Landscaping monitoring system and monitoring method
Through phased multi-view image acquisition and dynamic optimization mechanisms, combined with satellite remote sensing, drone back shooting and double-sided detection of fallen objects, the problems of low efficiency and incomplete identification in landscaping monitoring are solved, and efficient and accurate vegetation health management and early warning are achieved.
Patent Information
- Application Number
- CN202510611819.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-13
- Publication Date
- 2025-08-12
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing landscaping monitoring technology has low efficiency, limited coverage, and insufficient real-time performance, making it difficult to achieve multi-view data coordination and dynamic early warning, resulting in incomplete identification of vegetation diseases.
A staged multi-view image acquisition strategy is adopted, combined with satellite remote sensing, drone surfing and double-sided feature detection of fallen leaves, vegetation index and disease characteristics are extracted through the analysis module, and early warning signals are generated using the weighted fusion module, and the acquisition strategy is dynamically adjusted and weighted, combining lightweight algorithms and distributed storage.
It realizes stratified monitoring from macro to micro, improves the efficiency and accuracy of vegetation health management, reduces resource consumption and computing power burden, enhances the real-time and flexibility of early warnings, and supports personalized garden management.
Smart Images

Figure CN120472321A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of garden monitoring, and in particular to a garden greening monitoring system and a monitoring method. Background Art
[0002] Landscape monitoring technology is a crucial component of urban ecological management. Traditional methods rely primarily on manual surveys and single data sources, resulting in low efficiency, limited coverage, and insufficient real-time performance. In recent years, technological advancements have led to a shift in related patents and research toward intelligent monitoring and multi-source data integration, but significant limitations remain.
[0003] For example, early technologies relied on manual visual inspections or handheld devices such as chlorophyll meters to assess plant health, which was inefficient and highly subjective, making it difficult to achieve large-scale continuous monitoring. Some devices deployed sensors such as soil moisture, temperature, and humidity to obtain environmental parameters, but lacked quantitative analysis of the apparent state of vegetation, resulting in one-sided monitoring results. In the direction of intelligence, existing technologies have disclosed monitoring methods based on satellite imagery. Although they can achieve large-scale vegetation coverage assessments, they are limited by resolution and cannot capture microscopic details of lesions. The update cycle is long, making it difficult to meet the needs of real-time early warning. Existing image monitoring devices often use a single perspective, such as looking down or looking straight ahead, to capture images, resulting in the omission of key information such as trunk epidermal diseases and leaf wilting at lower levels. For example, drones can collect images of tree crowns, but they do not cover the branches below, which affects the early identification of diseases. Summary of the Invention
[0004] To this end, the present invention provides a landscaping monitoring system and a monitoring method to overcome the obvious deficiencies in the prior art in terms of multi-perspective data collaboration, computing power optimization and dynamic early warning mechanism.
[0005] To achieve the above-mentioned object, on the one hand, the present invention provides a landscaping monitoring system, comprising an acquisition module, an analysis module, an identification and selection module, a weighted fusion module and an early warning module; The acquisition module is used to acquire the following image data in stages: First-level images, remote sensing images collected from above the garden via satellite or drone; The second-level images are the overhead combined images of the garden, the front-side images of the fallen leaves, or the back-side images of the fallen leaves, which are selectively collected based on the correlation determination results; The analysis module is used to extract vegetation index, disease targets, leaf health characteristics and fallen plant status characteristics; an identification and selection module configured to: calculate a vegetation index based on the first-level images to identify preliminary risk areas; analyze correlation anomaly indicators of different image data in the preliminary risk areas, including the deviation rate between the vegetation index and historical data and the matching degree between thermal radiation and chlorophyll content; and generate a dynamic acquisition strategy based on the correlation anomaly indicators to trigger the acquisition module to perform targeted acquisition of the second-level images; The weighted fusion module is used to perform adaptive weighted fusion of multi-source features and calculate the comprehensive withering risk level; The warning module is used to generate a warning signal and a warning level.
