Leaf determination method and system based on image processing

Abstract

The application discloses a leaf determination method and system based on image processing and belongs to the technical field of leaf determination. The method comprises the following steps: collecting a leaf image and extracting morphological feature information of the leaf image; constructing a leaf vein skeleton recognition model, analyzing the morphological feature information, and obtaining a partition coordinate system of the leaf; and obtaining a corresponding health state evaluation result of the leaf according to a leaf recognition mechanism and in combination with the partition coordinate system of the leaf. The leaf recognition mechanism adopts a symmetry judgment method, a diffusion mode judgment method and a leaf vein guiding gradual change recognition method for output according to the partition coordinate system of the leaf. The application has the beneficial effect that the internal structural features of the leaf can be accurately recognized through the deep analysis of the leaf vein structure, especially the application of the main vein feature extraction method and the lateral vein feature analysis method, which provides an important basis for the early recognition of discoloration.

Description

Technical Field

[0001] This invention relates to the field of leaf measurement technology, and more specifically to a leaf measurement method and system based on image processing. Background Technology

[0002] Detecting the health status of plant leaves is a crucial technological requirement in modern agriculture and plant protection. Plant leaves discolor when they experience nutrient deficiencies, diseases, or pests. Traditional methods for detecting leaf discoloration rely primarily on manual visual observation and experience-based judgment. This approach is not only inefficient but also highly subjective, making it difficult to accurately identify and quantify early discoloration. With the rapid development of computer vision technology, image processing-based leaf detection methods have gradually become a research hotspot; however, existing technologies still have many limitations.

[0003] Most existing image processing methods rely on simple color analysis or texture feature extraction to identify discoloration. These methods often only detect surface features of discoloration and cannot delve into the internal structural characteristics of the leaves. Especially in the early stages of discoloration, the leaf surface may not yet show obvious chlorosis, yellowing, spots, or patches, making effective identification difficult with traditional methods. Furthermore, current technologies lack in-depth analysis of leaf vein structure. As the vascular bundle system of a plant, the structural characteristics of leaf veins are closely related to the health of the leaf, and discoloration often spreads along the veins. Summary of the Invention

[0004] Purpose of the invention: The purpose of this invention is to address the shortcomings of existing technologies that can only detect surface features caused by differences in leaf color due to nutrient deficiencies, diseases, or pests, and cannot deeply analyze the internal structural features of leaves. This invention provides a leaf measurement method and system based on image processing.

[0005] Technical solution: The image processing-based leaf measurement method of the present invention comprises the following steps:

[0006] Leaf images are acquired, and morphological feature information of the leaf images is extracted; a leaf vein skeleton recognition model is constructed, and the morphological feature information is analyzed to obtain the leaf partition coordinate system; based on the nutrient deficiency or pest and disease recognition mechanism, combined with the leaf partition coordinate system, the corresponding health status assessment result of the leaf is obtained.

[0007] The nutrient deficiency or pest and disease identification mechanism outputs results based on the leaf zonal coordinate system using symmetry judgment method, diffusion pattern judgment method, and leaf vein guidance gradient identification method.

[0008] As a preferred embodiment of the image processing-based leaf measurement method of the present invention, the step of extracting morphological feature information of the leaf image includes: preprocessing the leaf image; extracting the edge information of the preprocessed leaf image and obtaining the complete outline of the leaf image based on the edge information; obtaining local features of the leaf within the complete outline of the leaf image; and combining the local features of the leaf to form complete morphological feature information.

[0009] As a preferred embodiment of the leaf measurement method based on image processing described in this invention, the leaf vein skeleton recognition model includes a main vein feature extraction method, a lateral vein feature analysis method, and a nutrient deficiency or pest and disease transmission path recognition method.

[0010] The method for extracting the main vein features includes: extracting the complete path of the leaf's main vein, establishing the main vein centerline from the petiole connection point to the leaf tip; measuring the width variation at each point on the main vein centerline to identify the distribution pattern of the main vein's thickness; calculating the offset angle between the main vein and the leaf's long axis to determine whether the main vein is deformed; and dividing the main vein into three types—strong segments, normal segments, and weak segments—based on the width distribution and offset angle of the main vein.

[0011] The lateral vein feature analysis method includes: identifying all lateral veins branching off from the main vein, measuring the angle between the lateral veins and the main vein; counting the number of lateral veins per unit length of the main vein, and calculating the lateral vein density distribution; comparing the number and angle of lateral veins on the left and right sides, and assessing the symmetry of the lateral vein distribution; and outputting lateral vein development health assessment data based on the lateral vein angle, density, and symmetry.

[0012] The method for identifying the transmission path of nutrient deficiency or pests includes: establishing a transmission network of nutrient deficiency or pests using the main vein and lateral veins as the path framework; setting transmission impedance points on the transmission network; and when a nutrient deficiency or pest area is detected, calculating the transmission path of nutrient deficiency or pests based on the location of the nutrient deficiency or pest, and predicting the range of transmission of nutrient deficiency or pests.

[0013] The beneficial effects of this preferred technical solution are as follows: The leaf vein skeleton recognition model, through the combination of main vein feature extraction, lateral vein feature analysis, and nutrient deficiency or pest and disease transmission path identification methods, achieves a comprehensive and in-depth analysis of the leaf vein structure. The main vein feature extraction method, by establishing the main vein centerline and measuring width changes, can identify the health status of the main vein and scientifically divide the main vein into strong, normal, and weak segments, providing a basis for nutrient deficiency or pest and disease risk assessment. The lateral vein feature analysis method, through the analysis of angle, density, and symmetry, assesses the development status of leaf veins and can detect abnormal leaf vein development and potential nutrient deficiency or pest and disease risks at an early stage. The nutrient deficiency or pest and disease transmission path identification method establishes a transmission network and impedance point mechanism, providing a scientific theoretical basis for predicting nutrient deficiency or pest and disease transmission, enabling the system to have a forward-looking ability to control nutrient deficiency or pests and diseases.

