Method for managing hydroponic lettuce and predicting yield of hydroponic lettuce

Abstract

The invention discloses a hydroponic lettuce management and yield prediction method, which comprises the steps of seedling raising, strong seedling transplanting, environmental factor management, lettuce yield prediction and the like, seedlings are raised through a self-made seedling raising solution, lettuce growth is cultivated through a cultivation nutrient solution, and meanwhile, factors such as temperature and humidity, illumination and carbon dioxide concentration of ambient air are subjected to combined monitoring and coupling regulation and control. The hydroponic management method is simple and easy to implement, growth and physiology of root systems, stems and leaves of crops are remarkably improved after the lettuce is planted, the chlorophyll and nitrogen content of the leaves is increased, and the yield of the lettuce is increased. The yield prediction model can monitor and predict the yield condition according to growth, is especially suitable for smart lettuce cultivation, and has a certain comprehensive effect on other hydroponic leaf vegetables.

Description

Technical Field

[0001] The present invention belongs to the fields of agricultural planting, facility horticulture cultivation, smart agriculture, etc., and specifically relates to a method for managing hydroponic lettuce and predicting its yield. Background Art

[0002] With the development of technology in the field of facility agriculture in recent years, the facility vegetable industry at home and abroad has gradually developed towards mechanization, digitization and intelligence. my country has a huge market demand for facility vegetables, and the facility vegetable industry has developed rapidly in recent years. China's facility vegetable planting has entered a stage of rapid development. With the improvement of people's awareness of ecological environment and health, the development of the vegetable industry has received more and more attention. At present, domestic facility vegetable planting has transformed from simple increase in production and expansion to diversification and high quality. The planting area has expanded year by year, and the technical level has been continuously improved. At the same time, some emerging formats have also been gradually introduced, bringing more development opportunities to the facility vegetable industry. The production management process of facility lettuce mainly includes four links: fine sowing, factory seedling cultivation, field management, and harvesting and transportation. At present, the form of "Internet of Things + Agriculture" has been vigorously promoted, which has greatly improved agricultural productivity. At present, various sensors are also used to monitor and control environmental factors such as temperature and humidity, CO2 concentration, and light intensity required by crops in real time, such as using PLC controllers and HMI human-machine interfaces to achieve precise control of water and fertilizer integration and harvesting of lettuce.

[0003] At present, excessive nitrogen input in the facility vegetable system has led to a series of problems such as soil salinization and acidification, and increased greenhouse gas emissions. This is mainly due to the excessive annual average nitrogen input of chemical fertilizers, organic fertilizers and irrigation water. The nitrogen surplus and nitrate nitrogen residue in the soil layer are far higher than the demand for crop planting, and the groundwater in the facility vegetable production area is seriously polluted. Nitrogen has a significant impact on the growth, yield and nutritional quality of leafy vegetables, especially lettuce. To a certain extent, with the increase of nitrogen levels, lettuce yield increases. However, after exceeding a certain peak, the yield decreases and the quality is also greatly affected. Obviously, the nitrogen content in the leaves is too high. At present, shortening the lettuce planting process can bring great yields, but the increase in production times also leads to excessive nitrogen input. However, the emergence of hydroponic cultivation mode can greatly reduce soil problems, but because it is not easy to maintain nitrogen form when configuring hydroponic solution, lettuce has a greater demand for ammonium nitrogen, but other environmental factors such as light, carbon dioxide concentration, temperature and humidity also have a great impact on hydroponic liquid nitrogen, which has a great impact on leaf photosynthesis, especially the formation of chlorophyll. Therefore, it is necessary to add appropriate amounts of fertilizers and nutrients to the nutrient solution for hydroponic lettuce, while maintaining appropriate environmental factors such as temperature, humidity, light, and air carbon dioxide concentration. Researchers at home and abroad have shown that different nutrient ratios in hydroponic solution can have a significant impact on growth rate, growth cycle, and yield. Summary of the invention

[0004] Purpose of the invention: The technical problem to be solved by the present invention is to provide a facility cultivation management method for improving the yield and nutrient nitrogen content of hydroponic lettuce in response to the shortcomings of the prior art, solve the problem of the influence of coupled management of multiple environmental factors on the growth of hydroponic lettuce, and realize high-yield and high-quality cultivation of facility lettuce.

[0005] In order to achieve the above-mentioned invention object, the technical solution adopted by the present invention is as follows: A method for managing hydroponic lettuce and predicting its yield comprises the following steps: (1) Seedling cultivation Sow the lettuce seeds evenly and directly in the rock wool, and then put them in a container filled with seedling solution. Keep the rock wool immersed in the seedling solution throughout the seedling period, and maintain the seedling environment temperature at 25-28℃; (2) Transplanting strong seedlings Select strong seedlings, insert the rock wool with seedlings into the matching planting basket first, and then transplant them together into the pipes or pots filled with hydroponic solution in the greenhouse, and keep the average temperature in the cultivation room at about 25-28℃ and 50-60% humidity during the day, and 20±1℃ and 65-70% humidity at night; (3) Environmental factor management Temperature, humidity, light, and air carbon dioxide concentration were controlled during the seedling slowing stage after transplanting, the stable growth stage of leafing, the rapid increase of plant height after the leaf number growth slowed down, the increase of single leaf area and biomass after the rapid increase of plant height, and the harvesting stage. Nutrient solution control: When transplanting seedlings for the first time, add coated slow-release urea fertilizer to the hydroponic solution and dilute it to make a nutrient solution; replace the hydroponic solution every 25-35 days according to the growth of the plants, and keep the remaining coated slow-release urea fertilizer; the hydroponic solution should have a total nitrogen content of ≥3 g / kg, a total phosphorus content of ≥100 mg / kg, a total potassium content of ≥120.0 mg / kg, a pH of 6.0-7.0, a conductivity of 2.0 mS / cm -3.1 mS / cm, and a salinity of 0.1%-0.15%; (4) Lettuce production forecast The yield of lettuce is predicted based on the content of nutrient solution indicators or based on plant growth indicators, and a prediction model for chlorophyll and nitrogen content of hydroponic lettuce during the harvest period is constructed using machine learning methods. The above-mentioned environmental factors are adjusted according to the predicted results of the yield, chlorophyll and nitrogen content of hydroponic lettuce.

