A method for evaluating rice taste value

Abstract

The invention discloses a rice taste value evaluation method, which belongs to the field of food analysis technology. The method collects multispectral images of rice in the differentiation period, heading period and yellow maturity period; screens and processes the multispectral images to obtain a complete multispectral orthophoto; performs image segmentation, selects different rice treatment areas, and calculates the average value of the spectral reflectance of all pixels in the area. The spectral reflectance of different bands is combined, and the reflection characteristics of the multispectral bands are used to obtain the vegetation index through band calculation; four rice quality indicators of experimental materials are measured; the data of the above different periods are combined, and different nitrogen fertilizer treatment gradients and rice quality indicators are combined to construct a model, and the model with the highest Pearson correlation coefficient is selected as the final prediction model. The taste value evaluation method evaluates the taste value of rice through the leaf "ripe color" index, and can achieve non-destructive evaluation of a large number of rice materials before harvest.

Description

Technical Field

[0001] The invention belongs to the technical field of food analysis, and in particular relates to a method for evaluating the taste value of rice. Background Art

[0002] Taste quality is one of the most important traits of rice quality. It refers to the comprehensive evaluation of sensory indicators such as the smell, color, shape, palatability and taste of rice after it is cooked under certain conditions (Ahmed et al., 2019). The main determinants are metabolites such as starch, protein and lipids in rice. Although the regulatory mechanism of rice taste quality is very complex, it is related to the synthesis and accumulation of metabolites in the grains. The substances in the grains are all converted from the photosynthetic products of the leaves; this shows that the formation of rice taste quality is closely related to the synthesis and accumulation of photosynthetic products.

[0003] The current methods for testing taste quality include manual sensory evaluation, instrumental evaluation and physical and chemical index evaluation. All of them require rice harvesting, drying, storage, grinding, steaming and finally instrument testing (or tasting and scoring). The above methods have many processes, are time-consuming (sometimes even more than two months), require special instruments and personnel, have high operating requirements and are inefficient.

[0004] The patent with publication number CN114136918B discloses a rice taste quality evaluation method based on near infrared, but it still requires the selection, shelling, husking and milling of samples to make polished rice, and then the near infrared spectrum of polished rice is associated with the taste value to build a model and then perform the test. The operation process is slow, the pre-treatment process is cumbersome, the sample consumption is large, and the sample needs to be destroyed.

[0005] In the breeding process of high-quality rice in South China, if the taste value of a large number of breeding materials can be evaluated in advance, the efficiency of new rice varieties will be significantly improved. In addition, in the complex practical production process, if the taste value of rice in each field is known in advance, it will be beneficial for farmers to sell at a suitable price and achieve accurate high quality and high price.

[0006] To this end, we proposed a rice taste evaluation method to solve the above problems. Summary of the invention

[0007] The purpose of the present invention is to solve the problem in the prior art that the taste value evaluation needs to be carried out after the rice is mature and the process is complicated and time-consuming, and a method for evaluating the taste value of rice is proposed.

[0008] In order to achieve the above object, the present invention adopts the following technical solutions:

[0009] A rice taste evaluation method comprises the following steps:

[0010] S1: Collect multispectral images of rice at the differentiation stage, heading stage and yellow maturity stage;

[0011] S2: screening the multispectral image obtained in step S1, and processing the screened image to obtain a complete multispectral orthophoto;

[0012] S3: Perform image segmentation on the complete multispectral orthophoto, select different rice treatment areas, and calculate the average spectral reflectance of all pixels in the area as the spectral reflectance data of the rice canopy in the area;

[0013] S4: By combining the spectral reflectance of different bands and utilizing the reflectance characteristics of multi-spectral bands, the vegetation index can be obtained through band calculation;

[0014] S5: After harvest, four rice quality indicators of the experimental materials were measured;

[0015] S6: The data from the above different periods were combined, combined with different nitrogen fertilizer treatment gradients and four rice quality indicators, and the early or late harvest model was constructed using the decision tree regression algorithm. The Pearson correlation coefficient was used as an indicator, and the model with the highest Pearson correlation coefficient was selected as the final prediction model. The prediction results of rice taste quality under different conditions were evaluated through the final prediction model.

[0016] Preferably, the processing of the screened images in step S2 includes point cloud processing, texture processing, geo-registration, seamless stitching and band synthesis.

[0017] Preferably, the vegetation index in step S4 includes normalized vegetation index, green band normalized vegetation index, ratio vegetation index, enhanced vegetation index, difference vegetation index, and improved nonlinear vegetation index.

[0018] Preferably, the calculation method of the enhanced vegetation index is as follows:

[0019] ;

[0020] Among them, NIR is the near-infrared band reflectivity, R is the infrared band reflectivity, B is the blue light band reflectivity, and EVI is the enhanced vegetation index.