[0014] As a preferred embodiment of the image processing-based leaf measurement method of the present invention, the step of establishing the partitioned coordinate system is as follows: the leaf is divided into two symmetrical regions, left and right, with the midrib centerline as the central axis; the symmetrical regions are divided into leaf tip region, leaf middle region, and leaf base region with the lateral veins as the boundary lines, and a partitioned coordinate system for the leaf is established; based on the distribution of strong, normal, and weak segments of the midrib, combined with the assessment data of the health of lateral vein development, nutrient deficiency or disease and pest susceptibility areas are marked in the partitioned coordinate system of the leaf; the propagation impedance points of the propagation network are matched with the partitioned coordinate system of the leaf to establish a partitioned propagation impedance map.

[0015] In a preferred embodiment of the image processing-based leaf measurement method of the present invention, the operation steps of the symmetry judgment method include:

[0016] Obtain the leaf's zonal coordinate system and calculate the color difference between the left leaf tip, leaf middle, and leaf base regions and their corresponding right-side regions. Establish a leaf state history database to store the color characteristics of the leaf's zonal coordinate system at different time points. Introduce a time-weighted decay function to calculate the deviation vector data between the current state and the historical state. When the deviation vector data increases in a certain region, combine the impedance value of that region in the zonal propagation impedance map and the assessment data of the health of lateral vein development to increase the nutrient deficiency or pest and disease risk level of that region. Determine acute nutrient deficiency or pest and disease, or chronic nutrient deficiency or pest and disease, based on the rate of change of the deviation vector data. Feedback the symmetry judgment results to the nutrient deficiency or pest and disease propagation path identification method to adjust the propagation impedance point values ​​of the propagation network. Output the evaluation score of the symmetry judgment method and the deviation vector data.

[0017] The beneficial effects of this preferred technical solution are as follows: the symmetry judgment method, by introducing historical data comparison and analysis in the time dimension, realizes the monitoring of changes in leaf health status; through the calculation and analysis of deviation vector data, it can promptly detect minute changes in leaf health status, realizing early warning of nutrient deficiency or pests and diseases; the function of distinguishing between acute and chronic abnormalities provides an important basis for targeted treatment of different types of nutrient deficiency or pests and diseases.

[0018] As a preferred embodiment of the image processing-based leaf measurement method of the present invention, the operation steps of the diffusion pattern judgment method include: identifying abnormal color regions in the leaf image; matching the position of the abnormal color region with the leaf's partition coordinate system to determine the partition to which the abnormal region belongs; judging the interaction relationship between the abnormal color region and the propagation network based on the partition propagation impedance diagram; when the abnormal color region is located in a weak section of the main vein, it is judged as a high-risk diffusion pattern; when the abnormal color region is located in a strong section of the main vein, it is judged as a low-risk diffusion pattern; analyzing the diffusion direction of the abnormal color region, and combining it with the propagation path predicted by the propagation network, judging the diffusion type as propagation along the vein, propagation across the vein, or random diffusion; comparing and verifying the diffusion pattern judgment result with the deviation vector data of the symmetry judgment method; and outputting the evaluation score and diffusion risk level data of the diffusion pattern judgment method.

[0019] The beneficial effects of this preferred technical solution are as follows: by locating and analyzing abnormal color areas, the identification and risk assessment of nutrient deficiency or pest spread patterns are realized; the technical approach of matching abnormal areas with the zoning coordinate system ensures the accuracy of spatial analysis, locating the specific location and zoning of nutrient deficiency or pest occurrence.

[0020] As a preferred embodiment of the image processing-based leaf measurement method of the present invention, the operation steps of the vein-guided gradient recognition method include: establishing a gradient detection zone along the propagation path of the propagation network, the width of the gradient detection zone being adjusted according to the width of the main vein and lateral veins; calculating the color gradient change rate within the gradient detection zone; marking the vein-guided gradient region when the color gradient change rate exceeds a preset threshold; determining the failure of the propagation impedance point when the vein-guided gradient region appears near the propagation impedance point, based on the location of the propagation impedance point; updating the partitioned propagation impedance map according to the number and distribution of the failed propagation impedance points; using the vein-guided gradient recognition result as a weight adjustment factor to correct the evaluation thresholds of the symmetry judgment method and the diffusion pattern judgment method; recalculating the propagation path and propagation speed of nutrient deficiency or pests and diseases using the updated partitioned propagation impedance map; and outputting the evaluation score of the vein-guided gradient recognition method and the updated propagation impedance data.

[0021] The beneficial effects of this preferred technical solution are as follows: by establishing a gradient detection zone and calculating the color gradient change rate, sensitive detection of minute changes around leaf veins is achieved; it can identify early signs of nutrient deficiency or pests that are difficult to detect with the naked eye, greatly improving the ability of early diagnosis.

[0022] As a preferred embodiment of the image processing-based leaf measurement method of the present invention, the output step of the nutrient deficiency or pest and disease identification mechanism includes:

[0023] Obtain the evaluation score and deviation vector data of the symmetry judgment method;

[0024] Obtain the assessment score and diffusion risk level data of the diffusion pattern judgment method;

[0025] Obtain the evaluation score and propagation impedance update data of the vein-guided gradient recognition method;

[0026] Based on the updated propagation impedance data, adjust the weight ratio of the evaluation scores for the symmetry judgment method and the diffusion mode judgment method.

[0027] The three assessment scores are weighted and calculated to obtain the comprehensive health status assessment score.

[0028] Based on the threshold range set by the comprehensive health status assessment score, the output leaf health status assessment result is healthy, slightly abnormal, moderately abnormal, or severely abnormal.

[0029] This invention provides a system for leaf measurement based on image processing.

