A seedling water supplementing judgment method, system and device based on dynamic change of leaf area index

Abstract

The application provides a seedling water supplement judgment method, system and device based on leaf area index dynamic change, which comprises the following steps: obtaining the RGB image, crop information and stress stage of the experimental group crop seedling, and the RGB image and crop information of the control group crop seedling; marking the effective area of the RGB image; performing accurate extraction according to the core vegetation index and the screening result of the HSV space green range; calculating the leaf area index according to the total pixel number in the effective area and the effective leaf pixel number of the pure leaf mask; calculating the leaf area index relative difference rate of the experimental group according to the leaf area index of the experimental group and the leaf area index of the control group; calculating the strong seedling coefficient of each group according to the crop information; selecting the leaf area index relative difference rate corresponding to the experimental group with the largest strong seedling coefficient as the most suitable water supplement interval, and selecting the leaf area index relative difference rate corresponding to the experimental group with the decreased strong seedling coefficient relative to the control group to determine the water supplement threshold. The application can realize the integrated regulation of "real-time monitoring-growth linkage-accurate water supplement".

Description

Technical Field

[0001] This invention belongs to the field of crop cultivation technology, specifically relating to a method, system, and device for determining seedling water replenishment based on dynamic changes in leaf area index. Background Technology

[0002] The rice seedling raising stage is a critical period that determines seedling quality and affects subsequent field growth and yield. The two-leaf-one-heart stage is the core node in the transition of seedlings from heterotrophic to autotrophic growth, requiring extremely precise water management. Moderate water stress (hardening off) can promote lateral root growth, thicken the stem base, and enhance resistance, while excessive water can lead to excessive vegetative growth, and insufficient water can cause irreversible growth damage. Therefore, obtaining the plant's water content to achieve precise water replenishment is one of the core technologies in rice seedling management.

[0003] Existing methods for judging water replenishment in rice seedling raising mainly rely on manual experience, judging the timing of water replenishment by observing the dryness and wetness of the substrate surface and the morphology of the seedling leaves. This has the following significant drawbacks: First, it is highly subjective, with different operators having different judgment standards, resulting in poor consistency in water management; second, it lacks quantitative basis, making it impossible to accurately match with the physiological needs of hardening off seedlings and making it difficult to achieve the goal of "controlling the top and promoting the bottom" to strengthen seedlings.

[0004] To address the limitations of manual watering, existing technologies include methods that use soil moisture sensors to directly measure substrate moisture content. However, these sensors are susceptible to substrate heterogeneity and installation location, resulting in insufficient data stability. Furthermore, watering strategies based solely on substrate moisture content fail to consider feedback from the seedlings' own growth status, thus failing to reflect the varying water requirements of seedlings at different growth stages. Additionally, existing technologies primarily employ two methods to accurately determine plant moisture content: drying and spectral inversion. However, both have significant limitations and are ill-suited to the practical needs of large-scale, intelligent seedling cultivation. The drying method is less practical in seedling cultivation scenarios. Firstly, it involves destructive sampling, disturbing the seedling root environment and affecting seedling growth. Secondly, it cannot obtain real-time plant moisture content data, making it unsuitable for standardized seedling cultivation processes requiring dynamic control. Spectral inversion is a non-destructive method for monitoring water content that has emerged in recent years. Its principle is based on the correlation between the reflection and absorption characteristics in specific spectral bands (such as the near-infrared band) and water content. Spectral data is collected using a spectrometer, and the water content is calculated using a pre-set inversion model. This method achieves non-destructive monitoring, acquiring data without disturbing the seedlings. However, its widespread application is constrained by several factors: Firstly, the equipment is expensive; a high-precision near-infrared spectrometer and hyperspectral imaging equipment typically cost tens to hundreds of thousands of yuan, far exceeding the economic affordability of ordinary seedling bases and farmers. Secondly, the equipment has high requirements for the operating environment; temperature and humidity fluctuations and dust pollution in the field or seedling greenhouse can affect the accuracy of spectral acquisition. Furthermore, the equipment requires professional personnel to operate, which is difficult for ordinary farmers to master, further limiting its large-scale promotion in grassroots seedling raising scenarios. In summary, neither of these factors can meet the "real-time, accurate, low-cost, and easily promoted" water content monitoring needs in standardized rice seedling raising, resulting in seedling water replenishment still relying heavily on manual experience, thus hindering the development of intelligent seedling raising technology. Therefore, developing a monitoring method that can acquire plant water content in real time, is cost-effective, and is suitable for grassroots promotion has become an urgent need for upgrading rice seedling cultivation technology. Summary of the Invention

[0005] The technical problem to be solved by the present invention is to provide a method, system and device for judging seedling water replenishment based on the dynamic change of leaf area index. By constructing a correlation model between the dynamic change of LAI and the plant water content, a scientific water replenishment threshold based on the feedback of plant water content is established, so as to realize the integrated regulation of "real-time monitoring-growth linkage-precise water replenishment".