[0006] As an optimal technical solution for the garden greening monitoring system, the acquisition module's acquisition method for the overhead combined image of the garden includes drone reverse acquisition configuration, image preprocessing, multi-view image fusion and data enhancement. The drone reverse acquisition configuration is equipped with a vertically upward camera and covers the area below the garden. The image preprocessing includes distortion correction and lighting equalization. The multi-view image fusion generates continuous images based on a feature point matching algorithm. The data enhancement extracts detailed features through super-resolution reconstruction.
[0007] As an optimal technical solution for the garden and greening monitoring system, the acquisition module acquires the front and back images of the fallen leaves, including dual-perspective acquisition, triggered acquisition control, image background removal processing and dual-perspective feature fusion steps. The dual-perspective acquisition is to install cameras on the top and bottom of the fallen leaves collector respectively. The triggered acquisition control is synchronously triggered by the infrared sensor to shoot. The image background removal processing is based on the semantic segmentation algorithm to separate the mesh background. The dual-perspective feature fusion generates a comprehensive health index through comparative analysis.
[0008] As an optimal technical solution for the garden and greening monitoring system, the method for determining the correlation anomaly index includes indicator definition and calculation, anomaly judgment, dynamic acquisition strategy generation and strategy optimization iteration steps. The indicator definition and calculation are based on the vegetation index deviation rate and thermal radiation residual. The anomaly judgment distinguishes correlation anomalies from single-dimensional anomalies through threshold comparison. The dynamic acquisition strategy generation triggers directional acquisition. The strategy optimization iteration adjusts the threshold according to historical data.
[0009] As an optimal technical solution for the garden and greening monitoring system, the weighted fusion module generates the wilting risk level in a manner that includes multi-source feature input, adaptive weighted fusion, risk level division and warning signal generation steps. The multi-source feature input includes vegetation index, disease density, leaf health score and fallen object index. The adaptive weighted fusion calculates the comprehensive risk value through a dynamic weight coefficient. The risk level division is based on a preset threshold interval, and the warning signal generation marks the risk area location.
[0010] As an optimal technical solution for the garden greening monitoring system, the garden greening monitoring system further includes a distributed storage module, which stores data according to timestamps and regional labels and updates weight coefficients through incremental learning.
[0011] As an optimal technical solution for the garden and greening monitoring system, the garden and greening monitoring system also includes a wireless communication module, which uses LoRa and NB-IoT dual-mode transmission protocols to encrypt and send early warning signals.
[0012] As an optimal technical solution for the garden and greening monitoring system, the garden and greening monitoring system further includes a user configuration interface module, which provides visual map annotation and custom threshold functions.
[0013] As an optimal technical solution for the garden greening monitoring system, the weight coefficient update method includes data sampling and labeling, feature importance analysis, incremental learning optimization, and weight verification and locking steps. The data sampling and labeling is to extract historical monitoring data and actual withering labels. The feature importance analysis is to generate initial weights using the random forest algorithm. The incremental learning optimization is to dynamically adjust the weights through the online gradient descent method. The weight verification and locking is based on the verification set accuracy improvement threshold.
[0014] In another aspect, the present invention provides a landscaping monitoring method, which is applied to the landscaping monitoring system described in any of the above solutions, comprising: Collect multi-view image data in stages; Analyze multi-source features; Dynamically adjust collection strategies and generate fusion warning signals; Among them, the phased acquisition includes remote sensing image acquisition, selective triggering of local image acquisition and synchronous acquisition of fallen object images; the analysis of multi-source features extracts vegetation index and disease density; the dynamic adjustment of acquisition strategy is based on correlation anomaly indicators; the fusion generates early warning signals and quantifies the risk level through an adaptive weighted algorithm.
[0015] Compared with the existing technology, the beneficial effect of the present invention is that the landscaping monitoring method and system of the present invention significantly improve the efficiency and accuracy of landscaping health management through multi-dimensional technological innovation. First, the system adopts a phased multi-perspective image acquisition strategy, integrating satellite remote sensing data, inverted drone overhead images and double-sided feature detection of dropped objects to form a hierarchical monitoring system from macro to micro. This design breaks through the field of view limitations of traditional single-perspective monitoring. It can not only quickly screen a large area of abnormal vegetation vitality, but also accurately capture the detailed features of trunk diseases, low-lying leaf wilting and microscopic lesions, significantly improving the comprehensiveness and reliability of early wilting identification.