[0030] To solve the above-mentioned technical problems, the present invention provides the following technical solution: a system for leaf measurement based on image processing, comprising:

[0031] The feature extraction module is responsible for acquiring leaf images and using image processing technology to extract the morphological features of the leaves from the acquired images.

[0032] The leaf vein skeleton recognition model construction module is used to build a leaf vein skeleton recognition model, and the extracted morphological feature information is input into the model for analysis.

[0033] The nutrient deficiency or pest and disease identification mechanism module outputs the corresponding health status assessment results of the leaves, which is used to determine whether the leaves have nutrient deficiency or pest and disease and their severity.

[0034] As a preferred embodiment of the image processing-based blade measurement system of the present invention, it further includes a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the steps of the image processing-based blade measurement method.

[0035] The beneficial effects of this invention are:

[0036] Firstly, through in-depth analysis of leaf vein structure, particularly the application of midrib feature extraction and lateral vein feature analysis, the internal structural characteristics of leaves can be accurately identified. This provides an important basis for the early identification of nutrient deficiencies or pests and diseases. Compared with traditional surface feature analysis methods, this invention can identify abnormal changes in leaf vein structure before nutrient deficiencies or pest and disease symptoms are obvious, greatly improving the accuracy of early diagnosis. By dividing the midrib into strong, normal, and weak segments, and analyzing the angle, density, and symmetry of lateral veins, the developmental status and health level of leaves can be comprehensively assessed.

[0037] Secondly, the zonal coordinate system established in this invention provides a scientific spatial framework for leaf analysis, dividing the leaf into apical, mid-leaf, and basal regions. Combined with left-right symmetry analysis, this enables refined evaluation of different leaf regions. This zoning method fully considers the physiological structural characteristics and functional differences of leaves, making the detection results more accurate and targeted. By establishing zonal propagation impedance maps, areas prone to nutrient deficiencies or pests and diseases can be accurately located, providing scientific guidance for preventative protection.

[0038] Furthermore, by proposing a method for identifying the transmission pathways of nutrient deficiencies or pests, establishing transmission networks, and setting transmission impedance points, the transmission pathways and spread range of nutrient deficiencies or pests can be accurately predicted. This is of great significance for developing effective control strategies for nutrient deficiencies or pests, and can help agricultural producers take targeted control measures before nutrient deficiencies or pests spread, thereby reducing losses.

[0039] Finally, this invention integrates three complementary mechanisms for identifying nutrient deficiencies or pests: symmetry judgment, diffusion pattern judgment, and vein-guided gradual change identification. Through multi-dimensional and multi-angle comprehensive analysis, it improves the accuracy and reliability of nutrient deficiency or pest identification. This multi-verification mechanism effectively reduces false positives and false negatives, resulting in higher practicality and credibility. The symmetry judgment method can identify progressive changes in nutrient deficiencies or pests through historical data comparison; the diffusion pattern judgment method can accurately analyze the diffusion type and risk level of nutrient deficiencies or pests; and the vein-guided gradual change identification method can dynamically adjust the propagation impedance parameters. The three mechanisms work together to form a complete system for identifying nutrient deficiencies or pests. Attached Figure Description

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0041] Figure 1 This is a flowchart illustrating an image processing-based leaf measurement method according to an embodiment of the present invention.

[0042] Figure 2 This is a schematic diagram illustrating the establishment of a partitioned coordinate system for a leaf measurement method based on image processing, as provided in an embodiment of the present invention.

[0043] Figure 3 This is a schematic diagram illustrating the prediction of nutrient deficiency or pest and disease transmission pathways using an image processing-based leaf measurement method, as provided in one embodiment of the present invention.

[0044] Figure 4 This is a schematic diagram of the output result interface of a leaf measurement method based on image processing provided in an embodiment of the present invention. Detailed Implementation

[0045] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the embodiments described.

[0046] Example 1, referring to Figures 1-4 This is one embodiment of the present invention, which provides a leaf measurement method based on image processing, including the following steps:

[0047] S1. Acquire leaf images and extract morphological feature information from the leaf images.

[0048] The steps for extracting morphological feature information from leaf images include:

[0049] Preprocess the leaf images;

[0050] Extract edge information from the preprocessed blade image, and obtain the complete outline of the blade image based on the edge information;

[0051] Local features of the leaf are obtained within the complete outline of the leaf image;

[0052] The local features of the blade are combined to form complete morphological feature information.

[0053] In this embodiment, the image format is first converted from the original RGB color image to a format suitable for processing. A Gaussian filter is then used to denoise the image, removing random noise generated during image acquisition. Next, illumination homogenization processing is performed to adjust the brightness distribution of the image, ensuring that the brightness differences between different parts of the leaf are controlled within a reasonable range. For areas with shadows or uneven illumination, local contrast enhancement is used.

[0054] In the edge information extraction stage, the color image is converted to a grayscale image. To obtain the complete blade contour, morphological processing is performed on the detected edges, using a 3×3 pixel structuring element for closing operations to connect broken edge segments. Subsequently, the largest connected contour is extracted from the set of edge points as the complete blade contour. The spacing between contour points is between 1 and 2 pixels to ensure the smoothness and accuracy of the contour.

[0055] For acquiring local features of the blade, a gridded sampling method is used for feature extraction within the defined blade outline. The blade region is divided into a 20×20 grid, and the size of each grid is dynamically adjusted according to the actual size of the blade to ensure that the grid covers the entire area of ​​the blade. Multiple local features, including color features, texture features, and shape features, are extracted within each grid.

[0056] Different weighting coefficients are assigned to each grid cell based on its position within the blade. The grid weight in the central region of the blade is set to 1.0, gradually decreasing towards the edges. The weight of the edge grid cells is set between 0.6 and 0.8. This weighting setting highlights the features of the main areas of the blade while preserving edge information. Simultaneously, local feature information of each grid cell is preserved to create a feature map, recording the spatial distribution of each feature on the blade.