[0006] This invention is implemented as follows: a method for determining seedling water replenishment based on dynamic changes in leaf area index, comprising the following steps: For each day after the two lobes and one heart stage, S1. Obtain RGB images, crop information, and stress stages of the experimental group crop seedlings, as well as RGB images and crop information of the control group crop seedlings. The crop information includes seedling height, stem base diameter, and above-ground dry weight. S2. Mark the effective regions of the RGB image and obtain the core vegetation index and HSV spatial green range screening results; S3. Based on the core vegetation index and HSV spatial green range screening results, a pure leaf mask is obtained by precise extraction. S4. Calculate the leaf area index based on the total number of pixels in the effective area and the effective leaf pixel count of the pure leaf mask; S5. Calculate the relative difference rate of leaf area index between the experimental group and the control group based on the leaf area index of the experimental group and the control group. S6. Calculate the seedling vigor coefficient for each group based on the crop information; S7. Select the relative difference rate of leaf area index corresponding to the experimental group with the largest seedling vigor coefficient as the optimal water replenishment range, and select the relative difference rate of leaf area index corresponding to the experimental group with a decrease in seedling vigor coefficient relative to the control group to determine the water replenishment threshold.

[0007] Furthermore, step S2 includes the following steps: S201. Select several vertices on the RGB image to form a closed ROI region, and generate an ROI mask through coordinate scaling. S202. Convert the ROI mask into RGB and HSV dual color spaces, and separate the pixel values ​​of the R, G, B channels and the pixel values ​​of the H, S, V channels. S203. Calculate the core vegetation index for each pixel based on the pixel values ​​of the R, G, and B channels. The core vegetation index includes the excess green index and the normalized difference index. S204. Based on the pixel values ​​of the H, S, and V channels and the preset threshold range of the H, S, and V channels, perform HSV space green range filtering to obtain the filtering results.

[0008] Furthermore, the expression for the coordinate scaling process is: , ; Where, x original y original The coordinates are the image coordinates obtained after scaling and restoration, where x and y are the coordinates of the mouse click on the image display interface. ratio This is the scaling factor; The formula for calculating the excess green index is as follows: ; The formula for calculating the normalized difference index is as follows: ; Where R, G, and B are the pixel values ​​of the R, G, and B channels, respectively, and ε is the minimum value; The expression for the HSV space green range filtering result is: ; Where i,j are the coordinates of the image pixels, and H, S, V are the pixel values ​​of the H, S, and V channels.

[0009] Furthermore, S3 includes the following steps: S301. Perform feature normalization and weighted fusion processing on the core vegetation index to obtain a comprehensive feature score map; S302. Perform adaptive threshold segmentation on the comprehensive feature score map to obtain a binary mask; S303. Perform morphological optimization processing on the binary mask to obtain an optimized image; S304. Perform connected component analysis on the optimized image to obtain a clean leaf mask.

[0010] Furthermore, step S301 includes the following steps: S30101. Perform feature normalization processing on the core vegetation index to obtain the normalization result; the expression of the normalization result is: , ; Among them, ExG norm For the normalization result of the excess green index, ExG min Let ExG be the minimum value of ExG in the interval [0,1] of a single image. max NDI is the maximum value of ExG in the interval [0,1] of a single image. norm The normalized result of the normalized difference index, NDI min The minimum NDI value in a single image within the interval [0,1] is the NDI value. max The maximum value of NDI in a single image within the interval [0,1]; S30102. The normalization result and the HSV space green range screening result are weighted and fused to obtain a comprehensive feature score map; the calculation formula of the comprehensive feature score map is: ; Among them, green_mask_hsv is the result of green range filtering in HSV space.

[0011] Furthermore, step S302 includes the following steps: S30201. Map the comprehensive feature score map to a grayscale image, and then perform Gaussian blurring processing through a preset Gaussian kernel to obtain the standard deviation; S30202. Calculate the weighted average of each pixel in the comprehensive feature score map based on the standard deviation to obtain the blurred feature map. The calculation formula is: ; Where σ is the standard deviation, and x' and y' are the coordinates of the pixels in the kernel relative to the center; S30203. The feature map is segmented using Otsu's method and converted into a binary image. The conversion formula is as follows: ; Where i',j' are the pixel coordinates of the feature map, 255 represents the leaf region, and 0 represents the background region.