[0006] Specifically, in step (1), the seedling solution preparation process is as follows: add rooting powder to distilled water at a volume weight ratio of 0.1-05g / 100ml, and add limestone with a particle size of less than 0.5cm at a weight ratio of 10-15% to the liquid, maintain the temperature at 35-40℃, and place it at a light intensity of 10000 lx-12000 lx for 2-3 days, then place it at -18℃ away from light for 5-10 days, take it out and place it in a refrigerator at 4℃ away from light for 5-10 days; when raising seedlings, take it out and dilute it 8-12 times, then put the rock wool with seeds into it, do not completely immerse it, maintain the liquid level at 0.5-1cm, and keep the rock wool completely moist; maintain the pH of the seedling solution at 6.5-7.5, the conductivity can be 1.3 mS / cm-2.4 mS / cm, and the salinity is 0.1-0.2%.

[0007] Specifically, in step (2), the hydroponic solution is kept to be immersed at 1 / 2 height of the planting basket and 1-2 cm below the lowest position of the pipe planting hole.

[0008] Specifically, in step (3), the temperature and humidity control method is as follows: during the 7-day seedling hardening stage after transplanting, the temperature is controlled at 20°C-25°C and the relative humidity is controlled at 60%-70%; after the seedling hardening stage, the temperature enters the leaf-pulling stable growth stage, the temperature is controlled at 25°C-30°C and the relative humidity is controlled at 70%-80%; after the leaf number growth slows down, the temperature is controlled at 20°C-30°C and the relative humidity is controlled at 60%-80%; after the plant height increases rapidly, the temperature enters the single leaf area and biomass increase stage, the temperature is controlled at 20°C-33°C and the relative humidity is controlled at 70%-85%; after entering the harvestable stage, the temperature is controlled at 28°C-35°C and the relative humidity is controlled at 75%-90%; when all growth stages exceed these ranges, the temperature is raised or lowered, the humidity is increased or water vapor is removed.

[0009] Specifically, in step (3), the light control method is: using a light meter to measure the light intensity in three time periods: 8:00-10:00 in the morning, 12:00-14:00 in the afternoon, and 16:00-18:00 in the evening, once every half an hour; in the 7 days of the seedling acclimatization stage after transplanting, the average light intensity in the three time periods shall not be higher than 6000 lx. If it is lower than 4000 lx, supplementary light must be provided; after the seedlings are acclimatized and enter the leaf-pulling and stable growth stage, the average light intensity in the three time periods shall not be higher than 8000 lx. If it is lower than 6000 lx, supplementary light shall be provided for 3-5 hours after 18:00 in the evening; in the stage of rapid increase in plant height after the growth of leaf number slows down, the average light intensity in the three time periods shall not be higher than 9000 lx. If it is lower than 7000 lx, supplementary light shall be provided for 3-5 hours after 18:00 in the evening; after the rapid increase in plant height, the stage of increasing single leaf area and biomass shall be entered. The average light intensity in the three time periods shall not be higher than 12000 lx. If it is lower than 8000 lx, supplementary light shall be provided for 3-5 hours after 18:00 in the evening. lx, supplementary lighting time is required for 6-8 hours after 18:00 in the evening; in the harvestable stage, the average light intensity in three time periods shall not exceed 8000 lx. If it is lower than 4000lx, supplementary lighting time is required for 1-3 hours after 18:00 in the evening; purple light is used for supplementary lighting in each stage.

[0010] Specifically, in step (3), the method for controlling the concentration of carbon dioxide in the air is as follows: starting real-time monitoring using a CO2 monitoring device after transplanting; during the 7-day seedling hardening stage after transplanting, the CO2 concentration in the air is controlled at 500 ppm-600 ppm; after the seedling hardening stage, when entering the leaf-pulling stable growth stage, the CO2 concentration in the air is controlled at 500 ppm-600 ppm; after the leaf number growth slows down and the plant height increases rapidly, the CO2 concentration in the air is controlled at 600 ppm-800 ppm; after the plant height increases rapidly and enters the stage of increasing single leaf area and biomass, the CO2 concentration in the air is controlled at 850 ppm-1000 ppm; after entering the harvestable stage, the CO2 concentration in the air is controlled at 1000 ppm-1200 ppm.

[0011] Specifically, in step (3), during the first week of seedling growth after transplanting, natural ventilation reaches the CO2 concentration requirement; starting from the second week, CO2 growth agent is applied to increase the carbon dioxide concentration in the air. When the CO2 concentration in the air (C air ) is lower than the minimum value C min When the CO2 gas bottle or dry ice is placed in the middle of the lettuce planting area, the height and distance from the lettuce are H. The CO2 gas coverage height should be within 2H, and the coverage volume should be V (m 3 ) = s*2H; adjust the CO2 gas bottle release rate v to 2-5 mg / min; the required CO2 weight MCO2 and time T are: MCO2 (mg) = [(C max +Cmin ) / 2-C air ]*V; T (min) = MCO2 / v; Among them, the value range of H is 30cm-60cm; Lettuce planting area s<100m 2 ; C min , C max The values ​​are the lowest and highest carbon dioxide concentrations in the air at different growth stages of lettuce, ranging from the seedling stage after transplanting to the stable growth stage of leafing. min , C max 500 ppm and 600 ppm respectively; in the stage of rapid increase in plant height after the growth of leaf number slowed down, C min , C max 600 ppm and 800 ppm respectively; after the rapid increase in plant height, it entered the stage of increasing single leaf area and biomass, C min , C max 850 ppm and 1000 ppm respectively; after entering the harvestable stage, C min , C max 1000 ppm and 1200 ppm respectively.

[0012] Specifically, in the nutrient solution control process of step (3), the N content in the hydroponic solution comes from ammonium chloride, the P comes from diamine, and the potassium comes from the ash washing liquid, which is prepared with distilled water according to the requirements of total nitrogen, total phosphorus, and total potassium. In addition, trace elements such as Mg, Ca, S, Fe, B, Mn, Zn, Cu, and Mo are added, and vitamin B1, B2, or B6 is added according to a weight volume ratio of 1-2 mg / 10000 ml; the ash washing liquid is the remaining filtrate of washing rice and wheat ash, and the upper clear liquid is taken after centrifugation and added to the hydroponic solution at a ratio of 1-2:10; the exogenously added coated slow-release urea fertilizer is a homemade 60-day slow-release urea fertilizer, wherein the nitrogen content is not less than 45%, and its release effect should be that the initial nutrient dissolution rate is not less than 18.5%, and the cumulative nutrient release rate is not less than 50%, 75%, 85%, and 90% on the 14th, 28th, 56th, and 84th days, respectively.