[0057] S2. Construct a leaf vein skeleton recognition model, analyze the morphological feature information, and obtain the leaf partition coordinate system.

[0058] The leaf vein skeleton recognition model includes a main vein feature extraction method, a lateral vein feature analysis method, and a nutrient deficiency or pest and disease transmission path recognition method.

[0059] The main vein feature extraction method includes:

[0060] Extract the complete path of the leaf midrib and establish the midrib centerline from the petiole connection point to the leaf tip.

[0061] Measure the width variation at each point on the center line of the main vein to identify the distribution pattern of the main vein's thickness;

[0062] Calculate the offset angle between the midrib and the long axis of the blade to determine whether the midrib is deformed;

[0063] Based on the width distribution and offset angle of the main vein, the main vein is divided into three types: strong segment, normal segment, and weak segment.

[0064] In the process of extracting the midrib feature, the first step is to locate the thickest "vessel" in the leaf image, which is the midrib. Starting from the petiole, where the leaf connects to the branch, trace it all the way to the very tip of the leaf. This process is like drawing a center line on the leaf from bottom to top with a pencil; this line represents the path of the midrib. To ensure the accuracy of this line, its position is automatically adjusted to ensure it always extends along the center of the midrib.

[0065] Next, the thickness of the midrib needs to be measured. Along the drawn center line of the midrib, the width is measured at short intervals. In healthy leaves, the midrib is usually thicker near the petiole and gradually tapers towards the leaf tip; this is the normal growth pattern. If a section suddenly becomes very thin, or if the width changes irregularly, it may indicate a problem in that area. The width data at each measurement point will be recorded, creating a curve showing the width variation.

[0066] At the same time, it's necessary to check if the midrib is growing "straight." Normally, the midrib should grow along the central axis of the leaf. The angle by which the midrib deviates from the ideal straight line will be calculated. If the deviation is too large, it indicates that the midrib may be deformed, which is usually a sign of abnormal growth or external force. Based on the measured width changes and deviation angles, the entire midrib is divided into three types of segments.

[0067] The lateral vein feature analysis method includes:

[0068] Identify all lateral vessels branching off from the main vessel and measure the angle between the lateral vessels and the main vessel;

[0069] Count the number of lateral veins per unit length of main vein and calculate the lateral vein density distribution;

[0070] Compare the number and angle of the lateral veins on the left and right sides to assess the symmetry of the lateral vein distribution;

[0071] Based on the angle, density, and symmetry of the lateral veins, output data assessing the health of lateral vein development;

[0072] Lateral vein feature analysis is similar to examining the branching of a tree. First, you need to locate all the "small blood vessels" branching off from the main vein—the lateral veins. These lateral veins extend outwards from the main vein. The system automatically identifies each lateral vein and measures the angle between it and the main vein. In normal leaves, these angles are typically between 45° and 60°. If the angle is too small or too large, it may indicate a problem with leaf development.

[0073] Next, the density of lateral veins needs to be counted. The system divides the main vein into several small segments, counts the number of lateral veins in each segment, and then calculates the average density. Healthy leaves usually have a relatively even distribution of lateral veins with a moderate density. If there are areas with an unusually large or small number of lateral veins, there may be developmental abnormalities.

[0074] The system also checks the symmetry of the lateral veins on both sides. A normal leaf should be bilaterally symmetrical; the number of lateral veins on the left should be roughly the same as the number on the right, and the angles should be roughly the same. If there is a significant difference between the left and right sides—for example, many lateral veins on the left but few on the right, or a large difference in angle—it indicates that the leaf may have been affected by uneven growth conditions or has some kind of abnormality. Finally, the system comprehensively considers the angle, density, and symmetry of the lateral veins to give a health score.

[0075] The method for identifying nutritional deficiencies or pest transmission pathways includes:

[0076] Using the main vein and lateral veins as the pathway framework, establish a transmission network for nutrient deficiencies or pests and diseases;

[0077] Set a propagation impedance point on the propagation network;

[0078] When an area with nutrient deficiency or pests is detected, its transmission path is calculated based on the location of the nutrient deficiency or pests, and the extent of its spread is predicted.

[0079] In this embodiment, the main vein and all lateral veins are first connected to construct a complete "road network." The main vein is the main road, and the lateral veins are branch roads; nutritional deficiencies or pests will spread along these roads. The "traffic capacity" of each road is marked in this network; thicker veins have stronger traffic capacity, while thinner veins have weaker traffic capacity.

[0080] Within this propagation network, several points of resistance to transmission must be established. These points are placed in relatively weak areas, such as weak sections of the main vein, areas with abnormal lateral vein density, or asymmetrical parts. These areas act like traffic bottlenecks, hindering the spread of nutrient deficiencies or pests. Stronger areas have lower resistance, allowing for faster spread of nutrient deficiencies or pests; weaker areas have higher resistance, resulting in slower spread.

[0081] When chlorosis, yellowing, spots, patches, or other abnormal areas are detected on leaves, a propagation path calculation is initiated. This calculates the most probable path for nutrient deficiency or pests to spread from their origin along the vascular network. Considering the impedance of each path, it predicts that the disease will preferentially spread along paths with low impedance—that is, spread rapidly along healthy blood vessels—while spreading more slowly in weaker areas. Ultimately, a propagation prediction map is generated, showing the extent to which nutrient deficiency or pests may spread in the coming days.

[0082] The steps for establishing the partitioned coordinate system are as follows:

[0083] The leaf is divided into two symmetrical regions, left and right, with the midrib centerline as the central axis.

[0084] Using the lateral veins as the dividing line, the symmetrical area is divided into the leaf tip area, leaf middle area, and leaf base area, and a zoned coordinate system for the leaf is established.

[0085] Based on the distribution of strong, normal, and weak segments of the main vein, and combined with the assessment data on the health of lateral vein development, areas prone to nutrient deficiency or pests and diseases are marked in the leaf's zonal coordinate system.

