddy, brown rice and polished rice are 0.9009, 0.9239 and 0.9275 respectively, and its accuracy and standard error are as expected, which can replace the traditional method to realize the rapid prediction of taste value index in low-generation rice breeding materials.

[0068] The rice taste detection system provided by the present invention is described below. The rice taste detection system described below and the rice taste detection method described above can be referred to each other.

[0069] Fig. 9 The schematic diagram of the structure of the rice taste detection system provided by the present invention is as follows: Fig. 9As shown, the present invention provides a rice taste detection system, comprising a spectral data acquisition module 901, a model determination module 902 and a taste detection module 903, wherein the spectral data acquisition module 901 is used to obtain target spectral data corresponding to the rice to be detected; the model determination module 902 is used to determine the corresponding rice taste near-infrared prediction regression model according to the sample type of the rice to be detected, wherein the rice taste near-infrared prediction regression model is constructed based on the correspondence between the sample spectral data of sample rice of different sample types and the sample taste meter correction value corresponding to the sample spectral data; the taste detection module 903 is used to input the target spectral data into the rice taste near-infrared prediction regression model to obtain the taste prediction value corresponding to the rice to be detected output by the rice taste near-infrared prediction regression model.

[0070] The rice taste detection system provided by the present invention obtains target spectral data corresponding to the rice to be detected, and determines the corresponding taste near-infrared prediction regression model according to the type of the rice to be detected, and then based on the taste near-infrared prediction regression model constructed by the correspondence between the sample spectral data and the sample taste correction value, uses the target spectral data to predict the taste value, thereby obtaining the rice taste detection result more efficiently and accurately.

[0071] The system provided by the present invention is used to execute the above-mentioned method embodiments. Please refer to the above-mentioned embodiments for the specific processes and detailed contents, which will not be repeated here.

[0072] Fig.10 A schematic diagram of the structure of an electronic device provided by the present invention, such as Fig.10 As shown, the electronic device may include: a processor 1001, a communication interface 1002, a memory 1003 and a communication bus 1004, wherein the processor 1001, the communication interface 1002 and the memory 1003 communicate with each other through the communication bus 1004. The processor 1001 may call the logic instructions in the memory 1003 to execute the rice taste detection method, which includes: obtaining target spectrum data corresponding to the rice to be detected; determining the corresponding rice taste near-infrared prediction regression model according to the sample type of the rice to be detected, wherein the rice taste near-infrared prediction regression model is constructed based on the correspondence between the sample spectrum data of the sample rice of different sample types and the sample taste meter correction value corresponding to the sample spectrum data; inputting the target spectrum data into the rice taste near-infrared prediction regression model to obtain the taste prediction value corresponding to the rice to be detected output by the rice taste near-infrared prediction regression model.

[0073] In addition, the logic instructions in the above-mentioned memory 1003 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc. Various media that can store program codes.

[0074] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the rice taste detection method provided by the above-mentioned methods, and the method includes: obtaining target spectral data corresponding to the rice to be detected; determining a corresponding near-infrared prediction regression model of rice taste according to the sample type of the rice to be detected, wherein the near-infrared prediction regression model of rice taste is constructed based on the correspondence between sample spectral data of sample rice of different sample types and sample taste meter correction values ​​corresponding to the sample spectral data; inputting the target spectral data into the near-infrared prediction regression model of rice taste to obtain the taste prediction value corresponding to the rice to be detected output by the near-infrared prediction regression model of rice taste.

[0075] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, is implemented to execute the rice taste detection method provided in the above-mentioned embodiments, the method comprising: obtaining target spectral data corresponding to the rice to be detected; determining a corresponding near-infrared prediction regression model of rice taste according to the sample type of the rice to be detected, wherein the near-infrared prediction regression model of rice taste is constructed based on the correspondence between sample spectral data of sample rice of different sample types and sample taste meter correction values ​​corresponding to the sample spectral data; inputting the target spectral data into the near-infrared prediction regression model of rice taste to obtain the taste prediction value corresponding to the rice to be detected output by the near-infrared prediction regression model of rice taste.