[0016] Furthermore, the system effectively reduces resource consumption and computing power through a dynamic optimization mechanism. Based on the coupled analysis of correlated anomaly indicators, it dynamically triggers a local image acquisition strategy, acquiring refined data only in high-risk areas, significantly reducing the amount of invalid data. Furthermore, by combining lightweight algorithms with regional segmentation computing technology, the system significantly reduces data processing complexity while ensuring monitoring accuracy, enabling the system to adapt to the real-time operation requirements of edge devices and providing flexible deployment solutions for garden scenarios of varying sizes.
[0017] Furthermore, the system's early warning mechanism enhances real-time performance and robustness through closed-loop feedback and adaptive models. Feature importance analysis and incremental learning techniques, based on historical data training, dynamically adjust the fusion weights of multi-source features, enabling the model to adapt to seasonal changes, tree species differences, and environmental fluctuations. Furthermore, a multi-level risk classification mechanism generates graded early warning signals through comprehensive quantitative indicators. Combined with the ability to label risk area locations, this provides intuitive decision-making support for maintenance personnel, significantly improving the efficiency of problem location and resolution.
[0018] Furthermore, the system's modular architecture and user-defined configurations enhance flexibility and scalability. A visual interface supports threshold adjustment, tree species prioritization, and data management, enabling the system to adapt to the personalized needs of diverse garden scenarios. Furthermore, independently deployed data collection, analysis, and early warning modules facilitate functional expansion and seamless integration with other intelligent management platforms, leaving room for compatibility with future technology upgrades.
[0019] Furthermore, from an ecological and economic perspective, this invention effectively reduces the ecological degradation caused by widespread vegetation mortality through early, accurate warnings and targeted maintenance interventions. Furthermore, the monitoring strategy, which optimizes resource allocation, reduces the duplication of manual inspections and equipment maintenance, providing technical support for the sustainable development of urban greening management. BRIEF DESCRIPTION OF THE DRAWINGS
[0020] Figure 1 This is a structural block diagram of a landscaping monitoring system according to an embodiment of the present invention; Figure 2 The figure is a flow chart of a landscaping monitoring method according to an embodiment of the present invention. DETAILED DESCRIPTION
[0021] In order to make the objects and advantages of the present invention more clearly understood, the present invention is further described below in conjunction with embodiments; it should be understood that the specific embodiments described herein are merely used to explain the present invention and are not intended to limit the present invention.
[0022] The preferred embodiments of the present invention are described below with reference to the accompanying drawings. It should be understood by those skilled in the art that these embodiments are only used to explain the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.
[0023] It should be noted that, in the description of the present invention, terms such as "up", "down", "left", "right", "inside", and "outside" indicating directions or positional relationships are based on the directions or positional relationships shown in the accompanying drawings. This is only for the convenience of description and does not indicate or imply that the device or element must have a specific orientation, be constructed and operated in a specific orientation. Therefore, it cannot be understood as a limitation on the present invention.
[0024] Furthermore, it should be noted that, in the description of the present invention, unless otherwise expressly specified or limited, the terms "mounted," "connected," and "connected" should be understood in a broad sense. For example, they may refer to fixed connections, detachable connections, or integral connections; mechanical connections or electrical connections; direct connections or indirect connections through an intermediate medium; and internal communication between two components. Those skilled in the art will understand the specific meanings of the above terms in the present invention based on specific circumstances.
[0025] See also Figure 1 As shown, it is a structural block diagram of the garden greening monitoring system in an embodiment of the present invention. The garden greening monitoring system includes an acquisition module, an analysis module, an identification and selection module, a weighted fusion module and an early warning module; The acquisition module is used to acquire the following image data in stages: First-level images, remote sensing images collected from above the garden via satellite or drone; The second-level images are the overhead combined images of the garden, the front-side images of the fallen leaves, or the back-side images of the fallen leaves, which are selectively collected based on the correlation determination results; The analysis module is used to extract vegetation index, disease targets, leaf health characteristics and fallen plant status characteristics; The identification and selection module is used to: calculate the vegetation index based on the first-level images to identify the preliminary risk area; analyze the correlation anomaly indicators of different image data in the preliminary risk area, including the deviation rate between the vegetation index and historical data, and the matching degree between thermal radiation and chlorophyll content; generate a dynamic acquisition strategy based on the correlation anomaly indicators, and trigger the acquisition module to carry out the targeted acquisition of the second-level images; The weighted fusion module is used to perform adaptive weighted fusion of multi-source features and calculate the comprehensive withering risk level; The early warning module is used to generate early warning signals and warning levels.