[0012] Furthermore, step S303 includes the following steps: S30301. The binary image is subjected to an opening operation of erosion followed by dilation using a preset square structuring element, resulting in an image after the opening operation, the expression of which is: ; Among them, Kernel n*n It is an n*n square structuring element. , x' and y' are the coordinates of the pixels within the kernel relative to the center, and i' and j' are the pixel coordinates of the feature map; S30302. Using a preset square structuring element, perform a dilation-erosion closing operation on the image after the opening operation to obtain an optimized image, the expression of which is: .

[0013] Furthermore, step S304 includes the following steps: S30401. Divide all continuous blade regions into several non-overlapping independent regions, and treat each independent region as a connected component. S30402. The effective region is filtered according to a preset minimum area threshold, and pixels belonging to connected components with an area not less than the minimum area threshold are retained as leaf region pixels, thus obtaining a clean leaf mask; the expression for the minimum area threshold is: ; Among them, A total This represents the total number of pixels in the image. H is the image height, and W is the image width; The expression for the pure leaf mask is: ; Among them, C i Let x' and y' be the i-th connected component, where x' and y' are the coordinates of the pixels within the kernel relative to the center.

[0014] Furthermore, the formula for calculating the leaf area index is as follows: ; Where, N total N represents the total number of pixels within the effective area. leaf This represents the effective number of leaf pixels in the pure leaf mask.

[0015] Furthermore, the formula for calculating the relative difference rate of the leaf area index is as follows: ; Among them, LAI p Leaf area index (LAI) is the control group. q The leaf area index is for the experimental group.

[0016] This invention also provides a seedling water replenishment judgment model construction system based on dynamic changes in leaf area index, comprising: The acquisition module is used to acquire RGB images, crop information and stress stage of crop seedlings in the experimental group, as well as RGB images and crop information of crop seedlings in the control group. The crop information includes seedling height, stem base diameter and above-ground dry weight. The first processing module is used to mark the effective regions of the RGB image and obtain the core vegetation index and HSV spatial green range screening results; The second processing module is used to accurately extract the pure leaf mask based on the core vegetation index and the HSV spatial green range screening results. The third processing module is used to calculate the leaf area index based on the total number of pixels in the effective area and the effective leaf pixel number of the pure leaf mask. The fourth processing module is used to calculate the relative difference rate of leaf area index between the experimental group and the control group based on the leaf area index of the experimental group. The fifth processing module is used to calculate the seedling vigor coefficient for each group based on the crop information; The sixth processing module is used to select the relative difference rate of leaf area index corresponding to the experimental group with the largest seedling vigor coefficient as the optimal water replenishment range, and to select the relative difference rate of leaf area index corresponding to the experimental group with a decrease in seedling vigor coefficient relative to the control group to determine the water replenishment threshold.

[0017] The present invention also provides a seedling water replenishment judgment model construction device based on dynamic changes in leaf area index, including a processor and a memory. The memory is used to store a computer program, and the processor is used to execute the computer program to implement the steps of the above method.

[0018] The beneficial effects of this invention are: 1. This invention precisely focuses on the core water content requirements of plants, significantly improving the scientific rigor and relevance of watering decisions. Breaking through the limitations of existing technologies that only focus on substrate water content, this invention uses plant water content as the core monitoring target. Through in-depth linkage analysis between leaf growth status (LAI) and plant water content, it effectively avoids the problem of insufficient or excessive watering caused by a disconnect between substrate and plant water content. Compared to traditional experience-based watering methods, this invention can accurately capture signals of seedling water stress, shifting watering decisions from "relying on experience" to "relying on physiological state feedback," effectively ensuring robust seedling growth during hardening-off and significantly improving the stability of seedling quality.