[0086] The propagation impedance points of the propagation network are matched with the partition coordinate system of the blade to establish a partition propagation impedance diagram.

[0087] In this embodiment, the leaf is first folded in half like origami. Specifically, the center line of the main vein is used as a "central axis," which is like drawing a vertical dividing line in the middle of the leaf, dividing the entire leaf into two symmetrical regions, left and right.

[0088] Next, these regions need to be further subdivided. Using lateral veins as "boundaries," the left and right regions are further divided into smaller functional areas. Specifically, several lateral veins near the leaf tip and several near the leaf base are identified. Using these lateral veins as boundaries, each side region is divided into three parts: the leaf tip region (the top of the leaf), the leaf middle region (the middle part of the leaf), and the leaf base region (the bottom of the leaf). This divides the entire leaf into six sub-regions: the leaf tip region, leaf middle region, and leaf base region on the left, and three corresponding regions on the right. Coordinate markers are then established for each region.

[0089] After completing the basic zoning, "dangerous areas" are marked, much like marking earthquake zones or flood-prone areas on a map. Considering the health of the main vein and the development of the lateral veins, areas prone to nutrient deficiencies or pests and diseases are marked in the zoning coordinate system. Specifically, if the main vein in a certain area is a weak segment, or if the lateral veins in that area are poorly developed (e.g., abnormal density, abnormal angle, asymmetry, etc.), that area is marked as a "prone area." These markings, like marking high-risk areas in red on a map, remind users to pay special attention to these areas. Different warning levels are also assigned to these areas based on the degree of risk: the highest risk areas are marked in dark red, medium risk in orange, and low risk in yellow.

[0090] Finally, a complete "propagation impedance map" needs to be established. Each impedance point is matched with the newly established zonal coordinate system to determine which specific region each impedance point is located in. Then, the average impedance value is calculated for each region. Regions with high impedance values ​​indicate nutrient deficiency or difficulty in pest and disease transmission, while regions with low impedance values ​​indicate easy transmission of pests and diseases. This creates a zonal propagation impedance map, much like a colored contour map, with different colors representing different impedance levels. Green areas indicate high impedance and slow transmission, while red areas indicate low impedance and fast transmission. This impedance map not only shows the ease or difficulty of nutrient deficiency or pest and disease transmission but also, combined with zonal division, allows users to clearly understand the role and risk level of each part of the leaf in the process of nutrient deficiency or pest and disease transmission.

[0091] S3. Based on the nutrient deficiency or pest and disease identification mechanism and combined with the leaf zoning coordinate system, obtain the corresponding health status assessment results of the leaves; the nutrient deficiency or pest and disease identification mechanism outputs the results using the symmetry judgment method, diffusion pattern judgment method, and leaf vein guidance gradient identification method based on the leaf zoning coordinate system.

[0092] The steps of the symmetry judgment method include:

[0093] Obtain the leaf's partition coordinate system and calculate the color difference between the left leaf tip area, leaf middle area, leaf base area and the corresponding right area;

[0094] Establish a blade state history record library to store the color features of the blade's partition coordinate system at different time points;

[0095] A time-weighted decay function is introduced to calculate the deviation vector data between the current state and the historical state;

[0096] When the deviation vector data increases in a certain area, the risk level of nutritional deficiency or abnormal pests and diseases in that area is increased by combining the impedance value of that area in the partitioned propagation impedance map and the assessment data of the health of lateral vein development.

[0097] The acute or chronic abnormality is determined based on the rate of change of the deviation vector data.

[0098] The symmetry judgment results are fed back to the nutrient deficiency or pest and disease transmission path identification method to adjust the transmission impedance point value of the transmission network.

[0099] Output the evaluation score and deviation vector data of the symmetry judgment method.

[0100] In this embodiment, the color differences between corresponding areas on the left and right sides of the leaf are first compared. Specifically, the color of the leaf tip area on the left is compared with that on the right, the middle area on the left is compared with that on the right, and the base area on the left is compared with that on the right. Healthy leaves should typically have similar colors on both sides. If one side shows a significant color change, such as turning yellow, brown, or developing spots, while the other side remains a normal green, this indicates an abnormality. The numerical values ​​of these color differences are then calculated; the larger the difference, the more severe the asymmetry.

[0101] To better assess the health trends of the leaves, a "health record"—a historical database of leaf conditions—was established. This database records the color characteristics of the leaves at different points in time, such as today's color, yesterday's color, and the color a week ago. The leaves are periodically "photographed for health checks," and the results of each check are saved. This allows for observation of trends in the leaf's health. Furthermore, different time-related data are assigned different levels of importance, with the most recent data being the most important and older data decreasing in importance.

[0102] When a color change in a certain area exceeds the normal range, if that area is inherently weak (e.g., a weak segment of the main vein or underdeveloped lateral veins), even a small color change could indicate a significant problem, raising the risk level. The speed of the color change is also used to determine the type of nutrient deficiency or pest / disease abnormality: rapid color changes within a short period may indicate an acute abnormality requiring immediate treatment; slow color changes may indicate a chronic abnormality requiring long-term monitoring. Finally, these assessments are fed back to the transmission path prediction system to adjust the values ​​of the transmission impedance points.

[0103] The steps of the diffusion mode determination method include:

[0104] Identify anomalous color regions in leaf images;

[0105] Match the location of the abnormal color area with the leaf's partition coordinate system to determine the partition to which the abnormal area belongs;

[0106] Based on the partitioned propagation impedance diagram, determine the interaction between the abnormal color regions and the propagation network;

[0107] When the abnormal color area is located in the weak section of the main vein, it is determined to be a high-risk diffusion pattern; when the abnormal color area is located in the strong section of the main vein, it is determined to be a low-risk diffusion pattern.