[0021] Preferably, the normalized vegetation index is calculated as follows:

[0022] ;

[0023] The ratio vegetation index is calculated as follows:

[0024] ;

[0025] Among them, NDVI is the normalized difference vegetation index, RVI is the ratio vegetation index, NIR is the near-infrared band reflectance, and R is the infrared band reflectance.

[0026] Preferably, the four rice quality indicators include polished rice rate, chalky grain size, crude protein content and taste value.

[0027] Preferably, combining the data of the aforementioned different periods includes combining data of different periods, different spectral bands and different vegetation indices.

[0028] Preferably, the different periods include differentiation period, heading period and yellow maturity period; the different spectral bands include blue light 450 nm, green 560 nm, red light 650 nm, red edge 730 nm, and near infrared 840 nm.

[0029] In summary, the technical effects and advantages of the present invention are as follows: the rice taste evaluation method utilizes the leaf "ripe color" index to evaluate the taste value of rice, and fully considers different light and temperature environments (early and late seasons), different fertilization gradients and other conditions. Compared with the previous rice taste value detection, it has the advantages of fewer processes, convenient operation, high efficiency, non-destructiveness, immediacy and high throughput. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a schematic diagram of the process in the present invention;

[0031] Figure 2 The early rice taste value prediction model in Example 1;

[0032] Figure 3 The late-season rice taste value prediction model in Example 1;

[0033] Figure 4 The crude protein content prediction model of early rice in Example 1;

[0034] Figure 5 The late rice crude protein content prediction model in Example 1;

[0035] Figure 6 The early rice head rice rate prediction model in Example 1;

[0036] Figure 7 The late rice head rice rate prediction model in Example 1;

[0037] Figure 8 The prediction model for the chalky grain rate of early rice in Example 1;

[0038] Fig. 9 This is the prediction model for the chalky grain rate of late-harvested rice in Example 1.

[0039] In the figure: R represents correlation and p represents statistical significance. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present invention will be described clearly and completely below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all the embodiments.

[0041] As shown in the figure, a method for evaluating the taste value of rice comprises the following steps:

[0042] S1: Collect multispectral images of rice at the differentiation stage, heading stage and yellow maturity stage;

[0043] S2: screening the multispectral image obtained in step S1, and processing the screened image to obtain a complete multispectral orthophoto;

[0044] S3: Perform image segmentation on the complete multispectral orthophoto, select different rice treatment areas, and calculate the average spectral reflectance of all pixels in the area as the spectral reflectance data of the rice canopy in the area;

[0045] S4: By combining the spectral reflectance of different bands and utilizing the reflectance characteristics of multi-spectral bands, the vegetation index can be obtained through band calculation;

[0046] S5: After harvest, four rice quality indicators of the experimental materials were measured;

[0047] S6: The data from the above different periods were combined, combined with different nitrogen fertilizer treatment gradients and four rice quality indicators, and the early or late harvest model was constructed using the decision tree regression algorithm. The Pearson correlation coefficient was used as an indicator, and the model with the highest Pearson correlation coefficient was selected as the final prediction model. The prediction results of rice taste quality under different conditions were evaluated through the final prediction model.

[0048] The processing of the filtered images in step S2 includes point cloud processing, texture processing, geo-registration, seamless stitching and band synthesis.

[0049] The vegetation index in step S4 includes normalized vegetation index, green band normalized vegetation index, ratio vegetation index, enhanced vegetation index, difference vegetation index, and improved nonlinear vegetation index.

[0050] The calculation method of the enhanced vegetation index is as follows:

[0051] ;

[0052] in, is the reflectivity in the near-infrared band, is the infrared reflectivity, is the reflectance of blue light band, and EVI is the enhanced vegetation index.

[0053] The normalized difference vegetation index is calculated as follows:

[0054] ;

[0055] The ratio vegetation index is calculated as follows:

[0056] ;

[0057] Among them, NDVI is the normalized vegetation index, RVI is the ratio vegetation index, is the reflectivity in the near-infrared band, is the reflectivity in the infrared band.

[0058] Rice quality indicators include whole rice rate, chalky grain size, crude protein content and taste value.

[0059] Combining the data from the aforementioned different periods includes combining different periods, different spectral bands and different vegetation indices. The different periods include differentiation period, heading period and yellow maturity period; the different spectral bands include blue light 450nm, green 560nm, red light 650nm, red edge 730nm and near infrared 840nm.

[0060] In this embodiment, the experiment was conducted at the Baiyun Base of the Rice Research Institute of Guangdong Academy of Agricultural Sciences in 2023. In the early and late seasons, the high-quality rice varieties of South China, Meixiangzhan No. 2 and Nanjingxiangzhan, were used as materials. The nitrogen application rates were 0 (N0), 60 (N60), 80 (N80), 120 (N120), and 240 kg / hm2. 2 There were five treatments, including (N240), with three replicates for each treatment, randomly arranged, and a plot area of ​​20 m 2 All experiments were conducted by manual transplanting, with a transplanting density of 20 cm × 20 cm, 2 seedlings / hole, and 30 cm high and 25 cm wide ridges between plots to prevent cross-fertilization between plots. Water was independently drained and irrigated, and field management was uniformly implemented.

