Nutrient and water intelligent regulation method for mango planting

Abstract

The present application relates to the nutrient and water intelligent control method of mango planting, belongs to the agricultural planting technical field. In view of the problems of low utilization rate of fertilization and irrigation, subjective experience and poor timeliness of leaf nutrient and water monitoring in the prior art, and waste of water and fertilizer resources, the method is that the fertilization area is located in the inner side of the irrigation area, the high-position camera is used to vertically shoot the leaf, and the green channel mean and threshold value are extracted based on HSV color segmentation to determine nitrogen deficiency, combined with Canny edge detection to calculate the leaf thickness and threshold value to determine water deficiency, and trigger the atomizer to spray 0.5-3% urea solution 20-30ml / m2 leaf or start the drip irrigation pipeline to irrigate 10-15L / m2 rhizosphere area for 30-45 minutes. The method realizes the water and fertilizer synergistic regulation of mango planting, and is mainly used for improving the nutrient and water utilization efficiency and management automation level of mango garden.

Description

Technical Field

[0001] This invention belongs to the field of intelligent agricultural planting technology. More specifically, this invention relates to a method for intelligent regulation of nutrients and water in mango planting. Background Technology

[0002] In the field of nutrient and water management in mango cultivation, traditional techniques have several prominent problems affecting planting efficiency and resource utilization. Firstly, regarding fertilization zone delineation, conventional methods often employ full-coverage fertilization centered on the trunk or empirical ring application, lacking a targeted design based on the root distribution characteristics of mango trees. Studies have shown that mango absorbing roots are mainly distributed along the edge of the canopy projection, while traditional fertilization zones are often concentrated in sparsely rooted areas closer to the trunk, affecting fertilizer utilization.

[0003] Secondly, in terms of nutrient status monitoring, current technologies mainly rely on manual visual observation of leaf color changes or soil sampling analysis. The former is significantly affected by environmental light conditions and the observer's subjective judgment, exhibiting a strong lag in recognizing early signs of nitrogen deficiency, often indicating that plants have already entered a moderate nitrogen deficiency state by the time these signs are visible to the naked eye. The latter suffers from long detection cycles (usually requiring 3-5 days) and the inability to provide real-time feedback. Effective methods for assessing leaf water content are even more lacking. Conventional leaf curling observation methods are insufficiently sensitive in the early stages of drought, while destructive detection methods such as weighing detached leaves are clearly unsuitable for continuous monitoring. These limitations of monitoring methods directly lead to inaccurate timing of water and fertilizer regulation, becoming a bottleneck restricting refined management.

[0004] The main difficulties encountered in implementing this technology are concentrated in the following two aspects: First, vision-based nutrient diagnosis technology is susceptible to environmental interference; image color stability is poor under natural light conditions, and factors such as leaf surface reflection and dust adhesion can affect the accuracy of color feature extraction. Second, the matching of water and fertilizer synergistic regulation parameters is complex, and the dynamic coupling mechanism between irrigation water volume and soil moisture transport rate still needs in-depth research. These technical challenges make it difficult for existing methods to achieve precise and adaptive regulation, thus hindering the improvement of mango planting and management levels. Summary of the Invention

[0005] One object of the present invention is to address at least the aforementioned deficiencies and to provide at least the advantages described below.

[0006] One purpose of this invention is to provide an intelligent regulation method for nutrients and water in mango cultivation, thereby improving the efficiency of nutrient and water use and the level of automation in mango orchard management.

[0007] Existing technologies for monitoring leaf nutrients and water rely on manual experience, resulting in poor timeliness and strong subjectivity, making it difficult to detect early nitrogen deficiency (delayed leaf yellowing) and water deficiency (insufficient sensitivity of leaf curling method). Furthermore, existing technologies struggle to achieve dynamic, synergistic regulation of water and fertilizer, leading to low management efficiency. This invention addresses these issues by dividing the canopy projection radius into fertilization zones (90-100%) and irrigation zones (100-110%). It combines leaf photography with HSV color segmentation (nitrogen deficiency) and Canny edge detection (water deficiency) for assessment, and utilizes automated response control via atomized spraying and drip irrigation to resolve monitoring delays and reliance on manual intervention.

[0008] This invention addresses the problem of image stability under light interference by using existing visual monitoring technologies. These technologies are significantly affected by fluctuations in ambient light (such as strong light reflection and noise on cloudy days), leading to high errors in leaf color analysis. Furthermore, fixed thresholds cannot adapt to the varying nitrogen requirements of mangoes at different growth stages (e.g., high nitrogen requirement during flower bud formation and low nitrogen requirement during ripening), easily causing misjudgments. The invention utilizes an external light sensor on the camera to dynamically adjust the exposure time (1 / 500-1 / 100 second) and aperture (f / 2.8-f / 8), combined with staged green channel thresholds (130 for flower bud formation to 90 for ripening). This approach matches the physiological characteristics of phenological stages, avoiding misjudgments caused by a "one-size-fits-all" threshold.

[0009] This invention provides a method for intelligent regulation of nutrients and water in mango cultivation, comprising: Based on the canopy projection diameter of the mango tree, the root zone is divided into a ring-shaped fertilization zone and a ring-shaped irrigation zone centered on the trunk, with the fertilization zone located inside the irrigation zone and the irrigation zone equipped with drip irrigation pipelines. Images of mango tree leaves are captured by a camera mounted on a field support frame. The camera is 1-1.5 meters above the mango tree and the shooting angle is vertically downward. The average green channel value of the leaves is extracted using the HSV color segmentation algorithm. When the average green channel value is lower than the average green channel value threshold, it is determined to be nitrogen deficiency. At the same time, the thickness of the leaves is calculated using the Canny edge detection algorithm. When the thickness is lower than the leaf thickness threshold, it is determined to be water deficiency. Based on the nitrogen deficiency determination result, the atomizer is controlled to spray a urea solution with a mass concentration of 0.5-3% onto the leaves, and the spraying amount of the urea solution is 20-30 ml per square meter of leaf area each time. The drip irrigation system is activated based on the water shortage determination result. The irrigation volume of the drip irrigation system is 10-15 liters per square meter of rhizosphere area, and the activation time of the drip irrigation system is 30-45 minutes each time.

[0010] Preferably, the camera has an external light intensity sensor, and the camera dynamically adjusts the exposure time to 1 / 500 second to 1 / 100 second and the aperture size to f / 2.8 to f / 8 according to the real-time measured light intensity; the camera's shooting position avoids the reflective area of ​​the leaf surface; the camera's white balance mode is set to automatic mode, and the green channel threshold in the HSV color segmentation algorithm is calibrated using data collected by an ambient temperature and humidity sensor. During the growth period of mango trees, the average threshold of the green channel for nitrogen deficiency determination is set in segments according to the growth stage: the average threshold of the green channel during the flower bud differentiation period is 120-130; the average threshold of the green channel during the flowering period is 110-120; the average threshold of the green channel during the fruit enlargement period is 100-110; and the average threshold of the green channel during the ripening period is 90-100.

[0011] Preferably, before extracting the green channel mean of the leaf using the HSV color segmentation algorithm, the reflective areas on the leaf surface are identified and marked by an image processing algorithm. The criteria for determining the reflective areas are that the pixel brightness value is higher than 200 and the saturation is lower than 30. When calculating the green channel mean, pixel data within the reflective area are excluded; The camera is set to shoot 1-2 hours after sunrise or 1-2 hours before sunset, and the ambient light intensity is 5000-10000 lux during shooting.