[0026] In the above-mentioned embodiments, the efficiency and accuracy of landscaping health management have been significantly improved through multi-dimensional technological innovation. First, the system adopts a phased multi-perspective image acquisition strategy, integrating satellite remote sensing data, inverted drone overhead images, and double-sided feature detection of dropped objects to form a hierarchical monitoring system from macro to micro. This design breaks through the field of view limitations of traditional single-perspective monitoring. It can not only quickly screen large areas of abnormal vegetation vitality, but also accurately capture the detailed features of trunk diseases, low-lying leaf wilting, and microscopic lesions, significantly improving the comprehensiveness and reliability of early wilting identification.
[0027] Specifically, the acquisition module achieves efficient data acquisition through a phased strategy. The first stage of image acquisition utilizes satellites or drones equipped with optical sensors to periodically acquire remote sensing images from above the garden. Multispectral analysis is then used to extract vegetation indices (NDVI and leaf area index in this example) to rapidly screen for areas of abnormal vegetation coverage. This stage completes preliminary risk area labeling through large-scale, low-resolution scanning, reducing the resource consumption of global, high-precision data acquisition. The second stage of targeted image acquisition dynamically triggers local, refined data collection based on correlation anomaly indicators output by the recognition and selection module. For example, when the vegetation index in the preliminary risk area exceeds the limit compared to historical data, and the thermal radiation data deviates significantly from the model-predicted chlorophyll content, the system automatically dispatches an inverted drone to capture a combined overhead image of the garden, or activates a debris collector to simultaneously acquire frontal and back images. By determining spatial-temporal correlations, additional data collection is performed only in high-risk areas, avoiding full data redundancy and significantly saving storage and transmission resources.
[0028] The analysis module extracts key features using a multi-level algorithm. Quantitative indicators such as NDVI and chlorophyll content are generated based on the spectral reflectance of remote sensing images to assess overall vegetation vitality. A lightweight object detection model (YOLO-Lite in this example) is used to analyze overhead garden images to locate tree trunk wormholes, lesions, and yellowing areas on lower leaves. Semantic segmentation is performed on frontal and side images of fallen objects to quantify microscopic parameters such as the percentage of lesion area and the density of discolored areas. Indirect health indicators such as the degree of decay and insect egg attachment are identified using back images of fallen objects. Combined with frontal and side features, these indicators distinguish between natural metabolism and pathological shedding. This process ensures multi-dimensional feature coverage from the macro to the micro scale, enhancing the comprehensiveness of health assessments. The recognition and selection module implements dynamic policy control. The vegetation index from the first-level image is compared with historical data for the same period to calculate the deviation rate. Thermal infrared sensor data and a chlorophyll content inversion model are used to verify the physical correlation between vegetation vitality and thermal radiation. If there is a significant discrepancy between the two, a hidden wilting risk is identified. The dynamic acquisition strategy triggers the second-level image acquisition command based on abnormal indicator thresholds. For example, if the deviation rate exceeds the limit and the thermal radiation residual continues to expand, priority is given to collecting a combination of overhead garden images to verify tree trunk diseases. If only the fallen object index is abnormal, then a double-sided image is acquired in a targeted manner. This strategy reduces misjudgments through a cross-validation mechanism and optimizes resource allocation efficiency.
[0029] Specifically, the acquisition module's acquisition method for the overhead composite image of the garden includes drone reverse acquisition configuration, image preprocessing, multi-perspective image fusion and data enhancement. The drone's reverse acquisition configuration is to carry a vertically upward camera and cover the area below the garden. Image preprocessing includes distortion correction and lighting equalization. Multi-perspective image fusion generates continuous images based on the feature point matching algorithm. Data enhancement extracts detailed features through super-resolution reconstruction.