[0013] Specifically, in step (4), the lettuce yield prediction model based on the nutrient solution index content is: Y=29.177+0.196EC1-333.155pH1+.219EC2+235.674pH2-0.116EC3+84.758pH3-0.131EC4+43.829 pH4-0.547N1-0.599N2+0.406N3-0.103N4(R2 =0.747, p <0.0001); Among them, Y is the fresh leaf biomass of a single lettuce plant, in g / plant; EC1, EC2, EC3, EC4 are the conductivity contents when the nutrient solution was replaced from the first to the fourth time, in uS / cm; pH1, pH2, pH3, pH4 are the pH values ​​when the nutrient solution was replaced from the first to the fourth time; N1, N2, N3, N4 are the total nitrogen contents when the nutrient solution was replaced from the first to the fourth time, in mg / L.

[0014] Specifically, in step (4), the lettuce yield prediction model based on plant growth indicators is: Y=212.440+2.345H 2 -0.384L 2 -0.003A 2 -59.852H+15.236L+1.741A (R 2 =919, p <0.001) Y is the fresh leaf biomass of a single lettuce plant, in g / plant; H, L, and A represent plant height, number of leaves, and maximum leaf area, in cm, pieces, and cm, respectively. 2 .

[0015] Specifically, the method for determining the chlorophyll and nitrogen content prediction model of hydroponic lettuce in the present invention is as follows: S1: Take photos Take a photo of the lettuce. The resolution of the photo must be 1080p or higher, and the background must be black. S2: Trimming pictures Read the image from the specified path, as well as the height and width of the image, and crop an area from 10% to 80% of the height in the vertical direction of the original image; S3: Segmentation Create a mask and background and foreground model, use GrabCut algorithm to segment image1, initialize segmentation by specifying a rectangular area, and find known foreground or possible foreground, use connected component analysis to find the largest connected component, equal to 1, unequal to 0, where 1 is foreground and 0 is background, and create the corresponding binary mask, then copy the original image to the connected component, set the part that does not belong to the largest connected component to 0, get an image containing only leaves, then convert it to a grayscale image, apply thresholding method to get a binary image; copy the original image, set the area that does not belong to the leaves in the binary image to 0 in the copied image, get the final segmented leaf image, and add borders at the top and bottom of the image in the vertical direction; finally return the processed leaf image to remove the background of the image; S4: Get the color value of the image Get the mean values ​​of R, G, and B of the leaves in the picture, and fill the calculated data into the corresponding position in the Excel table; S5: Calculate the color vegetation index Determine the independent variables of red, green and blue color vegetation index according to the color vegetation index R, G and B; first obtain the data in the R, G and B columns in the Excel table, and use the above R, G, B and the calculated parameters as the independent variable X, and the index leaf chlorophyll SPAD and leaf N content as the dependent variable Y respectively; S6: Model construction and indicator evaluation determination With the three RGB channel values ​​as independent variables and leaf chlorophyll SPAD value and leaf N content (mg / g) as dependent variables, elastic network, kernel ridge regression, gradient boosting regression GBRT, XGBoost and Light GBM models were constructed respectively, and the best prediction model was determined by calculating the evaluation index R². Beneficial Effects

[0016] (1) In order to shorten the cultivation time of hydroponic lettuce and achieve higher yield, the present invention conducts certain hydroponic experiments to compare the effects of different hydroponic liquid nitrogen levels, carbon dioxide concentrations, light, etc. on the growth and yield of lettuce, and develops a reasonable hydroponic management technology and leaf yield prediction model, which provides a theoretical basis for the reasonable application of nitrogen fertilizer, supplementation of carbon dioxide and light, etc. in facility hydroponic vegetables to obtain high-yield and high-quality lettuce, and also provides a reference for avoiding soil and water pollution caused by excessive application of nitrogen fertilizer.

[0017] (2) The seedling cultivation of the present invention can promote the growth of lettuce seedlings, and the nutrient solution cultivation and environmental management can increase the nitrogen content and yield of lettuce. In addition, the control of environmental temperature and humidity can also prevent the growth of certain pests and diseases and pathogens. The present invention is simple and easy to operate, and can not only increase the yield and quality of purple green vegetables, but also increase its commercial value, and reduce operating costs. The present invention is particularly suitable for factory-scale hydroponic lettuce cultivation, and can also achieve certain beneficial effects in other facility leaf vegetable cultivation. The operation technical process is simple, time-saving and labor-saving, and the facility lettuce is high-yield and high-quality. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] The present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments, and the above and / or other advantages of the present invention will become more clear.

[0019] Figure 1 2 are the temperature, humidity and carbon dioxide concentration conditions under different carbon dioxide treatments in the embodiment.

[0020] Figure 2 1 is a comparison of the picture before and after trimming in the embodiment.

[0021] Figure 3 This is the fitting situation of the picture color value and SPAD value in the embodiment.

[0022] Figure 4 It is the correlation between the predicted value of the SPAD value fitting model and the true value.

[0023] Figure 5 This is the fitting situation of the picture color value and leaf N content.

[0024] Figure 6 It is the correlation between the predicted value of leaf N content by fitting model and the true value. DETAILED DESCRIPTION

[0025] The present invention can be better understood with reference to the following examples.

[0026] This embodiment provides a facility cultivation method of a hydroponic lettuce management method, and the specific steps include: The invention provides a hydroponic management method for lettuce, a seedling raising and transplanting method, and an environmental control management technology, wherein the environmental control management technology mainly includes the management of hydroponic solution, the control management of air environmental factors, and the requirements for light management. Harvesting can begin at the end of the sixth week after transplanting. 1. Experimental setup

[0027] The experiment was carried out in August 2021 (E 118°52′, N 32°2′), using hydroponics to grow lettuce, and the lettuce variety came from the Jiangsu Academy of Agricultural Sciences. Hydroponic tubes were used for planting. Three treatments were set up, including no application of CO2 additives (C0), conventional natural ventilation carbon dioxide (C1), and application of CO2 additives (C2) of the present invention. The CO2 fertilizer used was Quzhou Guosheng brand carbon dioxide growth agent. Three nitrogen levels were set in each CO2 treatment: N0 (no exogenous urea), N3 (1.95 g / L exogenous urea), and N5 (3.25 g / L exogenous urea) were added to the hydroponic solution. A total of 9 treatments were set up, namely C0N0, C0N3, YC0N5, C1N0, C1N3, C1N5, C2N0, C2N3, and C2N5, and each treatment was repeated 6 times. Transplantation was carried out on August 13. Lettuce seedlings with uniform growth were selected, fixed with planting cotton and transplanted into hydroponic pots. Natural ventilation was carried out in the first week of transplantation for treatments C0 and C1, and management was carried out according to Table 1 for treatment C2. Harvesting began in the sixth week after transplantation.

