A garden health dynamic evaluation method and system based on multi-source information fusion

Abstract

This invention relates to the field of landscape ecological monitoring technology, and discloses a method and system for dynamic assessment of landscape health based on multi-source information fusion. The method includes: acquiring multi-source raw monitoring data; constructing a unified time-scaled raw sample set and a preprocessed dataset with consistent mechanisms; constructing multi-domain coupled latent decay features; performing dynamic assessment of health evolution; and generating graded early warning judgments and maintenance recommendations. Compared to existing technologies that mainly rely on single image features or manual inspections for static health judgments, especially in cases of latent degradation where tree appearance indicators are normal but internal structure has weakened, this invention addresses the technical problem of failing to achieve early identification and dynamic assessment. By introducing multi-source heterogeneous data fusion, consistent mechanism preprocessing, and a latent decay discrimination mechanism, this application achieves continuous tracking and early warning of the process of trees transitioning from mild physiological imbalance to structural decay.

Description

Technical Field

[0001] This invention relates to the field of landscape ecological monitoring technology, and in particular to a method and system for dynamic assessment of landscape health based on multi-source information fusion. Background Technology

[0002] Currently, the assessment of the health status of garden trees mainly relies on manual inspections or monitoring methods based on a single data source. For example, the health status of trees is judged by visually observing changes in leaf color, crown width, or dead branches, or by analyzing canopy vegetation indices based on drone imagery. While these methods are effective in cases of obvious disease or severe decline, they are essentially static assessment methods based on superficial characteristics and cannot reflect deeper health information such as changes in the internal structure of the tree and the root zone environment.

[0003] For example, in real-world garden settings, some trees may appear normal in terms of leaf color, canopy morphology, and other external indicators, but they may exhibit chronic degeneration phenomena such as decreased vascular tissue efficiency, localized trunk decay or hollowing, and root zone water-salt imbalance. Current technologies, even when incorporating multispectral imagery or simple multi-source data fusion methods, often remain at the level of feature stitching or weighted overlay, lacking a unified modeling of the mechanistic relationships between different modalities and failing to effectively distinguish the hidden degeneration state of "normal appearance but internal weakness."

[0004] Furthermore, most existing technologies rely on single-moment data for health assessment, lacking continuous evolution analysis of tree health status over time. This makes it difficult to identify the critical stages of transition from mild physiological imbalance to structural weakness, resulting in delayed early warning results. Maintenance measures are often only implemented after problems have already emerged, failing to meet the needs of refined and forward-looking landscape maintenance management.

[0005] Therefore, there is an urgent need for a garden health assessment method that can identify internal weakness and dynamically assess health evolution while maintaining normal external tree appearance indicators, in order to improve the accuracy, timeliness, and scientific nature of garden tree health status identification and maintenance decisions. Summary of the Invention

[0006] To address the aforementioned technical shortcomings, the purpose of this invention is to propose a dynamic assessment method for garden health based on multi-source information fusion. This method aims to solve the technical problem that existing technologies mainly rely on single image features or manual inspections for static health judgment, especially under conditions of hidden degradation where tree appearance indicators are normal but internal structure has weakened, making early identification and dynamic assessment impossible.

[0007] To solve the above-mentioned technical problems, the present invention adopts the following technical solution: The present invention provides a method for dynamic assessment of garden health based on multi-source information fusion.

[0008] The method for dynamic assessment of garden health based on multi-source information fusion includes: Step S10: Obtain multi-source raw monitoring data of the target trees during the target assessment period. Based on the multi-source raw monitoring data, use a unified object mapping and unified time benchmark reconstruction mechanism to construct a multi-source raw sample set and output a unified standard raw sample set. ; Step S20: Based on the unified standard original sample set The preprocessing task is performed using a mechanism-consistent preprocessing mechanism, and the output is a mechanism-consistent preprocessed dataset. ; Step S30: Preprocess the dataset based on mechanism consistency A feature construction task is performed using a multi-domain coupling-based mechanism for constructing latent decay features, and the latent decay features are output. ; Step S40: Based on the feature of concealed decay The dynamic evaluation task is performed using an adjacent time series transition discrimination mechanism, and the health evolution status is output. ; Step S50: Based on the health evolution state The system generates graded early warning judgments and maintenance and treatment suggestions, and outputs a set of target tree health early warning results. .

[0009] Preferably, in step S10, the multi-source raw monitoring data includes UAV multispectral canopy image data, trunk stress wave detection data, trunk resistance tomography detection data, shallow soil moisture content data, deep soil moisture content data, soil electrical conductivity data, micrometeorological monitoring data, and maintenance operation record data.