[0061] The multispectral image of the rice experimental area was collected by using the DJI Phantom 4 multispectral version drone (with 1 visible light sensor and 5 multispectral sensors, including 450nm blue light, 560nm green light, 650nm red light, 730nm red edge, and 840nm near infrared). To improve the imaging quality of the multispectral camera, a clear day was selected from 11:00 to 13:00, and the drone route was planned to pass directly above the target field. The aerial photography height was 20 m, the photo interval was 2s, the heading overlap was 80%, the lateral overlap was 90%, and the image resolution was 6.92 mm / pixel. The detection period was the differentiation period, heading period, and yellow maturity period.

[0062] First, the obtained multispectral images were screened, and the images with spatial location information were imported into PhotoScan software, then point cloud and texture processing were performed, and finally geo-registration, seamless stitching and band synthesis were performed to obtain a complete multispectral orthophoto map (DOM) of the rice field. The ENVI5.3 software ROI (Region of interest) tool was used for image segmentation, and the different rice treatment areas were selected. The average spectral reflectance of all pixels in the area was calculated as the spectral reflectance data of the rice canopy in the plot.

[0063] Vegetation Index (VI) is a mathematical combination of spectral reflectances of different bands. For example, according to the characteristics of MicaSense RedEdge bands, the Normalized Difference Vegetation Index (NDVI), Green Band Normalized Difference Vegetation Index (GNDVI), Ratio Vegetation Index (RVI), Enhanced Vegetation Index (EVI), Difference Vegetation Index (DVI) and Modified Nonlinear Vegetation Index (MNLI) are selected. The above vegetation index can be obtained through band calculation in ENVI software by using the reflectance characteristics of multispectral bands.

[0064] The specific calculation method is as follows:

[0065] Difference Vegetation Index (DVI): ;

[0066] Green Normalized Difference Vegetation Index (GNDVI): ;

[0067] Improved Non-Linear Vegetation Index (MNLI): ;

[0068] Normalized Difference Vegetation Index (NDVI): ;

[0069] Ratio Vegetation Index (RVI): ;

[0070] Enhanced Vegetation Index (EVI): ;

[0071] in, is the reflectivity in the near-infrared band, is the reflectivity in the infrared band, is the reflectivity of green light band, is the reflectivity in the blue light band.

[0072] In this embodiment, the rice taste value determination includes the following steps: after the test material is harvested, dried and stored for 2 months, a sample is taken from replicate I, the rice is milled into polished rice using a small rice mill, and the polished rice is steamed into cooked rice at a ratio of 1:1.2 in terms of rice-water mass ratio. The rice comprehensive taste value, appearance and mouthfeel are determined using a rice taste meter (SATA1B) produced by Japan Satake Co., Ltd., and the average value is taken after 2 repetitions.

[0073] The determination of the polished rice rate and crude protein content includes the following steps: after the rice is harvested, dried and stored for 2 months, the polished rice rate is determined according to the national standard method for testing the polished rice rate of rice (GB / T 21719-2008); the protein content of the rice is determined according to the national food safety standard method for determining protein in food (GB / T 24897-2010).

[0074] The data for each period include 11 columns, including 450nm, 560nm, 650nm, 730nm, 840nm, DVI, GNDVI, MNLI, NDVI, RVI and EVI. 2047 combinations can be formed by permuting and combining them, and 5 nitrogen fertilizer treatment gradients (nitrogen 0kg / hm 2 (N0), 60kg / hm 2 (N60), 80kg / hm 2 (N80), 120kg / hm 2 (N120), 240kg / hm 2 (N240)), as well as four rice quality indices (whole polished rice rate, chalky particle size, crude protein content and taste value). Each indicator was predicted by building a prediction model using the decision tree regression algorithm ExtraTreeRegressor. The Pearson correlation coefficient was used as an indicator, and the model with the highest Pearson correlation coefficient was selected as the final model.

[0075] The technical solutions in the above-mentioned embodiments of the present application have at least the following technical effects or advantages: under different nitrogen fertilizer gradients (N0-240), different varieties and two natural environments (early and late seasons), a model for predicting the taste value, protein content, whole polished rice rate and chalkiness of South China indica rice is established using a drone multispectral system, and the correlations all reach an extremely significant correlation level.

[0076] In terms of taste value prediction, Figure 2 , 3 As shown in the figure, the correlation of the early-season EVI index at the yellow maturity stage to predict the taste value under the fertilization condition of N60-N240 is 0.7, which is extremely significant. The late-season 560nm, 730nm, 840 and EVI indexes at the differentiation and heading stages are used to jointly predict the taste value of rice under the fertilization condition of N0-N240, with a correlation of 0.789, which is extremely significant.