[0028] First, lettuce seeds are evenly and directly sown in rock wool, and then placed in a container containing homemade seedling solution. The rock wool is kept fully containing seedling solution during the entire seedling period. The homemade seedling solution is the original solution, which is added with rooting powder in distilled water according to the volume weight (100ml: 0.1g), and limestone with a particle size of less than 0.5 cm is added to the liquid at 10%-15% by weight. Then keep it at a temperature of 35℃-40℃ and a light intensity of 10000 lux-12000 lux for 3 days. Then place it at -18℃ away from light for 1 week. After taking it out, place it in a refrigerator at 4℃ away from light for 1 week. When raising seedlings, take it out and dilute it about 10 times. Then put the rock wool with seeds in it, not completely immersed (maintaining a liquid level of 1cm), and keep the rock wool completely moist. Keep the pH of the seedling solution at 7.0, the conductivity can be 2.0 mS / cm, and the salinity is about 0.15%. Maintain the seedling environment temperature at 25℃.

[0029] Select strong seedlings for transplanting, insert the rock wool with seedlings into the matching planting basket first, and then transplant them together into the pot device filled with hydroponic solution in the greenhouse, where the hydroponic solution is kept soaking 1 / 2 of the height of the planting basket and 2cm below the lowest position of the pipe planting hole. Keep the average temperature in the cultivation room at about 25-28℃ during the day and the humidity at 55%-60%. The temperature at night is 20℃ and the humidity is 65%-70%. Add 1.95 g of coated slow-release urea fertilizer per liter of homemade hydroponic solution, and then add 500ml of distilled water per liter to dilute it to make a nutrient solution. Replace a homemade hydroponic solution every 30 days according to the growth of the plants, and keep the remaining coated slow-release urea fertilizer. Then the nutrient solution needs to start circulating at 8:00 am and stop circulating at 16:00 pm. The homemade hydroponic solution has a total nitrogen content of ≥3 g / kg, a total phosphorus content of ≥100 mg / kg, a total potassium content of ≥120.0 mg / kg, a pH of 6.0-7.0, a conductivity of 2.0 mS / cm -3.1 mS / cm, and a salinity of about 0.1%-0.15%. The specific method is: the N content in the hydroponic solution comes from ammonium chloride, P comes from diamine, and potassium comes from the ash cleaning solution, which is prepared with distilled water according to the requirements of total nitrogen, total phosphorus, and total potassium. In addition, certain trace elements such as MgO, CaO, S, Fe, B, Mn, Zn, Cu, Mo, etc. are added, and vitamins B1, B2, and B6 are added, according to a weight-volume ratio of 1 mg: 10000 ml. The ash cleaning solution is the remaining filtrate from washing rice and wheat ash. After centrifugation, the upper clear liquid is taken and added to the hydroponic night at a ratio of 1:10. In addition, the coated slow-release urea fertilizer added exogenously was a homemade 60-day slow-release urea fertilizer, which contained 45% nitrogen. According to the national standard, its release effect was 18.5% in the initial nutrient dissolution rate, and the cumulative nutrient release rates on the 3rd, 5th, 7th, 14th, 28th, 42nd, 56th, 84th, and 112th days were 25.89%, 28.48%, 30.50%, 53.10%, 78.10%, 83.94%, 87.83%, 92.59%, and 97.22%.

[0030] Under the C2 treatment, according to the requirements of the lettuce growth stage, the environmental factors management includes the control management of temperature and humidity in the air, CO2 concentration, light, etc. Among them, according to the requirements of the lettuce growth stage, the seedlings are slowed down for 7 days after transplanting, the temperature is controlled at 20℃-25℃, the relative humidity is controlled at 60%-70%, and the CO2 concentration in the air is controlled at 500 ppm-600 ppm. After the seedlings are slowed down, they enter the stable growth stage of leaf extraction (leaf extraction stage, about 5 days), the temperature is controlled at 25℃-30℃, the relative humidity is controlled at 70%-80%, and the CO2 concentration in the air is controlled at 500 ppm-600 ppm. After the growth of the number of leaves slows down, the plant height increases rapidly (about 5 days), the temperature is controlled at 20℃-30℃, the relative humidity is controlled at 60%-80%, and the CO2 concentration in the air is controlled at 600 ppm-800 ppm. After the plant height increases rapidly, it enters the stage of increasing single leaf area and biomass (about 7 days). The temperature is controlled at 20℃-33℃, the relative humidity is controlled at 70%-85%, and the CO2 concentration in the air is controlled at 850 ppm-1000 ppm. After that, it enters the harvestable stage, the temperature is controlled at 28℃-35℃, the relative humidity is controlled at 75%-90%, and the CO2 concentration in the air is controlled at 1000 ppm-1200 ppm. All growth stages that exceed these ranges must be heated or cooled, humidified or water vapor removed, CO2 increased or excess CO2 discharged. Use a light meter to measure the light in the morning (8:00-10:00), middle (12:00-14:00), and evening (16:00-18:00), once every half an hour. After transplanting, the seedlings will be slowed down for 7 days, and the average light in the three time periods should not be higher than 6000 lx. If it is lower than 4000 lx, supplementary light must be provided. After the seedlings have been slowly grown, they enter the stable growth stage of leafing (leafing stage, about 5 days), and the average light intensity in the three time periods is not higher than 8000 lx. If it is lower than 6000 lx, it is necessary to supplement the light for 4 hours after 18:00 in the evening. After the growth of the number of leaves slows down, the plant height increases rapidly (about 5 days), and the average light intensity in the three time periods is not higher than 9000 lx. If it is lower than 7000 lx, it is necessary to supplement the light for 4 hours after 18:00 in the evening. After the rapid increase in plant height, it enters the stage of increasing single leaf area and biomass (about 7 days), and the average light intensity in the three time periods is not higher than 12000 lx. If it is lower than 8000 lx, it is necessary to supplement the light for 6-8 hours after 18:00 in the evening. After that, it enters the harvestable stage, and the average light intensity in the three time periods is not higher than 8000 lx. If it is lower than 4000 lx, it is necessary to supplement the light for 2 hours after 18:00 in the evening. Purple light is used for supplementary lighting.

[0031] Other treatments are managed according to conventional facilities and no supplementary light is used. If the above requirements are not met, the internal and external sunshade nets and wet curtain fans must be turned on to cool and humidify, or the fans must be turned on to exhaust and dehumidify.