[0030] Specifically, the acquisition module acquires the front and back images of litter falls in a manner that includes dual-perspective acquisition, triggered acquisition control, image background removal processing, and dual-perspective feature fusion steps. The dual-perspective acquisition involves installing cameras on the top and bottom of the litter collector respectively. The triggered acquisition control is synchronously triggered by an infrared sensor to shoot. The image background removal processing is based on a semantic segmentation algorithm to separate the mesh background. The dual-perspective feature fusion generates a comprehensive health index through comparative analysis.
[0031] Specifically, the method of determining the correlation anomaly index includes indicator definition and calculation, anomaly judgment, dynamic collection strategy generation and strategy optimization iteration steps. The indicator definition and calculation are based on the vegetation index deviation rate and thermal radiation residual. Anomaly judgment distinguishes correlation anomalies from single-dimensional anomalies through threshold comparison. Dynamic collection strategy generation triggers directional collection. Strategy optimization iteration adjusts the threshold according to historical data.
[0032] Specifically, the weighted fusion module generates the wilting risk level in a way that includes multi-source feature input, adaptive weighted fusion, risk level classification, and early warning signal generation steps. The multi-source feature input includes vegetation index, disease density, leaf health score, and fallen object index. The adaptive weighted fusion calculates the comprehensive risk value through dynamic weight coefficients. The risk level classification is based on the preset threshold interval. The early warning signal generates and marks the location of the risk area. At the same time, a distributed storage module is configured. The distributed storage module stores data according to timestamps and regional labels and updates the weight coefficients through incremental learning. The weight coefficient update method includes data sampling and labeling, feature importance analysis, incremental learning optimization, and weight verification and locking steps. Data sampling and labeling is to extract historical monitoring data and actual wilting labels. Feature importance analysis is to generate initial weights using the random forest algorithm. Incremental learning optimization is to dynamically adjust weights through the online gradient descent method. Weight verification and locking are based on the accuracy improvement threshold of the validation set.
[0033] In detail, the weighted fusion module integrates four types of features: vegetation index, disease density, leaf health score and fallen object index, and realizes risk quantification through dynamic weight allocation. The initial weight is determined based on historical data training and adjusted in real time using incremental learning to adapt to seasonal changes and tree species differences. The risk level division is normalized using a preset threshold interval to generate three-level warning signals: low, medium and high, and the location of the risk area is marked to guide the graded response strategy. This module significantly improves the warning accuracy and decision-making efficiency through multi-dimensional data complementarity and dynamic optimization mechanism. Based on the above effects, the distributed storage module is used to store data according to time and regional labels to support long-term trend analysis and rapid retrieval. Combined with the incremental learning mechanism, the weight coefficient is regularly optimized based on new data to ensure that the model's prediction ability continues to increase and avoid performance degradation caused by environmental changes. The closed-loop feedback design of storage and computing gives the system the ability to self-evolve, ensuring the long-term applicability of the technical solution.
[0034] Furthermore, to ensure that it matches the natural changes, the weight coefficient update method realizes model optimization through multi-stage collaboration. The data sampling and annotation stage extracts historical monitoring data and its corresponding actual withering event labels to construct training and validation data sets. The feature importance analysis uses the random forest algorithm to evaluate the contribution of each feature to risk prediction and generate an initial weight distribution scheme. The incremental learning optimization stage uses the online gradient descent method to dynamically adjust the weight parameters to adapt to environmental changes based on the error feedback of real-time collected data and warning results. The weight verification and locking stage uses an independent verification set to test the performance of the adjusted model. Only when the accuracy rate exceeds the set threshold, the new weight parameters are locked and deployed to the warning module. This process balances the stability and adaptability of the model through a closed-loop feedback mechanism to ensure the scientificity and reliability of the weight update.
[0035] Specifically, the landscaping monitoring system also includes a wireless communication module, which uses LoRa and NB-IoT dual-mode transmission protocols to encrypt and send warning signals. It also includes a user configuration interface module, which provides visual map annotation and custom threshold functions.