[0012] Preferably, the data collected by the ambient temperature and humidity sensor is used to calibrate the mean threshold of the green channel for determining nitrogen deficiency in the HSV color segmentation algorithm, specifically including the following steps: The ambient temperature (T) and humidity (H) values ​​are acquired using an ambient temperature and humidity sensor. The average threshold G of the green channel for nitrogen deficiency determination is dynamically adjusted according to a preset linear relationship formula. G=G 固定 −0.5×(T−25)+0.3×(H−60), Among them, T ranges from 20-35℃, and H ranges from 50-80%. G 固定 Using the baseline threshold, the flower bud differentiation period G is set in segments according to the growth stages. 固定 The value is 130, and the flowering period is G. 固定 The value is 120, and the fruit enlargement period is G. 固定 The value is 110, and the maturity level is G. 固定 The value is 100. In the HSV color segmentation algorithm, the leaf area is first extracted based on the hue range of 36-86°, the saturation range of 20-80%, and the brightness range of 40-90%, and then the interference area of ​​new leaves with hue of 42-75°, saturation <40%, and brightness >50% is further excluded. Calculate the mean value of the green channel within the leaf area, and determine nitrogen deficiency when the mean value is lower than the calibrated threshold G; When T exceeds the range of 20-35℃ or H exceeds the range of 50-80%, directly use G. 固定 As a threshold for judgment.

[0013] Preferably, the ambient light intensity L, temperature T, and humidity H are collected in real time using an external light intensity sensor and an ambient temperature and humidity sensor. The leaf thickness threshold D for determining water shortage in the Canny edge detection algorithm is dynamically adjusted based on environmental parameters. The adjustment formula is: D = 0.3 + 0.02 × (T − 25) − 0.01 × (H − 60) − 0.0001 × (L − 7500). Among them, T ranges from 20 to 35°C, H ranges from 50 to 80%, and L ranges from 5000 to 10000 lux; When the calculated leaf thickness is lower than the corrected threshold D, it is determined to be water shortage, and the drip irrigation pipeline is turned on. If T, H, or L exceeds the specified range, a fixed threshold D = 0.3 mm is used for determination.

[0014] Preferably, it also includes: Spraying and irrigation interval control: During the flower bud differentiation period, flowering period, and fruit enlargement period, after each application of urea solution based on the nitrogen deficiency assessment result or on rainy days, lock the start command of the atomizer until the interval reaches 12-24 hours. Each time the drip irrigation system is turned on based on the water shortage assessment or on a rainy day, the irrigation command is locked until the interval reaches 12-24 hours. If a nitrogen or water deficiency is triggered again during the lockout period, only the number of times the deficiency is detected will be recorded, but no spraying or irrigation operations will be performed. Root fertilizer application reminder mechanism: The number of foliar fertilizer applications for each mango tree is counted. When nitrogen deficiency is detected for three consecutive times and the cumulative number of applications reaches five, a root fertilizer application reminder signal is generated for the fertilization area. The reminder signal includes the location marker of the fertilization area and the recommended fertilization amount. The recommended fertilization amount is dynamically set according to the current growth stage: 0.5-1.0 kg of root fertilizer per square meter of fertilization area during the flower bud differentiation period; 1.0-1.5 kg of root fertilizer per square meter of fertilization area during the flowering period; 1.5-2.0 kg of root fertilizer per square meter of fertilization area during the fruit enlargement period; and 1.0-1.5 kg of root fertilizer per square meter of fertilization area during the ripening period.

[0015] Preferably, the root fertilizer comprises 98% compound fertilizer, 1% superabsorbent resin and 1% boron, wherein the compound fertilizer has a nitrogen, phosphorus and potassium mass ratio of 2:1:1; when fertilizing, dig a trench 15-20 cm deep in the fertilization area, apply the root fertilizer evenly and then cover it with soil.

[0016] Preferably, the root fertilizer is in granular form, with a water-responsive slow-release coating on the surface of the fertilizer granules. The slow-release coating is prepared by mixing starch and polyvinyl alcohol in a mass ratio of 3:1 and then mixing it with a crosslinking agent to form a coating solution, which is then uniformly coated by a spraying process. The coating thickness is 0.2 mm ± 0.02 mm. The amount of crosslinking agent added is 5%-8% of the total mass of starch and polyvinyl alcohol, and the swelling rate of the coating is controlled within the range of 150%-200%.

[0017] Preferably, the method for preparing the crosslinking agent includes the following steps: 1) Weigh out glycerol and citric acid in a 1:1 mass ratio; 2) Add glycerol and citric acid to the reaction vessel and stir at 50-60℃ for 200-300 rpm for 30-40 minutes to obtain a uniform and transparent liquid crosslinking agent; 3) After cooling the liquid crosslinking agent to room temperature, seal and store it. When using, add it to the starch-polyvinyl alcohol mixture at 5%-8% of the total mass of starch-polyvinyl alcohol.

[0018] The present invention has at least the following beneficial effects: The method of this invention optimizes the distribution of water and fertilizer by dividing fertilization and irrigation areas, thereby reducing resource waste; it uses image analysis technology to determine nitrogen and water deficiency in real time, improving the timeliness of the determination; and it reduces the need for manual intervention and achieves automated management by controlling spraying and drip irrigation, thus reducing labor intensity.

[0019] This invention reduces reflective pixel data by dynamically excluding it, combining light intensity and shooting time window, and utilizing natural light conditions to reduce reflective interference, thus avoiding additional equipment costs. It solves the problem of localized overexposure or color distortion in images caused by leaf surface reflection, which interferes with the accuracy of nitrogen deficiency determination, thereby improving determination precision and reducing the false positive rate.

[0020] This invention distinguishes the differences in nitrogen requirements during the flower bud differentiation period, flowering period, fruit enlargement period, and ripening period by using a dynamic benchmark threshold (G), avoiding misjudgment caused by a fixed threshold. It not only solves the problem of environmental fluctuations but also matches the physiological characteristics of different phenological stages. In addition, it adds filtering conditions of hue 42-75°, saturation <40%, and brightness >50% to eliminate the interference of new leaves on the average value of the green channel.

[0021] The method of this invention reduces the impact of environmental interference on the judgment by dynamically adjusting the green channel threshold and leaf thickness threshold, thereby improving the accuracy of nitrogen and water deficiency identification. Staged threshold matching is used to match the physiological characteristics of mangoes at different growth stages to avoid misjudgments. Water and fertilizer synergistic regulation parameters optimize irrigation volume, reduce resource waste, and improve the level of management automation.

[0022] The method of this invention achieves dynamic coupling between fertilizer release and soil moisture through a water-responsive coating, improving the spatial synergy efficiency of water and fertilizer. The slow-release properties of the coating extend the fertilizer's effective period, reducing the need for frequent fertilization. The root fertilizer formula incorporates water-retaining components to enhance soil water-holding capacity and adapt to the nutrient requirements of mangoes at different growth stages.

[0023] Other advantages, objectives and features of the present invention will become apparent in part from the following description, and in part from those skilled in the art through study and practice of the invention. Attached Figure Description

[0024] Figure 1 This is a schematic diagram of the hardware installation implementation of the intelligent control method of the present invention. Among them, there is a light intensity sensor 1; a camera 2; a protective cover 3; a temperature and humidity sensor 4; a nozzle 5; a bracket 6; a fertilization area 7; and an irrigation area 8. Detailed Implementation

[0025] The present invention will be further described in detail below with reference to embodiments, so that those skilled in the art can implement it based on the description.

[0026] It should be noted that, unless otherwise specified, the experimental methods described in the following embodiments are conventional methods, and the reagents and materials described are commercially available. In the description of this invention, it should be noted that, unless otherwise explicitly specified and limited, the terms "installation," "connection," and "setting" should be interpreted broadly. For example, they can refer to fixed connection or setting, detachable connection or setting, or integral connection or setting. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances. The terms "lateral," "longitudinal," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this invention and simplifying the description. They do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0027] Figure 1 This demonstrates the implementation of an intelligent nutrient and water regulation method for mango cultivation.