[0076] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0077] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A rice taste detection method, characterized in that: include: Obtain target spectrum data corresponding to the rice to be detected; According to the sample type of the rice to be tested, a corresponding near-infrared prediction regression model of rice taste is determined, wherein the near-infrared prediction regression model of rice taste is constructed based on the correspondence between sample spectral data of sample rice of different sample types and sample taste meter correction values ​​corresponding to the sample spectral data; Inputting the target spectrum data into the rice taste near infrared prediction regression model to obtain a taste prediction value corresponding to the rice to be tested output by the rice taste near infrared prediction regression model; The rice taste near infrared prediction regression model is constructed by the following steps: Acquire sample spectral data corresponding to first sample rice of different sample types, wherein the sample types include rice type, brown rice type and polished rice type; the sample spectral data is constructed by spectrally preprocessing the original spectral data corresponding to the first sample rice; Obtaining a sample rice taste meter measurement value corresponding to the first sample rice; Based on the taste meter correction regression model, the sample rice taste meter measurement value is corrected to obtain the sample taste meter correction value; wherein the taste meter correction regression model is constructed based on the corresponding relationship between the sensory evaluation value and the rice taste meter measurement value; Based on partial least squares regression, the sample spectral data is associated with the sample taste meter correction value to construct the rice taste near-infrared prediction regression model corresponding to the first sample rice of different sample types; The taste meter correction regression model is constructed by the following steps: Obtaining a sensory evaluation value and a rice taste meter measurement value corresponding to a second rice sample, wherein the second rice sample and the first rice sample are rice of the same variety, mature season and location; The rice taste meter measurement value corresponding to the second rice sample is associated with the sensory evaluation value to establish the taste meter correction regression model, and the taste meter correction regression model is used to correct the rice taste meter measurement value.

2. The rice taste detection method according to claim 1, characterized in that: The method further comprises: The near-infrared diffuse reflectance spectrum of the first sample rice is scanned to obtain the original spectrum data corresponding to the first sample rice.

3. The rice taste detection method according to claim 1, characterized in that: The spectrum preprocessing includes: removing abnormal spectra in the original spectrum data, and performing spectrum data preprocessing on the original spectrum data after the abnormal spectra are removed based on derivative transformation and standard normal variable transformation.

4. The rice taste detection method according to claim 1, characterized in that: After inputting the target spectrum data into the rice taste near infrared prediction regression model to obtain the taste prediction value corresponding to the rice to be tested output by the rice taste near infrared prediction regression model, the method further includes: According to the taste prediction value corresponding to the rice to be detected and the target spectral data, the rice taste near-infrared prediction regression model is optimized and updated to obtain an optimized and updated rice taste near-infrared prediction regression model.

5. A rice taste detection system, characterized in that: include: A spectral data acquisition module is used to obtain target spectral data corresponding to the rice to be tested; A model determination module, for determining a corresponding near-infrared prediction regression model for rice taste according to the sample type of the rice to be detected, wherein the near-infrared prediction regression model for rice taste is constructed based on the correspondence between sample spectral data of sample rice of different sample types and sample taste meter correction values ​​corresponding to the sample spectral data; A taste detection module, used for inputting the target spectrum data into the rice taste near infrared prediction regression model, and obtaining a taste prediction value corresponding to the rice to be detected output by the rice taste near infrared prediction regression model; The rice taste near infrared prediction regression model is constructed by the following steps: Acquire sample spectral data corresponding to first sample rice of different sample types, wherein the sample types include rice type, brown rice type and polished rice type; the sample spectral data is constructed by spectrally preprocessing the original spectral data corresponding to the first sample rice; Obtaining a sample rice taste meter measurement value corresponding to the first sample rice; Based on the taste meter correction regression model, the sample rice taste meter measurement value is corrected to obtain the sample taste meter correction value; wherein the taste meter correction regression model is constructed based on the corresponding relationship between the sensory evaluation value and the rice taste meter measurement value; Based on partial least squares regression, the sample spectral data is associated with the sample taste meter correction value to construct the rice taste near-infrared prediction regression model corresponding to the first sample rice of different sample types; The taste meter correction regression model is constructed by the following steps: Obtaining a sensory evaluation value and a rice taste meter measurement value corresponding to a second rice sample, wherein the second rice sample and the first rice sample are rice of the same variety, mature season and location; The rice taste meter measurement value corresponding to the second rice sample is associated with the sensory evaluation value to establish the taste meter correction regression model, and the taste meter correction regression model is used to correct the rice taste meter measurement value.

6. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the rice taste detection method according to any one of claims 1 to 4 is implemented.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the rice taste detection method according to any one of claims 1 to 4 is implemented.

8. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the rice taste detection method according to any one of claims 1 to 4 is implemented.

Images