[0036] See also Figure 2 As shown, this embodiment further provides a landscaping monitoring method, which is applied to the landscaping monitoring system of any of the above solutions, including: Step S1, collecting multi-view image data in stages; Step S2, analyzing multi-source features; Step S3, dynamically adjust the acquisition strategy and generate a fusion warning signal; Among them, the phased acquisition includes remote sensing image acquisition, selective triggering of local image acquisition and synchronous acquisition of fallen object images, analysis of multi-source features to extract vegetation index and disease density, dynamic adjustment of acquisition strategy based on correlation anomaly indicators, and fusion generation of early warning signals to quantify risk levels through adaptive weighted algorithms.
[0037] In the above embodiment, the resource consumption and computing power burden are effectively reduced through the dynamic optimization mechanism. Based on the coupling analysis of correlation anomaly indicators, the local image acquisition strategy is dynamically triggered, and refined data is only acquired for high-risk areas, which greatly reduces the amount of invalid data. At the same time, combined with lightweight algorithms and regional segmentation calculation technology, the complexity of data processing is significantly reduced while ensuring monitoring accuracy, so that the system can adapt to the real-time operation requirements of edge devices and provide flexible deployment solutions for garden scenes of different sizes. Furthermore, the system's early warning mechanism enhances real-time and robustness through closed-loop feedback and adaptive models. Based on the feature importance analysis and incremental learning technology trained on historical data, the fusion weights of multi-source features are dynamically adjusted, so that the model can adapt to seasonal changes, tree species differences and environmental fluctuations. In addition, the multi-level risk division mechanism generates graded early warning signals through comprehensive quantitative indicators, combined with the risk area location marking function, to provide maintenance personnel with intuitive decision support, significantly improving the efficiency of problem location and disposal.
[0038] Building on the aforementioned benefits, modular architecture and user-defined configuration further enhance flexibility and scalability. A visual interface supports threshold adjustment, tree species priority setting, and data management, enabling the system to adapt to the personalized needs of different garden scenarios. At the same time, independently deployed collection, analysis, and early warning modules facilitate functional expansion or seamless integration with other intelligent management platforms, reserving compatibility space for future technological upgrades. Furthermore, from an ecological and economic perspective, through early, accurate early warning and targeted maintenance intervention, the degradation of ecological functions caused by large-scale vegetation death has been effectively reduced. At the same time, monitoring strategies that optimize resource allocation reduce the duplication of manual inspections and equipment maintenance, providing technical support for the sustainable development of urban greening management.
[0039] With the exception of the acquisition module, the modules described in the embodiments of this application can be implemented in software or hardware. The modules described can also be provided in a processor. For example, a processor can be described as including an analysis module, a recognition and selection module, a weighted fusion module, and an early warning module. The names of these modules do not, in some cases, limit the modules themselves.
[0040] The flowcharts and block diagrams in the accompanying drawings illustrate the possible implementation architecture, functions and operations of the devices, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, program segment or a part of code, and the module, program segment or a part of code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order than that marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of the boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based device that performs the specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.
[0041] Thus far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art may make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will fall within the scope of protection of the present invention.
[0042] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that the present invention is susceptible to various modifications and variations. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention are intended to be within the scope of protection of the present invention.
Claims
1. A landscaping monitoring system, characterized in that: It includes acquisition module, analysis module, identification and selection module, weighted fusion module and early warning module; The acquisition module is used to acquire the following image data in stages: First-level images, remote sensing images collected from above the garden via satellite or drone; The second-level images are the overhead combined images of the garden, the front-side images of the fallen leaves, or the back-side images of the fallen leaves, which are selectively collected based on the correlation determination results; The analysis module is used to extract vegetation index, disease targets, leaf health characteristics and fallen plant status characteristics; an identification and selection module, configured to: calculate the vegetation index based on the first-level images to identify preliminary risk areas; and analyze correlation anomaly indicators of different image data in the preliminary risk areas, including the deviation rate between the vegetation index and historical data, and the matching degree between thermal radiation and chlorophyll content; Generating a dynamic acquisition strategy according to the correlation anomaly indicator to trigger the acquisition module to perform directional acquisition of the second-level image; The weighted fusion module is used to perform adaptive weighted fusion of multi-source features and calculate the comprehensive withering risk level; The warning module is used to generate a warning signal and a warning level.