[0032] According to the actual situation, use CO2 growth agent to increase the concentration of carbon dioxide in the air, and use CO2 monitoring equipment to monitor in real time. air ) is lower than the lowest value C in the table min When the CO2 solid growth agent is placed in the lettuce planting area (area s<100m 2 ) in the middle, the height and the lettuce distance is H (30cm-60cm), the CO2 gas coverage height should be about 2H, then the coverage volume should be V (m 3 ) = s*2H. Adjust the CO2 gas bottle release rate v to 2-5 mg / min. The required CO2 weight MCO2 and time T are: MCO2 (mg) = [(C max +C min ) / 2-C air ]*V; T (min) = MCO2 / v.

[0033] Among them, the value range of H is 30cm-60cm; Lettuce planting area s<100m 2 ; C min , C max The values ​​are the lowest and highest carbon dioxide concentrations in the air at different growth stages of lettuce, ranging from the seedling stage after transplanting to the stable growth stage of leafing. min , C max 500 ppm and 600 ppm respectively; in the stage of rapid increase in plant height after the growth of leaf number slowed down, C min , C max 600 ppm and 800 ppm respectively; after the rapid increase in plant height, it entered the stage of increasing single leaf area and biomass, C min , C max 850 ppm and 1000 ppm respectively; after entering the harvestable stage, C min , C max 1000 ppm and 1200 ppm respectively. 2. Data collection and analysis

[0034] Use a CO2 monitor (Beijing Green Scale Technology Co., Ltd.) to monitor CO2 concentration, temperature, and humidity in real time on a mobile phone, and use a light meter to measure light intensity. Measure crop plant height and leaf number. After harvesting, use a ruler to measure plant height, leaf number, plant root length, leaf fresh weight (yield), and root fresh weight. After the leaves are dried, weigh the root dry weight and leaf dry weight. After planting, use a chlorophyll meter to measure chlorophyll content, leaf surface temperature, leaf nitrogen content, and leaf surface humidity; use a illuminometer to measure light intensity each time the chlorophyll content is measured, and select fixed leaves with uniform growth and no diseases or insect pests for measurement. Measure the nitrogen content in plant leaves at harvest. The average values ​​of various indicators on some dates are selected in this example to display, as shown in Table 2, Table 3 and Figure 1 To figure. 3. Results

[0035] (1) Environmental monitoring A. Average light intensity Table 1 Average light intensity in the experimental greenhouse (lux)

[0036]

[0037] Note: Since C0, C1, and C2 are located adjacent to each other in the same greenhouse, the light intensity is the same.

[0038] B. Temperature, humidity and air carbon dioxide concentration Figure 1 These are the temperature, humidity and carbon dioxide concentration under different carbon dioxide treatments.

[0039] C. Hydroponic solution indicators (harvest period) Table 2 Conditions before replacement of hydroponic solution at different stages of the whole process (average value)

[0040]

[0041] Note: EC1, EC2, EC3, EC4 are the conductivity contents when the nutrient solution is replaced from the first to the fourth time, and the unit is uS / cm; pH1, pH2, pH3, pH4 are the pH values ​​when the nutrient solution is replaced from the first to the fourth time; N1, N2, N3, N4 are the total nitrogen contents when the nutrient solution is replaced from the first to the fourth time, and the unit is mg / L.

[0042] (2) Growth and yield Table 3 Growth and yield at harvest (average value)

[0043]

[0044] Note: Y is the fresh leaf biomass of a single lettuce plant, in g / plant. H, L, and A represent plant height, number of leaves, and maximum leaf area at harvest, in cm, pieces, and cm, respectively. 2 . R 2 is the model fit goodness of fit, R 2 =0.864 indicates a high degree of fit. p <0.001 indicates that the prediction model has extremely high accuracy and the model indicators are extremely significantly correlated.

[0045] (3) Leaf chlorophyll SPAD value and nitrogen content (harvest period) Table 4 Conditions before replacement of hydroponic solution at different stages of the whole process (average value)

[0046]

[0047] Note: SPAD1, SPAD2, SPAD3, SPAD4, SPAD5 and SPAD6 are the average values ​​of leaf chlorophyll SPAD from the 1st week to the 6th week (dimensionless index, no unit), LN1, LN2, LN3, LN4, LN5 and LN6 are the average values ​​of leaf nitrogen content from the 1st week to the 6th week (in mg / g).

[0048] (4) Production model construction Lettuce yield prediction model based on nutrient solution index content: Y=29.177+0.196EC1-333.155pH1+.219EC2+235.674 pH2-0.116EC3+84.758 pH3-0.131EC4+43.829 pH4-0.547N1-0.599N2+0.406N3-0.103N4(R 2 =0.747, p <0.0001) Among them, Y is the fresh leaf biomass of a single lettuce plant, in g / plant. EC1, EC2, EC3, EC4 are the conductivity contents when the nutrient solution was replaced from the first to the fourth time, in uS / cm; pH1, pH2, pH3, pH4 are the pH values ​​when the nutrient solution was replaced from the first to the fourth time; N1, N2, N3, N4 are the total nitrogen contents when the nutrient solution was replaced from the first to the fourth time, in mg / L. R 2 is the model fit goodness of fit, R 2 =0.747 indicates a high degree of fit. p <0.0001 indicates that the prediction model has extremely high accuracy and the model indicators are extremely significantly correlated.

[0049] Lettuce yield prediction model based on plant growth indicators: Y=212.440+2.345H 2 -0.384L 2 -0.003A 2 -59.852H+15.236L+1.741A (R 2 =919, p <0.001) Y is the fresh leaf biomass of a single lettuce plant, in g / plant. H, L, and A represent plant height, number of leaves, and maximum leaf area at harvest, in cm, pieces, and cm, respectively. 2 . R 2 is the model fit goodness of fit, R 2 =0.864 indicates a high degree of fit. p <0.001 indicates that the prediction model has extremely high accuracy and the model indicators are extremely significantly correlated.

[0050] (5) Machine learning was used to construct a prediction model for leaf chlorophyll SPAD value and nitrogen content at harvest time.

[0051] The prediction of chlorophyll and nitrogen content under the above hydroponic lettuce management uses machine learning methods to build a prediction model for chlorophyll SPAD value and leaf nitrogen content. The specific steps are as follows: Step 1: Take a photo Take a photo of the lettuce. The resolution of the photo must be 1080p or higher, and the background must be black.

[0052] Step 2: Trim the image To trim a picture, the specific steps are: 1) Use image = cv2.imread(path) to read the image from the specified path. path is the input parameter of the function, which indicates the path of the image file.

[0053] 2) Use height1, width1 = image.shape[:2] to get the height and width of the read image.