[0028] Example 1: A method for intelligent regulation of nutrients and water in mango cultivation, including: Based on the projected diameter of the mango tree canopy, the root zone is divided into a ring-shaped fertilization zone 7 and a ring-shaped irrigation zone 8 centered on the trunk. The radius of the fertilization zone 7 is 90-100% of the canopy projection radius (that is, the outer radius of the ring-shaped fertilization zone 7 is 1.0 times the canopy projection radius, and the inner radius is 1.1 times the canopy projection radius). The radius of the irrigation zone 8 is 100-110% of the canopy projection radius (that is, the outer radius of the ring-shaped irrigation zone 8 is 1.1 times the canopy projection radius, and the inner radius is 1.0 times the canopy projection radius). Drip irrigation pipes are arranged in the irrigation zone 8. Images of mango tree leaves are captured by camera 2 mounted on a field support 6. The camera 2 is 1.5 meters higher than the mango tree and the shooting angle is vertically downward. The image resolution is 20 megapixels. The average green channel value of the leaves is extracted using the HSV color segmentation algorithm. When the average green channel value is lower than the green channel average value threshold of 100, it is determined to be nitrogen deficiency. At the same time, the thickness of the leaves is calculated using the Canny edge detection algorithm. When the thickness is lower than the leaf thickness threshold of 0.3 mm, it is determined to be water deficiency. Based on the nitrogen deficiency determination results, the atomizer is controlled to spray the leaves with a urea solution of 0.5-3% by mass. The amount of urea solution sprayed is 30 ml per square meter of leaf area each time. During the flower bud differentiation and flowering period, the leaves are sprayed with a 3% urea solution by mass, and during the fruit enlargement period, the leaves are sprayed with a 0.5% urea solution by mass. No spraying is done during the fruit ripening period. The drip irrigation system is activated based on the water shortage determination result. The irrigation volume of the drip irrigation system is 15 liters per square meter of rhizosphere area, and the activation time of the drip irrigation system is 30-45 minutes each time.

[0029] The camera 2 has an exposure time of 1 / 100 second and an aperture of f / 2.8; f is the aperture number (f-number), which is used to represent the size of the light-transmitting aperture of the camera 2.

[0030] Also includes: Spraying and irrigation interval control: During the flower bud differentiation period, flowering period, and fruit enlargement period, after each application of urea solution based on the nitrogen deficiency assessment result or on rainy days, lock the start command of the atomizer until the interval reaches 24 hours; during the fruit ripening period, based on the nitrogen deficiency assessment result, the nitrogen deficiency assessment directly generates 7 fertilizer application reminder signals for the fertilization area. Each time the drip irrigation system is turned on based on the water shortage assessment or on a rainy day, the irrigation command is locked until the interval reaches 24 hours. If a nitrogen or water deficiency is triggered again during the lockout period, only the number of times the deficiency is detected will be recorded, but no spraying or irrigation operations will be performed. Root fertilizer application reminder mechanism: The number of foliar fertilizer applications for each mango tree is counted. When nitrogen deficiency is detected for three consecutive times and the cumulative number of applications reaches five, a fertilizer application reminder signal is generated for seven trees in the fertilization area. The reminder signal includes the location marker of fertilization zone 7 and the recommended fertilization amount, which is dynamically set according to the current growth stage: 1.0 kg of root fertilizer per square meter of fertilization zone 7 during the flower bud differentiation period; 1.5 kg of root fertilizer per square meter of fertilization zone 7 during the flowering period; 2.0 kg of root fertilizer per square meter of fertilization zone 7 during the fruit enlargement period; and 1.5 kg of root fertilizer per square meter of fertilization zone 7 during the ripening period. The root fertilizer comprises 98% compound fertilizer, 1% superabsorbent resin, and 1.0% boron, wherein the compound fertilizer has a nitrogen, phosphorus, and potassium mass ratio of 2:1:1. When fertilizing, dig a trench 15-20 cm deep in fertilization zone 7, apply the root fertilizer evenly, and then cover with soil.

[0031] During implementation: Camera 2 can be erected at intervals on pre-installed field supports 6 in the mango orchard, below the protective cover 3. The support 6 can be 1.5 meters above the top of the mango tree to ensure that camera 2 is shooting vertically downwards at the leaves. The atomizer nozzle 5 can be installed on the support 6 1 meter above the tree canopy. The nozzle 5 can be a rotatable nozzle. The lens of camera 2 can be fixed downwards to avoid image distortion due to angular deviation. The field support 6 can be set 2-3 meters away from the main trunk of the mango tree to ensure that the shooting range covers the target leaf area. The drip irrigation pipeline can be laid along the ring irrigation area 8, buried at a depth of about 5-10 cm. The dripper spacing can be set to 30-50 cm, evenly distributed in an area with a radius equal to 100-110% of the tree canopy projection radius. The fertilization area 7 is located inside the drip irrigation pipeline, with a ring area radius of 90-100% of the tree canopy projection radius and a trench depth of 15-20 cm.

[0032] Leaf images captured by camera 2 can be analyzed using the image processing module. The green channel mean threshold is set to 100; if the calculated value is below this threshold, it is determined to be nitrogen deficiency. Leaf thickness is calculated using Canny edge detection, with a thickness threshold set to 0.3 mm; if it is below this value, it is determined to be water deficiency. The urea solution application rate can be controlled at 30 ml per square meter of leaf area. The spray atomizer can be installed on a movable bracket 6 1 meter above the tree canopy. The drip irrigation system provides 15 liters per square meter of rhizosphere area per irrigation cycle, with an operating time of 30-45 minutes. The exposure time can be set to 1 / 100 second, with an aperture of f / 2.8.

[0033] For compound fertilizer, a commercially available general-purpose compound fertilizer with a nitrogen-phosphorus-potassium ratio of 2:1:1 can be selected. For superabsorbent polymer (SAP) resin, sodium polyacrylate products can be selected. For boron, borax or boric acid can be selected. Root fertilizer should be mixed with 98% compound fertilizer, 1% SAP resin, and 1% boron by weight and applied to fertilization area 7. The lockout time for spraying and irrigation can be set as follows: repeated operation is prohibited within 24 hours after spraying, and restarting is prohibited within 24 hours after drip irrigation. When nitrogen deficiency is detected three times consecutively and a total of five sprays are applied, the system will generate a root fertilizer application reminder signal, and the recommended amount will be dynamically adjusted according to the growth stage.

[0034] The method of this invention optimizes the distribution of water and fertilizer by dividing the fertilization and irrigation zones into eight areas, thereby reducing resource waste; it uses image analysis technology to determine nitrogen and water deficiency in real time, improving the timeliness of the determination; and it reduces the need for manual intervention and achieves automated management by controlling spraying and drip irrigation, thereby reducing labor intensity.