2. The landscaping monitoring system according to claim 1, characterized in that: The acquisition module acquires the combined image of the garden from above in a manner that includes reverse acquisition configuration of a drone, image preprocessing, multi-view image fusion, and data enhancement. The reverse acquisition configuration of the drone is to carry a vertically upward camera and cover the area below the garden. The image preprocessing includes distortion correction and illumination equalization. The multi-view image fusion generates continuous images based on a feature point matching algorithm. The data enhancement extracts detail features through super-resolution reconstruction.
3. The landscaping monitoring system according to claim 1, characterized in that: The acquisition module acquires the front side image and the back side image of the fallen leaves in a manner that includes dual-perspective acquisition, triggered acquisition control, image background removal processing and dual-perspective feature fusion steps. The dual-perspective acquisition is to install cameras on the top and bottom of the fallen leaves collector respectively. The triggered acquisition control is to trigger the shooting synchronously by the infrared sensor. The image background removal processing is based on the semantic segmentation algorithm to separate the mesh background. The dual-perspective feature fusion generates a comprehensive health index through comparative analysis.
4. The landscaping monitoring system according to any one of claims 1 or 2, characterized in that: The method for determining the correlation anomaly index includes the steps of index definition and calculation, anomaly judgment, dynamic acquisition strategy generation and strategy optimization iteration. The index definition and calculation are based on the vegetation index deviation rate and thermal radiation residual. The anomaly judgment distinguishes correlation anomalies from single-dimensional anomalies through threshold comparison. The dynamic acquisition strategy generation triggers directional acquisition. The strategy optimization iteration adjusts the threshold according to historical data.
5. The landscaping monitoring system according to claim 1, characterized in that: The weighted fusion module generates the wilting risk level in a manner that includes multi-source feature input, adaptive weighted fusion, risk level division, and warning signal generation steps. The multi-source feature input includes vegetation index, disease density, leaf health score, and fallen object index. The adaptive weighted fusion calculates the comprehensive risk value through a dynamic weight coefficient. The risk level division is based on a preset threshold interval. The warning signal generation marks the location of the risk area.
6. The landscaping monitoring system according to claim 1, characterized in that: The landscaping monitoring system further includes a distributed storage module, which stores data according to time stamps and regional labels and updates weight coefficients through incremental learning.
7. The landscaping monitoring system according to claim 1, characterized in that: The landscaping monitoring system also includes a wireless communication module, which uses LoRa and NB-IoT dual-mode transmission protocols to encrypt and send early warning signals.
8. The landscaping monitoring system according to claim 1, characterized in that: The landscaping monitoring system further includes a user configuration interface module, which provides visual map annotation and custom threshold functions.
9. The landscaping monitoring system according to claim 5, characterized in that The updating method of the weight coefficient includes data sampling and labeling, feature importance analysis, incremental learning optimization and weight verification and locking steps. The data sampling and labeling is to extract historical monitoring data and actual withering labels. The feature importance analysis is to generate initial weights using the random forest algorithm. The incremental learning optimization is to dynamically adjust the weights through the online gradient descent method. The weight verification and locking is based on the verification set accuracy improvement threshold.
10. A landscaping monitoring method, characterized in that: The landscaping monitoring system according to any one of claims 1 to 9 comprises: Collect multi-view image data in stages; Analyze multi-source features; Dynamically adjust collection strategies and generate fusion warning signals; Among them, the phased acquisition includes remote sensing image acquisition, selective triggering of local image acquisition and synchronous acquisition of fallen object images; the analysis of multi-source features extracts vegetation index and disease density; the dynamic adjustment of acquisition strategy is based on correlation anomaly indicators; the fusion generates early warning signals and quantifies the risk level through an adaptive weighted algorithm.
Citation Information
Cited By
Apple anthrax leaf blight identification five-dimensional data fusion method based on inspection robot
CN120726402A
Remote intelligent building construction monitoring system based on 5G telephone communication and edge computing
CN121037412A
Screening method for testing suitable fertilizing amount of rape
CN121278626A
A method for screening suitable fertilization amount of rape
CN121278626B
Landscape landscaping monitoring method based on image recognition
CN121564552A