[0054] 3) Use image1 = image[int(0.1*height1):int(0.8*height1),:] to crop an area of ​​10% to 80% of the height from the original image in the vertical direction to reduce the processed area. The comparison of the above test image before and after trimming is as follows Figure 2 shown.

[0055] Step 3: Segment the image 1) Create a mask and background and foreground models. Use the GrabCut algorithm to segment image1, and initialize the segmentation by specifying a rectangle (based on graph cuts). And use np.where to find known foreground or possible foreground.

[0056] 2) Use cv2.connectedComponents to return num_labels (the number of labeled connected areas), labeled_mask (an image of the same size as the input image), and use connected component analysis to find the largest connected component, np.where(labeled_mask == largest_component_label, 1, 0). If they are equal, they are 1, and if they are not equal, they are 0, that is, 1 is foreground and 0 is background. Create the corresponding binary mask largest_component_mask.

[0057] 3) By copying the original image image1 into largest and setting the parts that do not belong to the largest connected component to 0, we get an image containing only leaves.

[0058] 4) Convert largest to grayscale image gray_largest and apply Otsu’s thresholding method to get binary image thresholded.

[0059] 5) segmented_leaf = image1.copy(): Copy the original image.

[0060] 6) segmented_leaf[thresholded == 0] = 0: Set the areas in the binary image that do not belong to leaves to 0 in the copied image to obtain the final segmented leaf image.

[0061] 7) segmented_leaf = cv2.copyMakeBorder(segmented_leaf, int(0.1*height1), int(0.2*height1), 0,0, cv2.BORDER_CONSTANT, value=(0, 0, 0)): Add borders at the top and bottom of the image in the vertical direction to retain more information.

[0062] 8) Return the processed leaf image segmented_leaf.

[0063] 9) Use rembg.remove in rembg to remove the background of img.

[0064] Step 4: Get the image color value Run main.py to obtain the mean values ​​of R, G, and B of the plants in the picture and fill the calculated data into the corresponding positions in the excel table. Low-resolution images can be converted into high-resolution images for recognition.

[0065] Step 5: Calculate the color vegetation index According to the color vegetation index R, G, B, determine the independent variables of red, green, and blue color vegetation index. Run predict_N.py, first obtain the data in the R, G, and B columns in the Excel table, and use the above R, G, B and the calculated parameters as the independent variable X, the indicator chlorophyll SPAD (unitless, dimensionless indicator), leaf N content (mg / g) as the dependent variable Y, for the training of the following fitting model. And use R 2 as an evaluation indicator.

[0066] Step 6: Model construction and indicator evaluation determination The XGBoost and Light GBM models were constructed with the RGB three channel values ​​as independent variables and the chlorophyll SPAD value or leaf N content as the dependent variable, and the evaluation index R² was calculated to evaluate the quality of the model.

[0067] 1) With the RGB three channel values ​​as independent variables, leaf chlorophyll SPAD value and leaf N content (mg / g) as dependent variables, elastic network, kernel ridge regression, gradient boosting regression GBRT, XGBoost, Light GBM and other fitting models are constructed respectively. The code is shown below, and the evaluation index R² is calculated to evaluate the quality of the model.

[0068] #ElasticNetwork ENet = ElasticNet(alpha=0.01,l1_ratio=0.9,random_state=3) ENet.fit (X,Y) Y_predict1 = np.round(ENet.predict(X),decimals=4) R1 = np.round(r2_score(Y,Y_predict1),4) #Kernel Ridge Regression KRR=KernelRidge(alpha=0.05,kernel='polynomial',degree=2,coef0=0.5) KRR.fit(X,Y) Y_predict2 = np.round(KRR.predict(X),decimals=4) R2 = np.round(r2_score(Y,Y_predict2),4) # Gradient Boosting Regression GBRT GBoost=GradientBoostingRegressor(n_estimators=100,learning_rate=0.03,max_depth=3,max_features='sqrt',loss='huber',random_state=5) GBoost.fit(X, Y) Y_predict3 = np.round(GBoost.predict(X), decimals=4) R3 = np.round(r2_score(Y,Y_predict3),4) # # Random Forest Regression # RF = RandomForestRegressor(n_estimators=100, random_state=5) # RF.fit(X, Y) # Y_predict3 = np.round(RF.predict(X), decimals=4) # R3 = np.round(r2_score(Y,Y_predict3),4) # Extreme Gradient Boosting XGBoost model_xgb=xgb.XGBRegressor(colsample_bytree=0.5,gamma=0.05,learning_rate=0.05,max_depth=4,min_child_weight=1.8, n_estimators=3000,reg_alpha=0.4,reg_lambda=0.8,subsample=0.5,random_state=7,nthread=-1) model_xgb.fit(X, Y) Y_predict4 = np.round(model_xgb.predict(X), decimals=4) R4 = np.round(r2_score(Y,Y_predict4),4) #Lightgbm model_lgb = lgb.LGBMRegressor(objective='regression',num_leaves=3,learning_rate=0.1,n_estimators=2200,max_bin=55,bagging_fraction=0.8, bagging_frep=5,feature_fraction=0.2,feature_fraction_seed=4,bagging_seed=9,min_data_in_leaf=4, min_sum_hessian_in_leaf=4,verbose=-1) model_lgb.fit(X, Y) Y_predict5 = np.round(model_lgb.predict(X), decimals=4) R5 = np.round(r2_score(Y,Y_predict5),4) In the above code, X represents the parameter calculated by the color value extracted by main.py, Y represents the leaf chlorophyll SPAD or leaf N content (mg / g) index, fit(X, Y) means calling the true value of the column where Y is located in Excel, and Y_predict is the fitted prediction model.

[0069] A. Fitting of actual and predicted values ​​of chlorophyll SPAD model Run main.py to get the mean values ​​of R, G, and B of the plants in the picture and fill the calculated data into the corresponding positions in the Excel table. Figure 3 , Figure 4 Shown The scatter plots of the actual and predicted values ​​of elastic network, kernel ridge regression, gradient boosting regression GBRT, XGBoost, and Light GBM are shown in the figure below, from which we can see that after elastic network training and fitting, R 2 They are 0.5165, 0.8992, 0.9979, 0.9916 and 0.9923 respectively.

[0070] B. Fitting of the actual value and predicted value of leaf N content model construction Run main.py to get the mean values ​​of R, G, and B of the plants in the picture and fill the calculated data into the corresponding positions in the Excel table. Figure 5 , Figure 6 shown.

[0071] The scatter plots of the actual and predicted values ​​of elastic network, kernel ridge regression, gradient boosting regression GBRT, XGBoost, and Light GBM are shown in the figure below, from which we can see that after elastic network training and fitting, R 2 They are 0.2455, 0.2546, 0.7153, 0.7705 and 0.7783 respectively.