[0108] Analyze the diffusion direction of the abnormal color region, and combine it with the propagation path predicted by the propagation network to determine whether the diffusion type is along the vein, across the vein, or random diffusion.

[0109] The diffusion mode judgment results are compared and verified with the deviation vector data of the symmetry judgment method;

[0110] Output the assessment score and diffusion risk level data of the diffusion pattern judgment method.

[0111] In this embodiment, the first step is to locate areas with abnormal colors in the leaf image, such as areas with yellowed edges, yellow spots, brown spots, or black dots that are clearly different from normal green. This process is similar to finding colored stains on a green sheet of paper, identifying the location, size, and color characteristics of these "stains." After locating the abnormal areas, it is determined which section of the leaf these areas are located in.

[0112] If the abnormal area appears near a weak section of the main vein, it is considered a high-risk spread pattern, indicating that nutrient deficiency or pests are likely to spread rapidly. Conversely, if the abnormal area appears in a strong section of the main vein, it is considered a low-risk spread pattern, indicating that nutrient deficiency or pests may spread more slowly. The spread of the abnormal area is also observed: if nutrient deficiency or pests spread along the veins, it is called "along-vein spread"; if it spreads across veins, it is called "cross-vein spread"; and if it spreads indiscriminately without a clear direction, it is called "random spread".

[0113] To ensure the accuracy of the judgment, the analysis results of the diffusion pattern are compared and verified with the results of the symmetry judgment. If the conclusions of the two methods are consistent, the judgment is correct; if the conclusions are inconsistent, further analysis is needed to determine the reasons, requiring adjustment of the judgment parameters or more detailed examination. Ultimately, a comprehensive diffusion risk level assessment will be provided, informing users of the likelihood of nutrient deficiency or pest / disease spread and the necessary level of control measures.

[0114] The operation steps of the leaf vein-guided gradient recognition method include:

[0115] A gradient detection band is established along the propagation path of the propagation network, and the width of the gradient detection band is adjusted according to the width of the main vein and the lateral veins;

[0116] Calculate the rate of change of color gradient within the gradient detection zone;

[0117] When the rate of change of the color gradient exceeds a preset threshold, it is marked as a vein-guided gradient region;

[0118] Based on the location of the propagation impedance point, when the leaf vein guiding gradient area appears near the propagation impedance point, the propagation impedance point is determined to be ineffective.

[0119] Update the partitioned propagation impedance map based on the number and distribution of failed propagation impedance points;

[0120] The results of leaf vein guidance gradient recognition are used as weight adjustment factors to correct the evaluation thresholds of the symmetry judgment method and the diffusion pattern judgment method.

[0121] Apply the updated partitioned propagation impedance map to recalculate the propagation paths and speeds of nutrient deficiencies or pests and diseases;

[0122] Output the evaluation score and propagation impedance update data of the leaf vein guidance gradient recognition method.

[0123] In this embodiment, an "observation zone," or gradient detection zone, is established along each leaf vein. The width of this observation zone automatically adjusts according to the thickness of the leaf vein; the observation zone is wider for thicker main veins and narrower for thinner lateral veins. This ensures that the zone covers areas around the leaf veins that may be nutrient deficient or infested with pests and diseases, without including too much irrelevant information.

[0124] Within these observation zones, under normal circumstances, the color around the leaf veins should transition smoothly, like the natural gradation of a rainbow. However, if nutrient deficiency or pests / diseases occur, the color change becomes abrupt, like a clear dividing line suddenly appearing in a rainbow. The "steepness" of this color change is calculated. If the change is too abrupt, exceeding the normal range, it is marked as a "vein-oriented gradual change zone," which is usually an early signal that nutrient deficiency or pests / diseases are beginning to affect leaf vein function.

[0125] When vein-guided gradient areas are detected, their positional relationship with previously set propagation impedance points is also examined. If gradient anomalies occur near impedance points, it indicates that this "checkpoint" may have been breached by nutrient deficiency or pests, and it needs to be marked as "failed." Based on the number and distribution of failed impedance points, the entire propagation impedance map is redrawn, updating the predicted paths and speeds of nutrient deficiency or pest propagation. Simultaneously, the results of this gradient identification are used to adjust the sensitivity of the other two judgment methods. If many early signs of nutrient deficiency or pests are detected, the alert level is raised, making symmetry judgment and diffusion pattern judgment more sensitive for earlier problem detection. Finally, a comprehensive assessment result is output, including the degree of gradient anomalies detected and updated propagation impedance information, providing users with the latest nutrient deficiency or pest risk assessment.

[0126] The output steps of the nutrient deficiency or pest / disease identification mechanism include:

[0127] Obtain the evaluation score and deviation vector data of the symmetry judgment method;

[0128] Obtain the assessment score and diffusion risk level data of the diffusion pattern judgment method;

[0129] Obtain the evaluation score and propagation impedance update data of the vein-guided gradient recognition method;

[0130] Based on the updated propagation impedance data, adjust the weight ratio of the evaluation scores for the symmetry judgment method and the diffusion mode judgment method.

[0131] The three assessment scores are weighted and calculated to obtain the comprehensive health status assessment score.

[0132] Based on the threshold range set by the comprehensive health status assessment score, the output leaf health status assessment result is healthy, slightly abnormal, moderately abnormal, or severely abnormal.

[0133] In this embodiment, the evaluation score from the symmetry assessment method and specific data on left-right asymmetry are obtained. This data tells us whether the left and right sides of the leaf are developing evenly. Simultaneously, the evaluation score and risk level information from the diffusion pattern assessment method are collected. This data indicates how dangerous the discovered abnormal areas are and how nutrient deficiencies or pests may spread. Finally, the evaluation score from the vein-guided gradient identification method and the latest propagation impedance update information are obtained. This data reflects subtle changes around the veins and the latest status of the propagation path.