[0019] 2. This invention establishes a low-cost, non-destructive real-time monitoring system with strong feasibility for grassroots implementation. This invention abandons the expensive equipment of spectral inversion methods and the destructive sampling of drying methods, employing a low-cost depth camera paired with a moisture sensor of controllable precision, combined with a self-designed multi-feature fusion leaf segmentation algorithm, to achieve indirect quantitative monitoring of plant water content. This algorithm integrates RGB space ExG, NDI index, and HSV space green mask features, and through Gaussian blur denoising, Otsu's adaptive threshold segmentation, morphological optimization, and connected component screening, effectively improves leaf segmentation accuracy. Furthermore, while ensuring monitoring accuracy, it significantly reduces equipment purchase and operating costs. The entire monitoring system requires no professional maintenance personnel, and its non-destructive design enables continuous tracking of the same batch of seedlings throughout their entire growth period, fully adapting to the large-scale application needs of grassroots seedling nurseries and farmers, solving the bottlenecks of existing technologies being "expensive, difficult, and indestructible" for widespread adoption.

[0020] 3. This invention establishes a growth-water linkage water replenishment threshold system to achieve intelligent and precise water replenishment regulation. Breaking away from the limitations of traditional single-indicator methods for determining water replenishment timing, this invention uses the LAI (Labor Intake) of the unstressed control group as a benchmark. It quantifies the degree of water stress in plants by calculating the relative difference rate of LAI, and then combines this with the seedling vigor coefficient at the experimental endpoint for dual-dimensional verification, accurately establishing the "optimal water replenishment range" and the "critical stress threshold." This system achieves a closed-loop linkage between "plant physiological state - water demand - water replenishment timing," achieving seedling hardening through mild stress while avoiding irreversible damage to seedlings from excessive stress. Compared to traditional water replenishment methods, it also saves irrigation water, enhances seedling resistance, lays a good foundation for subsequent growth after transplanting, and helps upgrade rice seedling cultivation towards intelligence, water conservation, and high efficiency. Attached Figure Description

[0021] Figure 1 This is a flowchart of the method in this invention; Figure 2This is a schematic diagram of the ROI selection for the seedling tray in this invention; Figure 3 This is a schematic diagram illustrating the seedling image processing and leaf segmentation effects in this invention. Figure 4 This is a graph showing the dynamic changes in LAI under different water treatments in this invention. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings. The following embodiments are only used to more clearly illustrate the technical solution of the present invention, and should not be used to limit the scope of protection of the present invention.

[0023] Taking the Nanjing 9108 rice variety as an example, a standard 9-inch seedling tray (60cm×30cm×3cm) was used, with 150g of seeds sown in each tray. The volumetric moisture content sensor probe was inserted into the substrate in the center of the seedling tray (1cm deep) to ensure close contact with the substrate. An Intel RealSense D455 depth camera was fixed 50cm directly above the seedling tray, with the lens perpendicular to the plane of the tray, and the field of view completely covering the entire tray of seedlings. This experiment set up one control group (T1) and three experimental groups (T2, T3, T4). The control group kept the substrate moist throughout the experiment (moisture content 40%-50%). The experimental groups were subjected to mild stress (when the substrate moisture content dropped to 10% or below, water was added to moisten the same day), moderate stress (when the substrate moisture content dropped to 10% or below, water was added to moisten the substrate every 24 hours), and severe stress (when the substrate moisture content dropped to 10% or below, water was added to moisten the substrate every 48 hours) after the two-leaf stage.

[0024] Example 1: As Figures 1 to 4 As shown, this invention provides a method for determining seedling water replenishment based on dynamic changes in leaf area index, comprising the following steps: For each day after the two lobes and one heart stage, S1. Obtain RGB images, crop information, and stress stages of the experimental group crop seedlings, as well as RGB images and crop information of the control group crop seedlings. The crop information includes seedling height, stem base diameter, and above-ground dry weight.

[0025] S2. Effective region labeling is performed on the RGB image, and the core vegetation index and HSV spatial green range screening results are obtained. This improves the distinguishability between leaves, soil, and impurities, providing high-quality data support for subsequent accurate leaf area segmentation. Specifically, S2 includes the following steps: S201. Using an interactive point selection method, four vertices are selected on the RGB image to form a closed Region of Interest (ROI) (i.e., the effective planting area of ​​the seedling tray), and an ROI mask is generated through coordinate scaling. Specifically, the expression for the coordinate scaling process is: , ; Where, x original y original The coordinates are the image coordinates obtained after scaling and restoration, where x and y are the coordinates of the mouse click on the image display interface. ratio This is the scaling factor.

[0026] S202. Convert the ROI mask into RGB and HSV dual color spaces, and separate the pixel values ​​of the R, G, B channels and the pixel values ​​of the H, S, V channels.

[0027] S203. Calculate the core vegetation index for each pixel based on the R, G, and B channel pixel values. The core vegetation index includes an excess green index and a normalized difference index. The formula for calculating the excess green index is: ; The formula for calculating the normalized difference index is as follows: ; Where R, G, and B are the pixel values ​​of the R, G, and B channels, respectively, and ε is the minimum value to avoid a denominator of 0.