[0072] The current growth index parameters, chlorophyll SPAD value, and leaf nitrogen content were measured, and the current lettuce yield, chlorophyll SPAD value, and leaf nitrogen content were calculated using the yield, chlorophyll SPAD value, and nitrogen prediction model, thereby obtaining the cultivation condition parameters that need to be adjusted.

[0073] The present invention provides a method and idea for hydroponic lettuce management and yield prediction. There are many methods and approaches to implement the technical solution. The above is only a preferred embodiment of the present invention. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principle of the present invention. These improvements and modifications should also be regarded as the protection scope of the present invention. All components not specified in this embodiment can be implemented by existing technologies.

Claims

1. A method for managing and predicting the yield of hydroponic lettuce, characterized in that: The steps include: (1) Seedling cultivation Sow the lettuce seeds evenly and directly in the rock wool, and then put them in a container filled with seedling solution. Keep the rock wool immersed in the seedling solution throughout the seedling period, and maintain the seedling environment temperature at 25-28℃; (2) Transplanting strong seedlings Select strong seedlings, insert the rock wool with seedlings into the matching planting basket first, and then transplant them together into the pipe or pot device filled with hydroponic solution in the greenhouse. Keep the hydroponic solution immersing 1 / 2 height of the planting basket and 1-2cm below the lowest position of the pipe planting hole. Keep the average temperature in the cultivation room at about 25-28℃ during the day and 50-60% humidity, and the temperature at 20±1℃ and 65-70% humidity at night; (3) Environmental factor management Temperature, humidity, light, and air carbon dioxide concentration were controlled during the seedling slowing stage after transplanting, the stable growth stage of leafing, the rapid increase of plant height after the leaf number growth slowed down, the increase of single leaf area and biomass after the rapid increase of plant height, and the harvesting stage. Nutrient solution control: When transplanting seedlings for the first time, add coated slow-release urea fertilizer to the hydroponic solution and dilute it to make a nutrient solution; replace the hydroponic solution every 25-35 days according to the growth of the plants, and keep the remaining coated slow-release urea fertilizer; the hydroponic solution should have a total nitrogen content of ≥3 g / kg, a total phosphorus content of ≥100 mg / kg, a total potassium content of ≥120.0 mg / kg, a pH of 6.0-7.0, a conductivity of 2.0 mS / cm -3.1 mS / cm, and a salinity of 0.1%-0.15%; (4) Lettuce production forecast The yield of lettuce is predicted based on the content of nutrient solution indicators or based on plant growth indicators, and a prediction model for chlorophyll and nitrogen content of hydroponic lettuce during the harvest period is constructed using machine learning methods. The above-mentioned environmental factors are adjusted according to the predicted results of the yield, chlorophyll and nitrogen content of hydroponic lettuce.

2. The method for managing and predicting the yield of hydroponic lettuce according to claim 1, characterized in that: In step (1), the preparation process of the seedling solution is as follows: adding rooting powder to distilled water at a volume weight ratio of 0.1-05g / 100ml, and adding limestone with a particle size of less than 0.5cm at a weight ratio of 10-15% to the liquid, maintaining the temperature at 35-40°C, and placing it at a light intensity of 10000lx-12000lx for 2-3 days, then placing it at -18°C away from light for 5-10 days, and then placing it in a refrigerator at 4°C away from light for 5-10 days; when raising seedlings, take it out and dilute it 8-12 times, and then put the rock wool with seeds into it, not completely immersing it, maintaining a liquid level height of 0.5-1cm, and keeping the rock wool completely moist; maintaining the pH of the seedling solution at 6.5-7.5, the conductivity can be 1.3 mS / cm-2.4 mS / cm, and the salinity is 0.1-0.2%.

3. The method for managing and predicting the yield of hydroponic lettuce according to claim 1, characterized in that: In step (3), the temperature and humidity control method is as follows: during the 7-day seedling hardening stage after transplanting, the temperature is controlled at 20°C-25°C and the relative humidity is controlled at 60%-70%; after the seedling hardening stage, the temperature enters the leaf-pulling stable growth stage, the temperature is controlled at 25°C-30°C and the relative humidity is controlled at 70%-80%; after the leaf number growth slows down, the temperature is controlled at 20°C-30°C and the relative humidity is controlled at 60%-80%; after the plant height increases rapidly, the temperature enters the single leaf area and biomass increase stage, the temperature is controlled at 20°C-33°C and the relative humidity is controlled at 70%-85%; after entering the harvestable stage, the temperature is controlled at 28°C-35°C and the relative humidity is controlled at 75%-90%; when all growth stages exceed these ranges, the temperature is raised or lowered, the humidity is increased or water vapor is removed.

4. The method for managing and predicting the yield of hydroponic lettuce according to claim 1, characterized in that: In step (3), the light control method is: use a light meter to measure the light intensity in three time periods: 8:00-10:00 in the morning, 12:00-14:00 in the afternoon, and 16:00-18:00 in the evening, once every half an hour; in the 7 days of the seedling acclimatization stage after transplanting, the average light intensity in the three time periods shall not be higher than 6000 lx. If it is lower than 4000 lx, supplementary light must be provided; after the seedlings are acclimatized and enter the leaf-pulling and stable growth stage, the average light intensity in the three time periods shall not be higher than 8000 lx. If it is lower than 6000 lx, supplementary light shall be provided for 3-5 hours after 18:00 in the evening; in the stage of rapid increase in plant height after the growth of leaf number slows down, the average light intensity in the three time periods shall not be higher than 9000 lx. If it is lower than 7000 lx, supplementary light shall be provided for 3-5 hours after 18:00 in the evening; after the rapid increase in plant height, the stage of increasing single leaf area and biomass shall be entered. The average light intensity in the three time periods shall not be higher than 12000 lx. If it is lower than 8000 lx, supplementary light shall be provided for 3-5 hours after 18:00 in the evening. lx, supplementary lighting time is required for 6-8 hours after 18:00 in the evening; in the harvestable stage, the average light intensity in three time periods shall not exceed 8000 lx. If it is lower than 4000 lx, supplementary lighting time is required for 1-3 hours after 18:00 in the evening; purple light is used for supplementary lighting in each stage.