[0035] Example 2: A method for intelligent regulation of nutrients and water in mango cultivation, including: Based on the canopy projection diameter of the mango tree, the root zone is divided into a ring-shaped fertilization zone 7 and a ring-shaped irrigation zone 8 centered on the trunk. The radius of the fertilization zone 7 is 90-100% of the canopy projection radius, and the radius of the irrigation zone 8 is 100-110% of the canopy projection radius. Drip irrigation pipelines are arranged in the irrigation zone 8. Images of mango tree leaves are captured by camera 2 mounted on a field support 6. The camera 2 is 1.5 meters higher than the mango tree and the shooting angle is vertically downward. The image resolution is 20 megapixels. The average green channel value of the leaves is extracted using the HSV color segmentation algorithm. When the average green channel value is lower than the average green channel value threshold, it is determined to be nitrogen deficiency. At the same time, the thickness of the leaves is calculated using the Canny edge detection algorithm. When the thickness is lower than the leaf thickness threshold, it is determined to be water deficiency. Based on the nitrogen deficiency determination results, the atomizer is controlled to spray the leaves with a urea solution of 0.5-3% by mass. The amount of urea solution sprayed is 30 ml per square meter of leaf area each time. During the flower bud differentiation and flowering period, the leaves are sprayed with a 3% urea solution by mass, and during the fruit enlargement period, the leaves are sprayed with a 0.5% urea solution by mass. No spraying is done during the fruit ripening period. The drip irrigation system is activated based on the water shortage determination result. The irrigation volume of the drip irrigation system is 15 liters per square meter of rhizosphere area, and the activation time of the drip irrigation system is 30-45 minutes each time.

[0036] Also includes: Spraying and irrigation interval control: During the flower bud differentiation period, flowering period, and fruit enlargement period, after each application of urea solution based on the nitrogen deficiency assessment result or on rainy days, lock the start command of the atomizer until the interval reaches 24 hours; during the fruit ripening period, based on the nitrogen deficiency assessment result, the nitrogen deficiency assessment directly generates 7 fertilizer application reminder signals for the fertilization area. Each time the drip irrigation system is turned on based on the water shortage assessment or on a rainy day, the irrigation command is locked until the interval reaches 24 hours. If a nitrogen or water deficiency is triggered again during the lockout period, only the number of times the deficiency is detected will be recorded, but no spraying or irrigation operations will be performed. Root fertilizer application reminder mechanism: The number of foliar fertilizer applications for each mango tree is counted. When nitrogen deficiency is detected for three consecutive times and the cumulative number of applications reaches five, a fertilizer application reminder signal is generated for seven trees in the fertilization area. The reminder signal includes the location marker of fertilization zone 7 and the recommended fertilization amount, which is dynamically set according to the current growth stage: 1.0 kg of root fertilizer per square meter of fertilization zone 7 during the flower bud differentiation period; 1.5 kg of root fertilizer per square meter of fertilization zone 7 during the flowering period; 2.0 kg of root fertilizer per square meter of fertilization zone 7 during the fruit enlargement period; and 1.5 kg of root fertilizer per square meter of fertilization zone 7 during the ripening period. The root fertilizer comprises 98% compound fertilizer, 1% superabsorbent resin, and 1.0% boron, wherein the compound fertilizer has a nitrogen, phosphorus, and potassium mass ratio of 2:1:1. When fertilizing, dig a trench 15-20 cm deep in fertilization zone 7, apply the root fertilizer evenly, and then cover with soil.

[0037] The camera 2 has an external light intensity sensor 1, and the camera 2 dynamically adjusts the exposure time from 1 / 500 second to 1 / 100 second and the aperture from f / 2.8 to f / 8 according to the real-time measured light intensity. The white balance mode of the camera 2 is set to automatic mode, and the green channel threshold in the HSV color segmentation algorithm is calibrated using data collected by the ambient temperature and humidity sensor 4. During the mango tree's growth period, the green channel mean threshold for nitrogen deficiency determination is set in segments according to the growth stage: The average threshold value of the green channel during flower bud differentiation is 130; the average threshold value of the green channel during flowering is 120; the average threshold value of the green channel during fruit enlargement is 110; and the average threshold value of the green channel during ripening is 100. The camera 2 is positioned to avoid the reflective areas of the leaf surface, and the vertical distance error between it and the mango leaf does not exceed ±5 cm.

[0038] The specific method by which the camera 2 dynamically adjusts the exposure time and aperture size based on the real-time measured light intensity is as follows: The light intensity L is divided into the following ranges, and corresponding exposure time and aperture values ​​are set: High light range (7500≤L≤10000 lux), exposure time is set to 1 / 500 to 1 / 250 second, and aperture is adjusted to f / 8 to f / 5.6; Medium light range (6000≤L<7500 lux), exposure time is set to 1 / 250 to 1 / 125 second, and aperture is adjusted to f / 5.6 to f / 4; Low light range (5000≤L<6000 lux), exposure time is set to 1 / 125 to 1 / 100 second, and aperture is adjusted to f / 4 to f / 2.8; When L<5000 lux or L>10000 lux, fixed parameters are used: exposure time 1 / 100 second, aperture f / 2.8, and a warning signal is triggered to indicate abnormal light.

[0039] Based on typical light experiment data from mango fields, the range of natural light variation from sunrise to sunset was covered by 5000-10000 lux. Segmented thresholds were used to balance the camera's exposure triangle (ISO / aperture / shutter speed) to ensure clear leaf image details. In high light conditions, a combination of short exposure time (1 / 500 second) and small aperture (f / 8) suppressed strong light reflection and prevented overexposure of reflective areas on the leaf surface. In low light conditions, a combination of long exposure time (1 / 100 second) and large aperture (f / 2.8) increased light intake and reduced image noise. In medium light conditions, a compromise was made to balance depth of field and motion blur control.

[0040] The camera 2 avoids the reflective area on the leaf surface by shooting in the following way: before extracting the average green channel value of the leaf using the HSV color segmentation algorithm, the reflective area on the leaf surface is identified and marked by the image processing algorithm. The criteria for determining the reflective area are that the pixel brightness value is higher than 200 and the saturation is lower than 30. When calculating the green channel mean, pixel data within the reflective area are excluded; The shooting time of the camera 2 is set to 1-2 hours after sunrise or 1-2 hours before sunset, and the ambient light intensity is 5000-10000 lux during shooting.

[0041] This invention reduces reflective pixel data by dynamically excluding it, combining light intensity and shooting time window, and utilizing natural light conditions to reduce reflective interference, thus avoiding additional equipment costs. It solves the problem of localized overexposure or color distortion in images caused by leaf surface reflection, which interferes with the accuracy of nitrogen deficiency determination, thereby improving determination precision and reducing the false positive rate.

[0042] The data collected by the ambient temperature and humidity sensor 4 is used to calibrate the mean threshold of the green channel for determining nitrogen deficiency in the HSV color segmentation algorithm, specifically including the following steps: The ambient temperature value T (in °C) and ambient humidity value H (in %) are acquired by the ambient temperature and humidity sensor 4. The mean threshold G of the green channel for nitrogen deficiency determination is dynamically adjusted according to a preset linear relationship formula. G=G 固定 −0.5×(T−25)+0.3×(H−60), Among them, T ranges from 20-35℃, and H ranges from 50-80%. G 固定 Using the baseline threshold, the flower bud differentiation period G is set in segments according to the growth stages. 固定 The value is 130, and the flowering period is G. 固定 The value is 120, and the fruit enlargement period is G. 固定 The value is 110, and the maturity level is G. 固定 The value is 100. In the HSV color segmentation algorithm, the leaf area is first extracted based on the hue range of 36-86°, the saturation range of 20-80%, and the brightness range of 40-90%, and then the interference area of ​​new leaves with hue of 42-75°, saturation <40%, and brightness >50% is further excluded. Calculate the mean value of the green channel within the leaf area, and determine nitrogen deficiency when the mean value is lower than the calibrated threshold G; When T exceeds the range of 20-35℃ or H exceeds the range of 50-80%, directly use G. 固定 As a threshold for judgment.