[0028] S204. Based on the pixel values ​​of the H, S, and V channels and the preset threshold ranges of the H, S, and V channels (H∈[35,85], S∈[40,255], V∈[30,255]), perform HSV space green range filtering to obtain the filtering result. The expression for the HSV space green range filtering result is: ; Where i,j are the coordinates of the image pixels, and H, S, V are the pixel values ​​of the H, S, and V channels.

[0029] S3. Based on the core vegetation index and HSV spatial green range screening results, precise extraction is performed to obtain a pure leaf mask. Specifically, S3 includes the following steps: S301. Perform feature normalization and weighted fusion processing on the core vegetation index to obtain a comprehensive feature score map. Specifically, S301 includes the following steps: S30101. The core vegetation index is normalized to the [0,1] interval using min-max standardization. The expression for the normalization result is: , ; Among them, ExG norm For the normalization result of the excess green index, ExG min The minimum value of ExG in a single image, ExG maxNDI is the maximum value of ExG in a single image. norm The normalized result of the normalized difference index, NDI min The minimum NDI value in a single image. max This represents the maximum value of NDI in a single image.

[0030] S30102. The normalization result and the HSV space green range screening result are weighted and fused to obtain a comprehensive feature score map. The calculation formula for the comprehensive feature score map is: ; Among them, green_mask_hsv is the result of green range filtering in HSV space.

[0031] S302. Perform adaptive threshold segmentation on the comprehensive feature score map to obtain a binary mask. Specifically, S302 includes the following steps: S30201. The comprehensive feature score map is mapped to an 8-bit grayscale image, and then Gaussian blurring is performed using a 5*5 Gaussian kernel. The standard deviation parameter is set to 0. At this time, the algorithm automatically calculates the standard deviation based on the kernel size. In this invention, the standard deviation corresponding to the 5*5 kernel is 1.1, which avoids the subjectivity of manually setting the standard deviation and ensures that the blurring effect is adapted to the fine segmentation requirements of seedling leaves.

[0032] S30202. Calculate the weighted average of each pixel in the comprehensive feature score map based on the standard deviation to obtain the blurred feature map. The calculation formula is: ; Where σ is the standard deviation, and x' and y' are the coordinates of the pixels within the kernel relative to the center.

[0033] S30203. The feature map is segmented using Otsu's method. Otsu's method is an adaptive threshold segmentation algorithm that finds a threshold `otsu_thresh` by traversing all possible grayscale values ​​(0-255) to maximize the inter-class variance between the segmented "leaf region" and "background region" (the larger the inter-class variance, the more significant the difference between the two classes). A key advantage of Otsu's method is its automatic adaptation to the pixel distribution of each feature map, solving the problem of incompatible thresholds under different lighting conditions and seedling densities. Finally, using Otsu's `otsu_thresh` as the boundary, the blurred feature map is converted into a binary image. The conversion formula is: ; Where i',j' are the pixel coordinates of the feature map, 255 represents the leaf region, and 0 represents the background region.

[0034] S303. Perform morphological optimization processing on the binary mask to obtain an optimized image. This step is the core step for noise removal and mask repair after leaf segmentation. Specifically, S303 includes the following steps: S30301. An opening operation, involving erosion followed by dilation, is performed on the binary image using a 3*3 square structuring element to obtain the opened image. The core of this step is to remove noise without altering the main shape of the blade. Its expression is: ; Among them, Kernel 3*3 It is a 3x3 square structural element. , x' and y' are the coordinates of the pixels within the kernel relative to the center, and i' and j' are the pixel coordinates of the feature map.

[0035] S30302. After removing small noises such as matrix particles and light spots to make the leaf mask edges cleaner, a 3*3 square structuring element is used to perform a dilation-erosion closing operation on the image after the opening operation to obtain an optimized image. The core of this step is to fill the holes without changing the main shape of the leaf. Its expression is: .

[0036] The etching process removes bright spots smaller than 3x3 and slightly shrinks the blade edges. The expansion process fills in dark cavities smaller than 3x3 inside the blade and slightly expands the blade edges.

[0037] S304. Even after morphological optimization, the leaf mask may still contain pseudo-leaf regions (such as residual small noise points or tiny connected regions of matrix reflection). These regions are not real seedling leaves and will lead to an inflated LAI calculation. Therefore, connected component analysis needs to be performed on the optimized image to obtain a clean leaf mask. Specifically, S304 includes the following steps: S30401. Divide all continuous blade regions into several non-overlapping independent regions, and treat each independent region as a connected component.