[0077] Protein content is an important physical and chemical indicator of eating quality. Figure 4 , 5 As shown in the figure, the early crop used 560nm, 840nm, MNLI and EVI indexes at the yellow maturity stage to detect the protein content of rice under the condition of fertilization N0-N240, and the correlation was 0.736, which was extremely significant. The late crop used 560nm, 730nm, 840nm and EVI indexes at the differentiation and heading stages to predict the protein content of rice under the condition of nitrogen fertilizer N0-N240, and the correlation was 0.838, which was extremely significant.

[0078] In addition, in terms of the milled rice rate, Figure 6 , 7 As shown in the figure, the early-season rice head rice rate under the N60-N240 condition was predicted by using the 730nm index at the yellow maturity stage, and the correlation was 0.708, which was extremely significant. The late-season rice head rice rate under the N60-N240 condition was detected by using the 560nm and 730nm indexes at the differentiation and heading stages, and the correlation was 0.75, which was extremely significant. Figure 8 , 9 As shown, the early-season rice chalky grain rate under N0-N240 conditions was predicted by using the 840nm index at the yellow maturity stage, and the late-season rice chalky grain rate under N0-N120 conditions was detected by using the 560nm, 730nm, NDVI and RVI indexes at the heading and differentiation stages. The correlation was 0.714, which was extremely significant.

[0079] The working principle is as follows: During the process from heading stage to maturity stage, the leaves of South China Indica rice change from dark green to light yellow and enter the senescence stage. During this process, the leaves of some high-quality rice varieties age slowly (good ripe color), which is conducive to the synthesis and accumulation of metabolites in the grains, improving the taste value and nutritional quality of rice. Therefore, the leaf "ripe color" indicator is used to evaluate the taste quality of rice.

[0080] The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A method for evaluating the taste of rice, characterized in that: The steps include: S1: Collect multispectral images of indica rice in South China at the differentiation stage, heading stage and yellow maturity stage; S2: screening the multispectral image obtained in step S1, and processing the screened image to obtain a complete multispectral orthophoto; S3: Perform image segmentation on the complete multispectral orthophoto, select different rice treatment areas, and calculate the average spectral reflectance of all pixels in the area as the spectral reflectance data of the rice canopy in the area; S4: By combining the spectral reflectance of different bands and utilizing the reflectance characteristics of multi-spectral bands, the vegetation index is obtained through band calculation; S5: After harvest, the rice quality indicators of the experimental materials were measured; S6: The data from the above different periods were combined, combined with different nitrogen fertilizer treatment gradients and four rice quality indicators, and a prediction model was constructed using a decision tree regression algorithm. The Pearson correlation coefficient was used as an indicator, and the model with the highest Pearson correlation coefficient was selected as the final prediction model. The prediction results of rice taste quality under different conditions were evaluated through the final prediction model; Wherein, the rice quality indexes described in step S5 include polished rice rate, chalky grain size, crude protein content and taste value; For early-season rice, the enhanced vegetation index at the yellow-ripening stage is used to predict the taste value, while for late-season rice, the 560nm, 730nm, 840nm and enhanced vegetation index at the differentiation and heading stages are used to predict the taste value.

2. The rice taste evaluation method according to claim 1, characterized in that: The processing of the filtered images in step S2 includes point cloud processing, texture processing, geographic registration, seamless stitching and band synthesis.

3. The rice taste evaluation method according to claim 1, characterized in that: The vegetation index in step S4 includes normalized vegetation index, green band normalized vegetation index, ratio vegetation index, enhanced vegetation index, difference vegetation index, and improved nonlinear vegetation index.

4. The rice taste evaluation method according to claim 3, characterized in that: The calculation method of the enhanced vegetation index is as follows: ， in, is the reflectivity in the near-infrared band, is the infrared reflectivity, is the reflectance of blue light band, and EVI is the enhanced vegetation index.

5. The rice taste evaluation method according to claim 3, characterized in that: The normalized difference vegetation index is calculated as follows: ， The ratio vegetation index is calculated as follows: ， Among them, NDVI is the normalized vegetation index, RVI is the ratio vegetation index, is the reflectivity in the near-infrared band, is the reflectivity in the infrared band.

6. The rice taste evaluation method according to claim 1, characterized in that: The combining of the data from different periods in step S6 includes combining data from different periods, different spectral bands and different vegetation indices.

7. The rice taste evaluation method according to claim 6, characterized in that: The different periods include differentiation period, heading period and yellow maturity period; the different spectral bands include blue light 450 nm, green 560 nm, red light 650 nm, red edge 730 nm, and near infrared 840 nm.

Images