[0043] Formula design basis and definition: The temperature compensation term (-0.5×(T-25)) was used in controlled experiments to find that for every 1℃ increase in temperature (based on 25℃), chlorophyll decomposition in mango leaves accelerated, and the average green channel value decreased by approximately 0.5. For example, during the flower bud differentiation period (G... 固定 =130), if the ambient temperature T=30℃, the calibrated threshold G=130-0.5×(30-25)=127.5, to avoid misjudging nitrogen deficiency due to yellowing of leaves caused by high temperature. According to physiological basis, high temperature promotes transpiration, leaves close stomata, chloroplast activity decreases, and the intensity of the green band in the reflectance spectrum weakens. Humidity compensation term (+0.3×(H-60)), for every 1% increase in humidity (based on 60%), leaf surface water increases, light reflectance increases, and the average green channel value increases by about 0.3. For example, at maturity (G 固定At a humidity level of 100, if the ambient humidity H = 70%, the calibrated threshold G = 100 + 0.3 × (70 - 60) = 103, avoiding the underestimation of nitrogen deficiency due to darker leaf color in high humidity environments. Based on optical principles, high humidity reduces dust on the leaf surface, enhances specular reflection, and makes the green channel value closer to the true chlorophyll content. The HSV parameters are: Hue 36-86°, covering the typical hues of mango leaves from yellowish-green (immature) to dark green (mature), excluding interference from soil (brown, hue < 36°) and sky (blue, hue > 86°); Saturation > 20%, filtering out gray-white reflective areas (such as low-saturation pixels caused by dew or dust); Brightness > 40%, excluding interference from shadow areas (low-brightness pixels); and new leaf interference is eliminated by further filtering new leaf areas with hue 42-75°, saturation < 40%, and brightness > 50%. New leaves typically exhibit a light green color (yellowish hue, 42-75°), low saturation (<40%), and high brightness (>50%), which is significantly different from the spectral characteristics of mature leaves. This helps to avoid misjudgment caused by new leaves lowering the mean of the green channel.

[0044] Flower bud differentiation stage (G 固定 =130), this stage requires high nitrogen to support flower bud formation, the leaf chlorophyll content is the highest, and the average threshold of the green channel is correspondingly increased; flowering period (G 固定 =120), flowering consumes a large amount of nutrients, the leaves are slightly lighter in color, and the threshold is moderately lowered; during the fruit enlargement period (G 固定 =110), nutrients are transferred to the fruit, the nitrogen demand of leaves decreases, and the threshold is further lowered; at maturity (G 固定 =100), leaf physiological activity weakens, color naturally leans towards yellowish-green, and threshold is the lowest.

[0045] This invention distinguishes the differences in nitrogen requirements during the flower bud differentiation period, flowering period, fruit enlargement period, and ripening period by using a dynamic benchmark threshold (G), avoiding misjudgment caused by a fixed threshold. It not only solves the problem of environmental fluctuations but also matches the physiological characteristics of different phenological stages. In addition, it adds filtering conditions of hue 42-75°, saturation <40%, and brightness >50% to eliminate the interference of new leaves on the average value of the green channel.

[0046] Among them, the ambient light intensity L (in lux), temperature T (in °C) and humidity H (in %) are collected in real time through the external light intensity sensor 1 and ambient temperature and humidity sensor 4 of the camera 2. The leaf thickness threshold D for determining water shortage in the Canny edge detection algorithm is dynamically adjusted based on environmental parameters. The adjustment formula is: D = 0.3 + 0.02 × (T − 25) − 0.01 × (H − 60) − 0.0001 × (L − 7500). Among them, T ranges from 20 to 35°C, H ranges from 50 to 80%, and L ranges from 5000 to 10000 lux; When the calculated leaf thickness is lower than the corrected threshold D, it is determined to be water shortage, and the drip irrigation pipeline is turned on. If T, H, or L exceeds the specified range, a fixed threshold D = 0.3 mm is used for determination.

[0047] Formula design basis and definitions: Temperature compensation term (0.02×(T-25)): Based on the fact that for every 1℃ increase in temperature (based on 25℃), the thickness of mango leaves naturally decreases by approximately 0.02 mm due to increased transpiration. For example, at 30℃, the threshold D increases by 0.1 mm, i.e., (0.02×5), to avoid misjudging water shortage due to the natural thinning of leaves at high temperatures. When T<20℃ or T>35℃, the physiological activity of leaves tends to be abnormal or dormant, exceeding the effective range of the linear compensation model, so a fixed threshold is used. Humidity compensation term (-0.01×(H-60)): Based on the fact that for every 1% decrease in humidity (based on 60%), the thickness of leaves decreases by approximately 0.01 mm due to accelerated transpiration. For example, at 50% humidity, the threshold D decreases by 0.1 mm, i.e., (-0.01×(-10)), to avoid ignoring true water shortage in low humidity environments. When H < 50% or H > 80%, extreme humidity may induce non-physiological thickness changes (such as cell rupture or stomatal closure), so a fixed threshold is used. The light compensation term (-0.0001 × (L - 7500)) is based on the fact that for every 1000 lux increase in light intensity (with 7500 lux as a baseline), the slight shrinkage of the leaf caused by light intensity results in a thickness reduction of approximately 0.1 mm (0.0001 × 1000). For example, at 10000 lux of light, the threshold D is reduced by 0.25 mm to prevent misjudgments caused by optical measurement errors under strong light. When L < 5000 or L > 10000 lux, natural light conditions exceed the normal range and may introduce uncontrollable interference; therefore, a fixed threshold is used.

[0048] This invention improves the accuracy of threshold correction by quantifying the cumulative effects of temperature, humidity, and light on leaf thickness. It automatically optimizes the threshold within a normal environmental range (T: 20-35℃, H: 50-80%, L: 5000-10000 lux) and automatically switches to a fixed threshold under extreme conditions, ensuring system robustness. This effectively reduces accidental irrigation caused by environmental interference, saves water resources, avoids the negative impact of water stress on fruit tree growth, reuses sensor data, eliminates the need for additional hardware, and reduces system complexity and maintenance costs.

[0049] During implementation: The ambient temperature and humidity sensor 4, light intensity sensor 1, and camera 2 can all be installed below the protective cover 3. The entire system can be installed on multiple mango trees at intervals, or for sampling purposes. The protective cover 3 is fixed to the top of the field support 6. The field support 6 can be set 2-3 meters away from the main trunk of the mango tree, with its height 1.5 meters above the top of the tree. The lens of camera 2 points vertically downwards. The atomizer nozzle 5 can be installed on the support 6 1 meter above the tree canopy. The nozzle 5 can be a rotatable nozzle. The drip irrigation pipeline is laid along the ring irrigation area 8, buried 5-10 cm deep, with dripper spacing of 30-50 cm, evenly distributed within an area covering 100-110% of the tree canopy's projected radius. The fertilization area 7 is located inside the drip irrigation pipeline, with a ring area radius of 90-100% of the tree canopy's projected radius and a trench depth of 15-20 cm.

[0050] Leaf images captured by camera 2 are analyzed by the image processing module. The green channel mean threshold is dynamically set according to the mango's growth stages: 130 during flower bud differentiation, 120 during flowering, 110 during fruit enlargement, and 100 during ripening. Environmental temperature and humidity sensor 4 collects atmospheric temperature T (°C) and humidity H (%) in real time. The formula for calibrating the green channel threshold G is G=G 固定 −0.5×(T−25)+0.3×(H−60), where G 固定 Values ​​are taken according to the growth stage. Light intensity sensor 1 measures the light intensity L (in lux) in real time, dynamically adjusting the exposure time and aperture size of camera 2: 1 / 500 sec to 1 / 250 sec and f / 8 to f / 5.6 for high light range (7500-10000 lux); 1 / 250 sec to 1 / 125 sec and f / 5.6 to f / 4 for medium light range (6000-7500 lux); and 1 / 125 sec to 1 / 100 sec and f / 4 to f / 2.8 for low light range (5000-6000 lux). When outside the range, a fixed parameter of 1 / 100 sec and f / 2.8 is used, and a warning signal is triggered.