[0038] S30402. The effective region is filtered according to a preset minimum area threshold, and pixels belonging to connected components with an area not less than the minimum area threshold are retained as leaf region pixels, thus obtaining a clean leaf mask; the expression for the minimum area threshold is: ; Among them, A total This represents the total number of pixels in the image. H is the image height, and W is the image width.

[0039] The expression for the pure leaf mask is: ; Among them, C i Let x' and y' be the i-th connected component, where x' and y' are the coordinates of the pixels within the kernel relative to the center.

[0040] S4. Calculate the leaf area index (LAI) based on the total number of pixels within the effective area and the effective leaf pixels of the pure leaf mask. Starting from the two-leaf-one-heart stage of the seedlings, repeat steps S1-S4 daily to continuously collect and calculate LAI, constructing dynamic LAI datasets for different water treatment groups. Set up a T1 control group (with substrate moisture content maintained at 40%-50% of field capacity throughout the entire growth cycle). Use the LAI growth curve of the control group as the baseline for stress-free growth, clarifying the LAI variation range required for "moderate stress" in hardening off. Specifically, the formula for calculating the leaf area index is: ; Where, N total N represents the total number of pixels within the effective area. leaf This represents the effective number of leaf pixels in the pure leaf mask.

[0041] S5. Calculate the relative difference rate of leaf area index (LAI) between the experimental group and the control group. Real-time monitoring of LAI changes using differentiated watering strategies, recording growth recovery trajectories, effectively distinguishing the reversibility differences of stress, and providing dynamic data support for determining subsequent watering intervals and thresholds. Watering operations are performed according to preset watering trigger conditions for different treatment groups, with each watering operation uniformly restoring the substrate moisture content to 40%-50% of field capacity. After watering, continuous monitoring of LAI and moisture content changes is conducted, and growth recovery dynamics are recorded. Specifically, the formula for calculating the relative difference rate of leaf area index is: ; Among them, LAI p Leaf area index (LAI) is the control group. q The leaf area index is for the experimental group.

[0042] S6. Calculate the seedling vigor coefficient for each group based on the crop information. The calculation formula is as follows: .

[0043] S7. Select the relative difference rate of leaf area index corresponding to the experimental group with the largest seedling vigor coefficient as the optimal water replenishment range, and select the relative difference rate of leaf area index corresponding to the experimental group with a decrease in seedling vigor coefficient relative to the control group to determine the water replenishment threshold.

[0044] As shown in Table 1, the optimal watering range and watering threshold can be determined by comparing the seedling vigor coefficients among the groups. The optimal watering range represents the relative difference rate of LAI before watering corresponding to the experimental group with the highest seedling vigor coefficient. The watering threshold represents the extent of water shortage that will have an irreversible impact on the seedlings if watering is continued beyond this threshold, making it difficult for the seedlings to recover to normal growth levels. Taking this experiment as an example, the comparison of seedling vigor coefficients is as follows: T2 (mild stress) > T1 (control group) > T3 (moderate stress). T4 (severe stress) was already dead, and the seedling vigor coefficient could not be measured. Therefore, the relative difference rate of LAI corresponding to T2 can be used as the optimal watering range, with a value of 27.44%. In addition, since the seedling vigor coefficient of T3 (moderate stress) is less than that of T1 (control group), the relative difference rate of LAI corresponding to group T3 can be used as the watering threshold, with a value of 43.54%.

[0045] Table 1 Water Replenishment Threshold Judgment Table Group Day 3 LAI Day 4 LAI Ultimately LAI Remark T1 0.4432 0.4759 0.8407 No coercion T2 0.3216 0.4382 0.7774 Mild stress T3 0.3154 0.2687 0.726 Moderate stress (reversible) T4 0.3222 0.2573 0.1816 Severe stress Example 2: Based on the same inventive concept, this invention also provides a seedling water replenishment judgment model construction system based on dynamic changes in leaf area index, including: The acquisition module is used to acquire RGB images, crop information and stress stage of crop seedlings in the experimental group, as well as RGB images and crop information of crop seedlings in the control group. The crop information includes seedling height, stem base diameter and above-ground dry weight. The first processing module is used to mark the effective regions of the RGB image and obtain the core vegetation index and HSV spatial green range screening results; The second processing module is used to accurately extract the pure leaf mask based on the core vegetation index and the HSV spatial green range screening results. The third processing module is used to calculate the leaf area index based on the total number of pixels in the effective area and the effective leaf pixel number of the pure leaf mask. The fourth processing module is used to calculate the relative difference rate of leaf area index between the experimental group and the control group based on the leaf area index of the experimental group. The fifth processing module is used to calculate the seedling vigor coefficient for each group based on the crop information; The sixth processing module is used to select the relative difference rate of leaf area index corresponding to the experimental group with the largest seedling vigor coefficient as the optimal water replenishment range, and to select the relative difference rate of leaf area index corresponding to the experimental group with a decrease in seedling vigor coefficient relative to the control group to determine the water replenishment threshold.