[0134] After collecting all the data, the weight of each inspection result is determined by adjusting the weight ratio based on the latest propagation impedance update data provided by the vein-guided gradient identification method. If many propagation impedance points are found to be invalid, it indicates nutrient deficiency or strong pest and disease transmission ability, and the result of the diffusion pattern judgment method will be given more importance. If most of the propagation impedance points are still valid, it indicates that the vein structure is relatively stable, and the result of the symmetry judgment method will be more important.

[0135] Next, the three results are combined to obtain a final health score. This process uses a weighted average method. Assuming the symmetry assessment method gives 85 points, the diffusion pattern assessment method gives 70 points, and the vein-guided gradient recognition method gives 75 points, the system will calculate the composite score based on the previously determined weight ratios (e.g., 0.3, 0.4, 0.3). If a certain method has a higher weight, its impact on the final result is greater. In this way, a comprehensive health status assessment score between 0 and 100 is obtained.

[0136] Finally, based on this comprehensive score, a specific health status assessment is given. The system pre-sets four score ranges: 85 points and above is "healthy," indicating that the leaves are in good condition with no obvious problems; 70-85 points is "mildly abnormal," indicating some minor abnormalities that require observation but not major concern; 50-70 points is "moderately abnormal," indicating a clear problem requiring preventative measures; and below 50 points is "severely abnormal," indicating a serious problem requiring immediate emergency treatment. For example, if the calculated comprehensive score is 78 points, the system will output a "mildly abnormal" diagnosis and recommend that the user strengthen monitoring and take appropriate preventative measures. This tiered diagnosis allows users to clearly understand the health status of the leaves and know what level of response is needed.

[0137] In summary, through in-depth analysis of leaf vein structure, particularly the application of midrib feature extraction and lateral vein feature analysis, the internal structural characteristics of leaves can be accurately identified. This provides an important basis for the early identification of nutrient deficiencies or pests and diseases. Compared with traditional surface feature analysis methods, this invention can identify abnormal changes in leaf vein structure before nutrient deficiencies or pest and disease symptoms are obvious, greatly improving the accuracy of early diagnosis. By dividing the midrib into strong, normal, and weak segments, and analyzing the angle, density, and symmetry of lateral veins, the developmental status and health level of leaves can be comprehensively assessed.

[0138] The zonal coordinate system established in this invention provides a scientific spatial framework for leaf analysis, dividing the leaf into apical, mid-leaf, and basal regions. Combined with left-right symmetry analysis, it enables refined assessment of different leaf regions. This zoning method fully considers the physiological structural characteristics and functional differences of leaves, resulting in more accurate and targeted detection results. By establishing zonal propagation impedance maps, areas prone to nutrient deficiencies or pests and diseases can be accurately located, providing scientific guidance for preventative protection.

[0139] By proposing a method for identifying the transmission pathways of nutrient deficiencies or pests and diseases, establishing transmission networks, and setting transmission impedance points, the transmission pathways and spread range of nutrient deficiencies or pests and diseases can be accurately predicted. This is of great significance for developing effective control strategies for nutrient deficiencies or pests and diseases, and can help agricultural producers take targeted control measures before nutrient deficiencies or pests and diseases spread, thereby reducing losses caused by nutrient deficiencies or pests and diseases.

[0140] This invention integrates three complementary mechanisms for identifying nutrient deficiencies or pests: symmetry judgment, diffusion pattern judgment, and vein-guided gradual change identification. Through multi-dimensional and multi-angle comprehensive analysis, it improves the accuracy and reliability of nutrient deficiency or pest identification. This multi-verification mechanism effectively reduces false positives and false negatives, making the system more practical and reliable. The symmetry judgment method can identify progressive changes in nutrient deficiencies or pests through historical data comparison; the diffusion pattern judgment method can accurately analyze the diffusion type and risk level of nutrient deficiencies or pests; and the vein-guided gradual change identification method can dynamically adjust the propagation impedance parameters. The three mechanisms work together to form a complete system for identifying nutrient deficiencies or pests.

[0141] Example 2 is an embodiment of the present invention, which provides a system for leaf measurement based on image processing, comprising:

[0142] The feature extraction module is responsible for acquiring leaf images and using image processing technology to extract the morphological features of the leaves from the acquired images.

[0143] The leaf vein skeleton recognition model construction module is used to build a leaf vein skeleton recognition model, and the extracted morphological feature information is input into the model for analysis.

[0144] The nutrient deficiency or pest and disease identification mechanism module outputs the corresponding health status assessment results of the leaves, which is used to determine whether the leaves have nutrient deficiency or pest and disease and their severity.

[0145] It also includes a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the image processing-based blade measurement method.

[0146] As described above, although the invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the invention itself. Various changes in form and detail may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. A leaf measurement method based on image processing, characterized in that, Includes the following steps: Acquire leaf images and extract morphological feature information from the leaf images; A leaf vein skeleton recognition model is constructed, and the morphological feature information is analyzed to obtain the leaf's partition coordinate system; Based on the leaf identification mechanism and combined with the leaf's partition coordinate system, the health status assessment results corresponding to the leaves are obtained. The blade recognition mechanism outputs information based on the blade's partition coordinate system using a symmetry judgment method, a diffusion pattern judgment method, and a vein-guided gradual change recognition method. The leaf vein skeleton recognition model includes a main vein feature extraction method, a lateral vein feature analysis method, and a color change propagation path recognition method. The main vein feature extraction method includes: Extract the complete path of the leaf midrib and establish the midrib centerline from the petiole connection point to the leaf tip. Measure the width variation at each point on the center line of the main vein to identify the distribution pattern of the main vein's thickness; Calculate the offset angle between the midrib and the long axis of the blade to determine whether the midrib is deformed; Based on the width distribution and offset angle of the main vein, the main vein is divided into three types: strong segment, normal segment, and weak segment; The lateral vein feature analysis method includes: Identify all lateral vessels branching off from the main vessel and measure the angle between the lateral vessels and the main vessel; Count the number of lateral veins per unit length of main vein and calculate the lateral vein density distribution; Compare the number and angle of the lateral veins on the left and right sides to assess the symmetry of the lateral vein distribution; Based on the angle, density, and symmetry of the lateral veins, output data assessing the health of lateral vein development; The color-changing propagation path identification method includes: Using the main vein and lateral veins as the pathway framework, a color-changing propagation network is established; Set a propagation impedance point on the propagation network; When a color-changing area is detected, the propagation path of the color change is calculated based on the location of the color change, and the range of color change propagation is predicted.