5. The method for managing and predicting the yield of hydroponic lettuce according to claim 1, characterized in that: In step (3), the method for controlling the concentration of carbon dioxide in the air is as follows: starting real-time monitoring using a CO2 monitoring device after transplanting; during the 7-day seedling hardening stage after transplanting, the CO2 concentration in the air is controlled at 500 ppm-600 ppm; after the seedling hardening stage, when the plant enters the leaf-pulling stable growth stage, the CO2 concentration in the air is controlled at 500 ppm-600 ppm; after the leaf number growth slows down and the plant height increases rapidly, the CO2 concentration in the air is controlled at 600 ppm-800 ppm; after the plant height increases rapidly and enters the stage of increasing single leaf area and biomass, the CO2 concentration in the air is controlled at 850 ppm-1000 ppm; after entering the harvestable stage, the CO2 concentration in the air is controlled at 1000 ppm-1200 ppm.

6. The method for managing and predicting the yield of hydroponic lettuce according to claim 5, characterized in that: In step (3), during the first week of seedling acclimatization after transplanting, natural ventilation reaches the CO2 concentration requirement; starting from the second week, CO2 growth agent is applied to increase the carbon dioxide concentration in the air. When the CO2 concentration in the air (C air ) is lower than the minimum value C min When the CO2 gas bottle or dry ice is placed in the middle of the lettuce planting area, the height and distance from the lettuce are H. The CO2 gas coverage height should be within 2H, and the coverage volume should be V (m 3 ) = s*2H; adjust the CO2 gas bottle release rate v to 2-5 mg / min; the required CO2 weight MCO2 and time T are: MCO2 (mg)=[(C max +C min ) / 2-C air ]*V; T (min) = MCO2 / v; Among them, the value range of H is 30cm-60cm; Lettuce planting area s<100m 2 ; C min , C max The values ​​are the lowest and highest carbon dioxide concentrations in the air at different growth stages of lettuce, ranging from the seedling stage after transplanting to the stable growth stage of leafing. min , C max 500 ppm and 600 ppm respectively; in the stage of rapid increase in plant height after the growth of leaf number slowed down, C min , C max 600 ppm and 800 ppm respectively; after the rapid increase in plant height, it entered the stage of increasing single leaf area and biomass, C min , C max 850 ppm and 1000 ppm respectively; after entering the harvestable stage, C min , C max 1000 ppm and 1200 ppm respectively.

7. The method for managing and predicting the yield of hydroponic lettuce according to claim 1, characterized in that: In the process of nutrient solution control in step (3), the N content in the hydroponic solution comes from ammonium chloride, the P comes from diamine, and the potassium comes from the ash washing liquid, which is prepared with distilled water according to the requirements of total nitrogen, total phosphorus and total potassium. In addition, trace elements such as Mg, Ca, S, Fe, B, Mn, Zn, Cu and Mo are added, and vitamin B1, B2 or B6 is added according to a weight volume ratio of 1-2 mg / 10000 ml; the ash washing liquid is the remaining filtrate of washing rice and wheat ash, and the upper clear liquid is taken after centrifugation and added to the hydroponic solution at a ratio of 1-2:10; the exogenously added coated slow-release urea fertilizer is a homemade 60-day slow-release urea fertilizer, wherein the nitrogen content is not less than 45%, and its release effect should be that the initial nutrient dissolution rate is not less than 18.5%, and the cumulative nutrient release rate is not less than 50%, 75%, 85% and 90% on the 14th, 28th, 56th and 84th days, respectively.

8. The method for managing and predicting the yield of hydroponic lettuce according to claim 1, characterized in that: In step (4), the lettuce yield prediction model based on the nutrient solution index content is: Y=29.177+0.196EC1-333.155pH1+.219EC2+235.674pH2-0.116EC3+84.758pH3-0.131EC4+43.829 pH4-0.547N1-0.599N2+0.406N3-0.103N4 (R 2 =0.747, p <0.0001); Among them, Y is the fresh leaf biomass of a single lettuce plant, in g / plant; EC1, EC2, EC3, EC4 are the conductivity contents when the nutrient solution was replaced from the first to the fourth time, in uS / cm; pH1, pH2, pH3, pH4 are the pH values ​​when the nutrient solution was replaced from the first to the fourth time; N1, N2, N3, N4 are the total nitrogen contents when the nutrient solution was replaced from the first to the fourth time, in mg / L.

9. The method for managing and predicting the yield of hydroponic lettuce according to claim 1, characterized in that: In step (4), the lettuce yield prediction model based on plant growth indicators is: Y=212.440+2.345H 2 -0.384L 2 -0.003A 2 -59.852H+15.236L+1.741A (R 2 =919, p <0.001) Y is the fresh leaf biomass of a single lettuce plant, in g / plant; H, L, and A represent plant height, number of leaves, and maximum leaf area, in cm, pieces, and cm, respectively. 2 .

10. The method for managing and predicting the yield of hydroponic lettuce according to claim 1, characterized in that: In step (4), the method for determining the chlorophyll and nitrogen content prediction model of hydroponic lettuce is as follows: S1: Take photos Take a photo of the lettuce. The resolution of the photo must be 1080p or higher, and the background must be black. S2: Trimming pictures Read the image from the specified path, as well as the height and width of the image, and crop an area from 10% to 80% of the height in the vertical direction of the original image; S3: Segmentation Create a mask and background and foreground model, use GrabCut algorithm to segment image1, initialize segmentation by specifying a rectangular area, and find known foreground or possible foreground, use connected component analysis to find the largest connected component, equal to 1, unequal to 0, where 1 is foreground and 0 is background, and create the corresponding binary mask, then copy the original image to the connected component, set the part that does not belong to the largest connected component to 0, get an image containing only leaves, then convert it to a grayscale image, apply thresholding method to get a binary image; copy the original image, set the area that does not belong to the leaves in the binary image to 0 in the copied image, get the final segmented leaf image, and add borders at the top and bottom of the image in the vertical direction; finally return the processed leaf image to remove the background of the image; S4: Get the color value of the image Get the mean values ​​of R, G, and B of the leaves in the picture, and fill the calculated data into the corresponding position in the Excel table; S5: Calculate the color vegetation index Determine the independent variables of red, green and blue color vegetation index according to the color vegetation index R, G and B; first obtain the data in the R, G and B columns in the Excel table, and use the above R, G, B and the calculated parameters as the independent variable X, and the index leaf chlorophyll SPAD and leaf N content as the dependent variable Y respectively; S6: Model construction and indicator evaluation determination With the three RGB channel values ​​as independent variables and leaf chlorophyll SPAD value and leaf N content (mg / g) as dependent variables, elastic network, kernel ridge regression, gradient boosting regression GBRT, XGBoost and Light GBM models were constructed respectively, and the best prediction model was determined by calculating the evaluation index R².

Images