[0046] Example 3: Based on the same inventive concept, the present invention also provides a seedling water replenishment judgment model construction device based on dynamic changes in leaf area index, including a processor and a memory. The memory is used to store a computer program, and the processor is used to execute the computer program to implement the steps of the above method.

[0047] The above are merely preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the technical principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for determining seedling water replenishment based on dynamic changes in leaf area index, characterized in that, Includes the following steps: For each day after the two lobes and one heart stage, S1. Obtain RGB images, crop information, and stress stages of the experimental group crop seedlings, as well as RGB images and crop information of the control group crop seedlings. The crop information includes seedling height, stem base diameter, and above-ground dry weight. S2. Mark the effective regions of the RGB image and obtain the core vegetation index and HSV spatial green range screening results; S3. Based on the core vegetation index and HSV spatial green range screening results, a pure leaf mask is obtained by precise extraction. S4. Calculate the leaf area index based on the total number of pixels in the effective area and the effective leaf pixel count of the pure leaf mask; S5. Calculate the relative difference rate of leaf area index between the experimental group and the control group based on the leaf area index of the experimental group and the control group. S6. Calculate the seedling vigor coefficient for each group based on the crop information; S7. Select the relative difference rate of leaf area index corresponding to the experimental group with the largest seedling vigor coefficient as the optimal water replenishment range, and select the relative difference rate of leaf area index corresponding to the experimental group with a decrease in seedling vigor coefficient relative to the control group to determine the water replenishment threshold.

2. The method for determining seedling water replenishment based on dynamic changes in leaf area index according to claim 1, characterized in that, S2 includes the following steps: S201. Select several vertices on the RGB image to form a closed ROI region, and generate an ROI mask through coordinate scaling. S202. Convert the ROI mask into RGB and HSV dual color spaces, and separate the pixel values ​​of the R, G, B channels and the pixel values ​​of the H, S, V channels. S203. Calculate the core vegetation index for each pixel based on the pixel values ​​of the R, G, and B channels. The core vegetation index includes the excess green index and the normalized difference index. S204. Based on the pixel values ​​of the H, S, and V channels and the preset threshold range of the H, S, and V channels, perform HSV space green range filtering to obtain the filtering results.

3. The method for determining seedling water replenishment based on dynamic changes in leaf area index according to claim 2, characterized in that, The expression for the coordinate scaling process is: ， ； Where, x original y original The coordinates are the image coordinates obtained after scaling and restoration, where x and y are the coordinates of the mouse click on the image display interface. ratio This is the scaling factor; The formula for calculating the excess green index is as follows: ； The formula for calculating the normalized difference index is as follows: ； Where R, G, and B are the pixel values ​​of the R, G, and B channels, respectively, and ε is the minimum value; The expression for the HSV space green range filtering result is: ； Where i,j are the coordinates of the image pixels, and H, S, V are the pixel values ​​of the H, S, and V channels.

4. The method for determining seedling water replenishment based on dynamic changes in leaf area index according to claim 1, characterized in that, S3 includes the following steps: S301. Perform feature normalization and weighted fusion processing on the core vegetation index to obtain a comprehensive feature score map; S302. Perform adaptive threshold segmentation on the comprehensive feature score map to obtain a binary mask; S303. Perform morphological optimization processing on the binary mask to obtain an optimized image; S304. Perform connected component analysis on the optimized image to obtain a clean leaf mask.

5. The seedling water replenishment judgment method based on dynamic changes in leaf area index according to claim 4, characterized in that, S301 includes the following steps: S30101. Perform feature normalization processing on the core vegetation index to obtain the normalization result; the expression of the normalization result is: ， ； Among them, ExG norm For the normalization result of the excess green index, ExG min Let ExG be the minimum value of ExG in the interval [0,1] of a single image. max NDI is the maximum value of ExG in the interval [0,1] of a single image. norm The normalized result of the normalized difference index, NDI min The minimum NDI value in a single image within the interval [0,1] is the NDI value. max The maximum value of NDI in a single image within the interval [0,1]; S30102. The normalization result and the HSV space green range screening result are weighted and fused to obtain a comprehensive feature score map; the calculation formula of the comprehensive feature score map is: ； Among them, green_mask_hsv is the result of green range filtering in HSV space.