2. The leaf measurement method based on image processing as described in claim 1, characterized in that, The steps for extracting morphological feature information from leaf images include: Preprocess the leaf images; Extract edge information from the preprocessed blade image, and obtain the complete outline of the blade image based on the edge information; Local features of the leaf are obtained within the complete outline of the leaf image; The local features of the blade are combined to form complete morphological feature information.

3. The leaf measurement method based on image processing as described in claim 2, characterized in that, The steps for establishing the partitioned coordinate system are as follows: The leaf is divided into two symmetrical regions, left and right, with the midrib centerline as the central axis. Using the lateral veins as the dividing line, the symmetrical area is divided into the leaf tip area, leaf middle area, and leaf base area, and a zoned coordinate system for the leaf is established. Based on the distribution of strong, normal, and weak segments of the main vein, and combined with the assessment data on the health of lateral vein development, areas prone to discoloration are marked in the leaf's zonal coordinate system. The propagation impedance points of the propagation network are matched with the partition coordinate system of the blade to establish a partition propagation impedance diagram.

4. The leaf measurement method based on image processing as described in claim 3, characterized in that, The steps of the symmetry judgment method include: Obtain the leaf's partition coordinate system and calculate the color difference between the left leaf tip area, leaf middle area, leaf base area and the corresponding right area; Establish a blade state history record library to store the color features of the blade's partition coordinate system at different time points; A time-weighted decay function is introduced to calculate the deviation vector data between the current state and the historical state; When the deviation vector data increases in a certain area, the risk level of discoloration in that area is increased by combining the impedance value of that area in the partitioned propagation impedance map and the assessment data of the health of lateral vein development. The acute or chronic discoloration is determined based on the rate of change of the deviation vector data. The symmetry judgment result is fed back to the color-changing propagation path recognition method to adjust the propagation impedance point value of the propagation network; Output the evaluation score and deviation vector data of the symmetry judgment method.

5. The leaf measurement method based on image processing as described in claim 4, characterized in that, The steps of the diffusion mode determination method include: Identify anomalous color regions in leaf images; Match the location of the abnormal color area with the blade's partition coordinate system to determine the partition to which the abnormal color area belongs; Based on the partitioned propagation impedance diagram, determine the interaction between the abnormal color regions and the propagation network; When the abnormal color area is located in the weak section of the main vein, it is determined to be a high-risk diffusion pattern; when the abnormal color area is located in the strong section of the main vein, it is determined to be a low-risk diffusion pattern. Analyze the diffusion direction of the abnormal color region, and combine it with the propagation path predicted by the propagation network to determine whether the diffusion type is along the vein, across the vein, or random diffusion. The diffusion mode judgment results are compared and verified with the deviation vector data of the symmetry judgment method; Output the assessment score and diffusion risk level data of the diffusion pattern judgment method.

6. The leaf measurement method based on image processing as described in claim 5, characterized in that, The operation steps of the leaf vein-guided gradient recognition method include: A gradient detection band is established along the propagation path of the propagation network, and the width of the gradient detection band is adjusted according to the width of the main vein and the lateral veins; Calculate the rate of change of color gradient within the gradient detection zone; When the rate of change of the color gradient exceeds a preset threshold, it is marked as a vein-guided gradient region; Based on the location of the propagation impedance point, when the leaf vein guiding gradient area appears near the propagation impedance point, the propagation impedance point is determined to be ineffective. Update the partitioned propagation impedance map based on the number and distribution of failed propagation impedance points; The results of leaf vein guidance gradient recognition are used as weight adjustment factors to correct the evaluation thresholds of the symmetry judgment method and the diffusion pattern judgment method. Apply the updated partitioned propagation impedance map and recalculate the color-changing propagation path and propagation speed; Output the evaluation score and propagation impedance update data of the leaf vein guidance gradient recognition method.

7. The leaf measurement method based on image processing as described in claim 6, characterized in that, The output steps of the blade recognition mechanism include: Obtain the evaluation score and deviation vector data of the symmetry judgment method; Obtain the assessment score and diffusion risk level data of the diffusion pattern judgment method; Obtain the evaluation score and propagation impedance update data of the vein-guided gradient recognition method; Based on the updated propagation impedance data, adjust the weight ratio of the evaluation scores for the symmetry judgment method and the diffusion mode judgment method. The three assessment scores are weighted and calculated to obtain the comprehensive health status assessment score. Based on the threshold range set by the comprehensive health status assessment score, the output leaf health status assessment result is healthy, slightly abnormal, moderately abnormal, or severely abnormal.

8. A system for leaf measurement based on image processing, employing the leaf measurement method based on image processing as described in claim 1, characterized in that, include: The feature extraction module is responsible for acquiring leaf images and using image processing technology to extract the morphological features of the leaves from the acquired images. The leaf vein skeleton recognition model construction module is used to build a leaf vein skeleton recognition model, and the extracted morphological feature information is input into the model for analysis. The leaf recognition module outputs the health status assessment results of the corresponding leaves, which are used to determine whether the leaves are discolored and the severity of the discoloration.

9. The system for leaf measurement based on image processing as described in claim 8, characterized in that, It also includes a computer-readable storage medium storing computer-executable instructions that, when executed by a processor, implement the steps of the image processing-based blade measurement method according to any one of claims 1 to 7.

Images