[0051] The compound fertilizer can be a commercially available general-purpose compound fertilizer with a nitrogen-phosphorus-potassium ratio of 2:1:1. The superabsorbent polymer (SAP) resin can be sodium polyacrylate, and the boron element can be borax or boric acid. Root fertilizer should be applied to fertilization area 7 after mixing 98% compound fertilizer, 1% SAP resin, and 1% boron element by weight. The leaf thickness threshold D is dynamically corrected using the formula D=0.3+0.02×(T−25)−0.01×(H−60)−0.0001×(L−7500). A fixed threshold of 0.3 mm is used when T, H, and L exceed 20-35℃, 50-80%, and 5000-10000 lux, respectively. The spraying and irrigation lockout time is set as follows: repeated operation is prohibited within 24 hours after spraying, and restarting is prohibited within 24 hours after drip irrigation.

[0052] The method of this invention reduces the impact of environmental interference on the judgment by dynamically adjusting the green channel threshold and leaf thickness threshold, thereby improving the accuracy of nitrogen and water deficiency identification. The sensor is installed below the protective cover 3 to reduce weather interference and lower maintenance costs. Staged threshold matching is used to match the physiological characteristics of mangoes at different growth stages, avoiding misjudgments. Water and fertilizer synergistic regulation parameters optimize irrigation volume, reducing resource waste and improving the level of management automation.

[0053] Example 3: Example 3 is the same as Example 2, except that the root fertilizer in Example 3 is coated with a water-responsive slow-release film. Specifically, the surface of the root fertilizer granules is coated with a water-responsive slow-release film. The slow-release film is prepared by mixing starch and polyvinyl alcohol at a mass ratio of 3:1 and then preparing a coating solution with a crosslinking agent. The solution is then uniformly coated by a spraying process, and the film thickness is 0.2 mm ± 0.02 mm. The amount of crosslinking agent added is 5% of the total mass of starch and polyvinyl alcohol, which is used to control the swelling rate of the film within a range of 150%, so as to realize the dynamic change of fertilizer release rate with soil moisture content.

[0054] The preparation method of the crosslinking agent includes the following steps: 1) Weigh out glycerol and citric acid in a 1:1 mass ratio; 2) Add glycerol and citric acid to the reaction vessel and stir at 50°C for 30 minutes at a stirring rate of 200 rpm to obtain a uniform and transparent liquid crosslinking agent; 3) After cooling the liquid crosslinking agent to room temperature, seal and store it. When using, add it to the starch-polyvinyl alcohol mixture at 5%-8% of the total mass of starch-polyvinyl alcohol.

[0055] The process of coating fertilizer granules with a moisture-responsive slow-release coating includes the following steps: Compound fertilizer granules with a particle size of 2-4 mm are placed in a fluidized bed coating machine, and an airflow at a temperature of 50℃ is introduced to suspend the granules. A coating solution was prepared by mixing starch-polyvinyl alcohol mixture with crosslinking agent in a certain proportion, and then sprayed onto the surface of particles through a nozzle at a rate of 0.2 ml / s. During the spraying process, the airflow speed is controlled at 3 m / s and the spraying pressure is 0.1 MPa to ensure that the coating liquid evenly covers the particle surface. After spraying, cure for 10 minutes under 50℃ hot air to form a coating layer with a thickness of 0.2 mm ± 0.02 mm.

[0056] During implementation: Root fertilizer granules can be coated using a fluidized bed coating machine, which can be installed in the fertilizer processing workshop and connected to the fertilizer mixing equipment. Starch and polyvinyl alcohol are mixed at a mass ratio of 3:1, and a crosslinking agent is added to prepare the coating solution. The crosslinking agent is prepared by stirring glycerol and citric acid at a mass ratio of 1:1 at 50-60℃ for 30-40 minutes. The coating solution is sprayed through a nozzle onto the surface of compound fertilizer granules with a particle size of 2-4 mm at a spraying rate of 0.2-0.4 ml / s, a spraying pressure of 0.1-0.3 MPa, and an airflow velocity of 3-5 m / s. After spraying, the granules are cured under hot air at 50-60℃ for 10-15 minutes, forming a coating layer with a thickness of 0.2 mm ± 0.02 mm.

[0057] Slow-release coated root fertilizer can be applied to the ring-shaped fertilization zone 7 of mango trees. The radius of zone 7 should be 90-100% of the tree canopy projection radius, and the trench depth should be 15-20 cm. Apply the root fertilizer evenly into the trench according to the recommended amount for each growth stage, then cover with soil: 71.0 kg per square meter during flower bud differentiation, 1.5 kg during flowering, 2.0 kg during fruit enlargement, and 1.5 kg during ripening. For the compound fertilizer, a commercially available general-purpose product with a nitrogen-phosphorus-potassium ratio of 2:1:1 can be used. For the superabsorbent resin, sodium polyacrylate can be used, and for boron, borax or boric acid can be used. For the coating starch, corn starch can be used, and for polyvinyl alcohol, industrial grade 1788 can be used. The crosslinking agent should be added at 5% of the total mass of starch and polyvinyl alcohol.

[0058] When soil moisture content increases, the swelling rate of the coated root fertilizer increases to 150%, and pore expansion accelerates fertilizer release; when soil moisture decreases, the swelling rate decreases, and pore contraction slows down release. During irrigation, water penetrates into the fertilization zone 7, triggering coating swelling and achieving simultaneous water and fertilizer release. The cross-linking agent regulates the coating swelling rate, ensuring that the fertilizer release rate matches the dynamics of soil moisture and reducing nutrient diffusion loss.

[0059] The method of this invention achieves dynamic coupling between fertilizer release and soil moisture through a water-responsive coating, improving the spatial synergy efficiency of water and fertilizer. The slow-release properties of the coating extend the fertilizer's effective period, reducing the need for frequent fertilization. The root fertilizer formula incorporates water-retaining components to enhance soil water-holding capacity and adapt to the nutrient requirements of mangoes at different growth stages.

[0060] Comparative Example: The comparative example is the same as Example 3, except that the positions of the annular fertilization area and the annular irrigation area are swapped, so that the annular irrigation area is located inside the annular fertilization area.

[0061] Test subjects: Select orchard areas with the same mango variety (such as Tainong No. 1), trees all 5 years old, and consistent planting density, soil fertility (pH 6.0-6.5, organic matter content ≥2%), and basic management.

[0062] Exclude plants with severe pests and diseases or abnormal canopy morphology to ensure that the samples are in similar physiological states.

[0063] 150 mango trees were randomly divided into five groups (30 trees in each group). There were no significant differences among the groups in the average canopy projection diameter, initial leaf nitrogen content (3.2%-3.5%), and soil moisture content (18%-20%) (P>0.05).

[0064] Both the conventional group and the example group adopted the same basic management measures (such as pruning and weeding), only the water and fertilizer control methods were different.

[0065] Water and fertilizer utilization rate is defined as the ratio of total water consumption (liters) per unit yield (kg) to total fertilizer application (kg nitrogen, phosphorus, potassium). Water and fertilizer utilization rate = Total yield (kg) / Total water and fertilizer input (liters or kg).

[0066] Measurement methods: Water volume is recorded by water meter to measure the total amount of irrigation, minus rainfall (weather station data); fertilizer volume is recorded by the type and quality of each application of fertilizer, and calculated based on the effective components of nitrogen, phosphorus and potassium; yield is the total weight of fruit per plant at harvest time (excluding deformed fruit).