6. The method for determining seedling water replenishment based on dynamic changes in leaf area index according to claim 4, characterized in that, S302 includes the following steps: S30201. Map the comprehensive feature score map to a grayscale image, and then perform Gaussian blurring processing through a preset Gaussian kernel to obtain the standard deviation; S30202. Calculate the weighted average of each pixel in the comprehensive feature score map based on the standard deviation to obtain the blurred feature map. The calculation formula is: ； Where σ is the standard deviation, and x' and y' are the coordinates of the pixels in the kernel relative to the center; S30203. The feature map is segmented using Otsu's method and converted into a binary image. The conversion formula is as follows: ； Where i',j' are the pixel coordinates of the feature map, 255 represents the leaf region, and 0 represents the background region.

7. The method for determining seedling water replenishment based on dynamic changes in leaf area index according to claim 4, characterized in that, S303 includes the following steps: S30301. The binary image is subjected to an opening operation of erosion followed by dilation using a preset square structuring element, resulting in an image after the opening operation, the expression of which is: ； Among them, Kernel n*n It is an n*n square structuring element. , x' and y' are the coordinates of the pixels within the kernel relative to the center, and i' and j' are the pixel coordinates of the feature map; S30302. Using a preset square structuring element, perform a dilation-erosion closing operation on the image after the opening operation to obtain an optimized image, the expression of which is: 。 8. The method for determining seedling water replenishment based on dynamic changes in leaf area index according to claim 4, characterized in that, S304 includes the following steps: S30401. Divide all continuous blade regions into several non-overlapping independent regions, and treat each independent region as a connected component. S30402. The effective region is filtered according to a preset minimum area threshold, and pixels belonging to connected components with an area not less than the minimum area threshold are retained as leaf region pixels, thus obtaining a clean leaf mask; the expression for the minimum area threshold is: ； Among them, A total This represents the total number of pixels in the image. H is the image height, and W is the image width; The expression for the pure leaf mask is: ； Among them, C i Let x' and y' be the i-th connected component, where x' and y' are the coordinates of the pixels within the kernel relative to the center.

9. The method for determining seedling water replenishment based on dynamic changes in leaf area index according to claim 1, characterized in that, The formula for calculating the leaf area index is: ； Where, N total N represents the total number of pixels within the effective area. leaf This represents the effective number of leaf pixels in the pure leaf mask.

10. The method for determining seedling water replenishment based on dynamic changes in leaf area index according to claim 1, characterized in that, The formula for calculating the relative difference rate of leaf area index is: ； Among them, LAI p Leaf area index (LAI) is the control group. q The leaf area index is for the experimental group.

11. A seedling water replenishment judgment model construction system based on dynamic changes in leaf area index, characterized in that, include: The acquisition module is used to acquire RGB images, crop information and stress stage of crop seedlings in the experimental group, as well as RGB images and crop information of crop seedlings in the control group. The crop information includes seedling height, stem base diameter and above-ground dry weight. The first processing module is used to mark the effective regions of the RGB image and obtain the core vegetation index and HSV spatial green range screening results; The second processing module is used to accurately extract the pure leaf mask based on the core vegetation index and the HSV spatial green range screening results. The third processing module is used to calculate the leaf area index based on the total number of pixels in the effective area and the effective leaf pixel number of the pure leaf mask. The fourth processing module is used to calculate the relative difference rate of leaf area index between the experimental group and the control group based on the leaf area index of the experimental group. The fifth processing module is used to calculate the seedling vigor coefficient for each group based on the crop information; The sixth processing module is used to select the relative difference rate of leaf area index corresponding to the experimental group with the largest seedling vigor coefficient as the optimal water replenishment range, and to select the relative difference rate of leaf area index corresponding to the experimental group with a decrease in seedling vigor coefficient relative to the control group to determine the water replenishment threshold.

12. A device for constructing a seedling water replenishment judgment model based on dynamic changes in leaf area index, characterized in that, It includes a processor and a memory, the memory being used to store a computer program, and the processor being used to execute the computer program to implement the steps of the method according to any one of claims 1-10.

Images