[0067] Nitrogen deficiency: Leaf samples (mature leaves in the middle of the canopy) were collected weekly, and the total nitrogen content was determined by the Kjeldahl method. The laboratory results (critical value: 2.8%-3.0%) were used as the baseline, and the number of incorrect judgments was multiplied by the total number of judgments.

[0068] Water shortage: Rhizosphere water content is measured daily using a soil moisture sensor (buried 20 cm deep). The soil moisture sensor reading (field capacity <50% is considered true water shortage) is used as the baseline. The percentage of incorrect readings / total readings is multiplied by 100%. Cost-benefit composition: Equipment costs: Depreciation costs for camera 2, sensors, drip irrigation system, controller, etc. (amortized over 5 years).

[0069] Water and fertilizer costs: Irrigation water fees, compound fertilizer, urea, and slow-release root fertilizer purchase costs.

[0070] Labor costs: labor costs for manual inspection, fertilization, and irrigation in the traditional group; labor costs for equipment maintenance in the example group.

[0071] Total cost: equipment depreciation (amortized over 5 years) + water and fertilizer purchase + labor costs.

[0072] Revenue item: The selling price of mangoes is calculated based on the market average price of 8 yuan / kg.

[0073] The following are the statistical data on water and fertilizer management effects and yield of the four groups of mango trees (traditional group, Example 1 group, Example 2 group, Example 3 group, and comparative group) during the 12-month comparative trial: Water and fertilizer utilization rate: The traditional group requires 8.2 units of water and fertilizer resources to produce 1 kg of mangoes, with a water and fertilizer utilization rate of 1:8.2. Flood irrigation and excessive fertilization result in the lowest utilization rate.

[0074] In Example 1, the water-fertilizer utilization rate was increased to 1:6.5 through an automated drip irrigation and spraying system, but it was still limited by a fixed threshold.

[0075] After introducing a dynamic environmental compensation model in Example 2, the utilization rate was further optimized to 1:4.3.

[0076] Example 3 combines water-responsive slow-release root fertilizer, which improves the efficiency of water and fertilizer synergistic release and achieves a utilization rate of 1:3.1.

[0077] The fertilization area of ​​the comparative group was located outside the irrigation area, which easily led to the outward diffusion and loss of nutrients, affecting the utilization rate. The utilization rate was calculated to be 1:4.0.

[0078] False positive rate: The misjudgment rates of nitrogen deficiency and water deficiency in the traditional group were 31.2% and 27.4%, respectively, mainly because early stress is difficult to identify with the naked eye.

[0079] In Example 1, the false positive rate dropped to 17.5% (nitrogen deficiency) and 15.6% (water deficiency), but the fixed threshold was easily affected by light.

[0080] In Example 2, through dynamic threshold correction (temperature, humidity, and light compensation), the misjudgment rate was significantly reduced to 7.9% (nitrogen deficiency) and 6.3% (water deficiency).

[0081] The misjudgment rate of Group 3 in Example 3 was the same as that of Group 2 in Example 2 (7.4% nitrogen deficiency, 6.8% water deficiency), indicating that the sustained-release coating did not interfere with the judgment.

[0082] The misjudgment rate of the comparative group was the same as that of the Example 3 group (7.5% nitrogen deficiency, 6.6% water deficiency).

[0083] Costs and benefits: The traditional group has the lowest total cost per plant (85 yuan), but labor costs account for as much as 45%, and the net profit is 63 yuan per plant.

[0084] In Example 1, the equipment investment and misjudgment increased to 120 yuan per plant, and the net profit was 63.2 yuan per plant.

[0085] In Example 2, due to sensor and algorithm optimization, the total cost increased to 135 yuan per plant, while the increased yield led to a revenue of 78.6 yuan per plant.

[0086] Example 3 had the highest total cost (150 yuan / plant), but the slow-release root fertilizer reduced the frequency of fertilization, and the yield per plant reached 29.4 kg (±2.0), increasing the net profit to 85.2 yuan / plant.

[0087] The control group had the highest total cost (156 yuan / plant), but its net profit increased to 74.4 yuan / plant.

[0088] Yield and fruit quality: The average yield per plant in the traditional group was 18.5 kg (±1.2 kg), and the average fruit weight was 320 g (±25 g).

[0089] In Example 1, the average yield per plant increased to 22.9 kg (±1.5 kg), with an average weight of 335 g (±20 g).

[0090] In Example 2, the average yield per plant reached 26.7 kg (±1.8 kg), with an average weight of 350 g (±18 g).

[0091] In Example 3, the highest average yield per plant was 29.4 kg (±2.0 kg), with an average weight of 365 g (±15 g). The fruit uniformity was improved by 10%, which also helped to increase the market price.

[0092] The average yield per plant in the control group was 28.8 kg (±2.2 kg), with an average weight of 364 g (±14 g).

[0093] The experimental results show that the traditional group suffered from the lowest water evaporation loss and fertilizer surface runoff due to flood irrigation, resulting in the lowest utilization rate. Furthermore, the traditional group had a high rate of human error, affecting the timeliness of water and fertilizer replenishment and increasing labor costs.

[0094] Although the net profit of Example 1 group was not significantly increased compared with the traditional group, the automation control could significantly reduce labor intensity and alleviate the labor burden on farmers.

[0095] In Examples 2 and 3, dynamic threshold compensation for temperature (T), humidity (H), and light (L) interference resulted in high consistency between the judgment results and laboratory results, improving the accuracy of nitrogen and water deficiency identification, meeting the nutrient requirements of mangoes, and contributing to increased yield. Example 3, through dynamic threshold optimization and synergy with slow-release root fertilizer, showed the best performance in reducing misjudgment rate, water and fertilizer utilization rate, and per-plant yield, with a 58.9% increase in yield compared to the traditional group. The slow-release coating increases its swelling rate and expands its pores when soil moisture content increases, accelerating the release of compound fertilizer; conversely, it decreases its swelling rate and shrinks its pores when soil moisture content decreases, slowing down the release of compound fertilizer and extending its fertilizer effect cycle and utilization rate.

[0096] Compared to the comparative example, Example 3 applied slow-release coated root fertilizer in the fertilization zone located inside the irrigation area. This allowed the slow-release coated root fertilizer, located inside the irrigation zone, to be triggered by swelling and release through irrigation water infiltration, achieving spatial coupling of water and fertilizer. This reduced nutrient level and outward diffusion losses, improved water and fertilizer utilization, and resulted in better relative yield and profit. The root fertilizer formula combined slow-release and water-retaining components to extend the fertilizer's effectiveness period and adapt to the nutrient requirements of mangoes at different growth stages. The slow-release coating technology, through water-fertilizer coupling release, reduced nutrient loss and verified the feasibility of its large-scale application.

[0097] Although embodiments of the present invention have been disclosed above, they are not limited to the applications listed in the specification and embodiments. It can be applied to various fields suitable for the present invention. Further modifications can be readily implemented by those skilled in the art.

Claims

1. A method for intelligent regulation of nutrients and water in mango cultivation, characterized in that, include: Based on the canopy projection diameter of the mango tree, the root zone is divided into a ring-shaped fertilization zone and a ring-shaped irrigation zone centered on the trunk, with the fertilization zone located inside the irrigation zone and the irrigation zone equipped with drip irrigation pipelines. Images of mango tree leaves are captured by a camera mounted on a field support frame. The camera is 1-1.5 meters above the mango tree and the shooting angle is vertically downward. The average green channel value of the leaves is extracted using the HSV color segmentation algorithm. When the average green channel value is lower than the average green channel value threshold, it is determined to be nitrogen deficiency. At the same time, the thickness of the leaves is calculated using the Canny edge detection algorithm. When the thickness is lower than the leaf thickness threshold, it is determined to be water deficiency. Based on the nitrogen deficiency determination result, the atomizer is controlled to spray a urea solution with a mass concentration of 0.5-3% onto the leaves, and the spraying amount of the urea solution is 20-30 ml per square meter of leaf area each time. The drip irrigation system is activated based on the water shortage determination result. The irrigation volume of the drip irrigation system is 10-15 liters per square meter of rhizosphere area, and the activation time of the drip irrigation system is 30-45 minutes each time. The green channel threshold in the HSV color segmentation algorithm is calibrated using data collected from an ambient temperature and humidity sensor. The ambient temperature (T) and humidity (H) values ​​are acquired using an ambient temperature and humidity sensor. The mean threshold G for the green channel used to determine nitrogen deficiency is dynamically adjusted according to a preset linear relationship formula: G = G0 固定 -0.5×(T-25)+0.3×(H-60), Among them, T ranges from 20-35℃, and H ranges from 50-80%. G 固定 Using the baseline threshold, the flower bud differentiation period G is set in segments according to the growth stages. 固定 The value is 130, and the flowering period is G. 固定 The value is 120, and the fruit enlargement period is G. 固定 The value is 110, and the maturity level is G. 固定 The value is 100. In the HSV color segmentation algorithm, the leaf area is first extracted based on the hue range of 36-86°, the saturation range of 20-80%, and the brightness range of 40-90%, and then the interference area of ​​new leaves with hue of 42-75°, saturation <40%, and brightness >50% is further excluded. Calculate the mean value of the green channel within the leaf area, and determine nitrogen deficiency when the mean value is lower than the calibrated threshold G; When T exceeds the range of 20-35℃ or H exceeds the range of 50-80%, directly use G. 固定 As a threshold for judgment; The camera uses an external light intensity sensor and an ambient temperature and humidity sensor to collect ambient light intensity (L), temperature (T), and humidity (H) in real time. Based on these environmental parameters, the Canny edge detection algorithm dynamically adjusts the leaf thickness threshold (D) for determining water shortage. The adjustment formula is: D = 0.3 + 0.02 × (T - 25) - 0.01 × (H - 60) - 0.0001 × (L - 7500), where T ranges from 20-35℃, H ranges from 50-80%, and L ranges from 5000-10000 lux. When the calculated leaf thickness is lower than the adjusted threshold D, water shortage is identified, and the drip irrigation system is activated. If T, H, or L exceeds these ranges, a fixed threshold of D = 0.3 mm is used for the determination.

2. The intelligent nutrient and water regulation method for mango cultivation as described in claim 1, characterized in that, The camera has an external light intensity sensor, and the camera dynamically adjusts the exposure time from 1 / 500 second to 1 / 100 second and the aperture from f / 2.8 to f / 8 based on the real-time measured light intensity. The camera's shooting position avoids the reflective areas on the leaf surface. The camera's white balance mode is set to automatic mode, and the green channel threshold in the HSV color segmentation algorithm is calibrated using data collected by an ambient temperature and humidity sensor.

3. The intelligent nutrient and water regulation method for mango cultivation as described in claim 2, characterized in that, Before extracting the green channel mean of the leaf using the HSV color segmentation algorithm, the reflective areas on the leaf surface are identified and marked by the image processing algorithm. The criteria for determining the reflective areas are that the pixel brightness value is higher than 200 and the saturation is lower than 30. When calculating the green channel mean, pixel data within the reflective area are excluded; The camera is set to shoot 1-2 hours after sunrise or 1-2 hours before sunset, and the ambient light intensity is 5000-10000 lux during shooting.

4. The intelligent nutrient and water regulation method for mango cultivation as described in any one of claims 1-3, characterized in that, Also includes: Spraying and irrigation interval control: During the flower bud differentiation period, flowering period, and fruit enlargement period, after each application of urea solution based on the nitrogen deficiency assessment result or on rainy days, lock the start command of the atomizer until the interval reaches 12-24 hours. Each time the drip irrigation system is turned on based on the water shortage assessment or on a rainy day, the irrigation command is locked until the interval reaches 12-24 hours. If a nitrogen or water deficiency is triggered again during the lockout period, only the number of times the deficiency is detected will be recorded, but no spraying or irrigation operations will be performed. Root fertilizer application reminder mechanism: The number of foliar fertilizer applications for each mango tree is counted. When nitrogen deficiency is detected for three consecutive times and the cumulative number of applications reaches five, a root fertilizer application reminder signal is generated for the fertilization area. The reminder signal includes the location marker of the fertilization area and the recommended fertilization amount. The recommended fertilization amount is dynamically set according to the current growth stage: 0.5-1.0 kg of root fertilizer per square meter of fertilization area during the flower bud differentiation period; 1.0-1.5 kg of root fertilizer per square meter of fertilization area during the flowering period; 1.5-2.0 kg of root fertilizer per square meter of fertilization area during the fruit enlargement period; and 1.0-1.5 kg of root fertilizer per square meter of fertilization area during the ripening period.

5. The intelligent nutrient and water regulation method for mango cultivation as described in claim 4, characterized in that, The root fertilizer comprises 98% compound fertilizer, 1% superabsorbent resin, and 1% boron, wherein the compound fertilizer contains nitrogen, phosphorus, and potassium in a mass ratio of 2:1:

1. When fertilizing, dig a trench 15-20 cm deep in the fertilization area, apply the root fertilizer evenly, and then cover it with soil.

6. The intelligent nutrient and water regulation method for mango cultivation as described in claim 5, characterized in that, The root fertilizer is in granular form, with a water-responsive slow-release coating on the surface of the fertilizer granules. The slow-release coating is prepared by mixing starch and polyvinyl alcohol in a mass ratio of 3:1 and then preparing a coating solution with a crosslinking agent. The solution is then uniformly coated by a spraying process, and the coating thickness is 0.2 mm ± 0.02 mm. The amount of crosslinking agent added is 5%-8% of the total mass of starch and polyvinyl alcohol, and the swelling rate of the coating is controlled within the range of 150%-200%.

7. The intelligent nutrient and water regulation method for mango cultivation as described in claim 6, characterized in that, The preparation method of the crosslinking agent includes the following steps: 1) Weigh out glycerol and citric acid in a 1:1 mass ratio; 2) Add glycerol and citric acid to the reaction vessel and stir at 50-60℃ for 200-300 rpm for 30-40 minutes to obtain a uniform and transparent liquid crosslinking agent; 3) After cooling the liquid crosslinking agent to room temperature, seal and store it. When using, add it to the starch-polyvinyl alcohol mixture at 5%-8% of the total mass of starch-polyvinyl alcohol.

8. The intelligent nutrient and water regulation method for mango cultivation as described in claim 6 or 7, characterized in that, The process of coating the fertilizer granules with a moisture-responsive slow-release coating specifically includes the following steps: Compound fertilizer granules with a particle size of 2-4 mm are placed in a fluidized bed coating machine, and an airflow at a temperature of 40-50℃ is introduced to suspend the granules. A coating solution is prepared by mixing starch-polyvinyl alcohol mixture with crosslinking agent in a certain proportion, and then sprayed onto the surface of particles through a nozzle at a rate of 0.2-0.4 ml / s. During the spraying process, the airflow speed is controlled at 3-5 m / s and the spraying pressure is 0.1-0.3 MPa to ensure that the coating liquid evenly covers the particle surface; After spraying, cure under hot air at 50-60℃ for 10-15 minutes to form a coating layer with a thickness of 0.2 mm ± 0.02 mm.

Images