[0010] Preferably, in step S10, multi-source raw monitoring data of the target trees during the target assessment period are obtained. Based on the multi-source raw monitoring data, a unified object mapping and unified time benchmark reconstruction mechanism is used to perform the multi-source raw sample set construction task, and a unified target raw sample set is output. The steps specifically include: Step S101: Obtain multi-source raw monitoring data of the target trees during the target assessment period, and assign a unique TreeID to each type of data in the multi-source raw monitoring data. Step S102: Obtain the sampling time corresponding to the unique identifier TreeID of the target tree. Collection location and sampling level labels Based on the time of data collection Collection location Sampling level labels The original multi-source sample sequence is constructed using the TreeID, the unique identifier of the target tree. Step S103: According to the preset evaluation time window The original multi-source sample sequences are subjected to uniform time resampling and time-stamp alignment to obtain a uniform time-stamped original sample set. , ;in, Indicates time A subset of drone canopy imagery data, Indicates time A subset of tree trunk stress wave data, Indicates time A subset of trunk resistance tomography data, Indicates time A subset of soil salinity data, Indicates time A subset of micro-meteorological data, Indicates time A subset of maintenance operation records.

[0011] Preferably, in step S20, the original sample set is based on a unified standard. The preprocessing task is performed using a mechanism-consistent preprocessing mechanism, and the output is a mechanism-consistent preprocessed dataset. The steps specifically include: Step S201: Analyze the original sample set with unified standards. The data for each modality are subjected to quality checks, specifically including: performing radiometric correction, geometric correction, and canopy region segmentation on UAV multispectral canopy image data; removing abnormal propagation time difference on trunk stress wave detection data; removing contact anomalies and compensating for temperature drift on trunk resistivity tomography data; removing abnormal peaks and imputing missing values ​​on soil water and salt data and micrometeorological data; and performing event structured coding on maintenance operation record data. Step S202: Perform physical quantity normalization processing on the unified standard original sample set after quality verification, and output the basic feature sequence with consistent mechanism; Step S203: Construct a maintenance intervention inhibition factor based on maintenance operation record data, and perform maintenance intervention inhibition correction on the mechanism-consistent basic feature sequences according to the maintenance intervention inhibition factor using an exponential decay weighted correction method, outputting a mechanism-consistent preprocessed dataset. .

[0012] Preferably, step S202, which involves performing physical quantity normalization processing on the unified standard original sample set after quality verification and outputting a basic feature sequence with consistent mechanism, specifically includes: Step S021: Calculate the time for the UAV multispectral canopy image data after quality verification. Normalized Difference Vegetation Index ; Step S2022: Obtain the trunk stress wave propagation distance based on the quality-verified trunk stress wave detection data. With propagation time difference And calculate the propagation speed of stress waves in the tree trunk. ; Step S2023: Calculate the average conductivity of the trunk cross section based on the quality-verified trunk resistance tomography data. ; Step S2024: Calculate the temperature-compensated soil electrical conductivity based on the quality-verified soil electrical conductivity data. ; Step S2025: For the air temperature in the meteorological monitoring data after quality verification... and relative humidity Calculate vapor pressure deficit ; Step S2026: Based on Normalized Difference Vegetation Index Tree trunk stress wave propagation speed Average electrical conductivity of tree trunk cross section Soil electrical conductivity after temperature compensation and vapor pressure deficit Perform physical quantity normalization processing to output a basic feature sequence with consistent mechanism.

[0013] Preferably, in step S202, ;in, Indicates time Near-infrared reflectivity, Indicates time Reflectivity in the red light band; ; ;in, Indicates time The original soil electrical conductivity, Indicates time Soil temperature, Indicates the temperature compensation coefficient; ;in, Indicates temperature The saturated vapor pressure function under the given conditions; ;in, Indicates the cross section of the tree trunk. Each tomographic unit at time... The resistivity value, This indicates the total number of cross-sectional tomographic units.

[0014] Preferably, in step S30, the dataset is preprocessed based on mechanistic consistency. A feature construction task is performed using a multi-domain coupling-based mechanism for constructing latent decay features, and the latent decay features are output. The steps specifically include: Step S301: Preprocess the dataset based on mechanism consistency Normalized Difference Vegetation Index and the preset reference baseline Constructing apparent deviation features , ; Step S302: Preprocess the dataset based on mechanism consistency Tree trunk stress wave propagation speed Average electrical conductivity of tree trunk cross section Soil electrical conductivity after temperature compensation and vapor pressure deficit Construct external weakening features of the tree trunk respectively Characteristics of internal weakness in the tree trunk Characteristics of water and salt stress in the root zone and characteristics of cumulative environmental stress ; Step S303: Based on apparent deviation features External signs of tree trunk weakness Characteristics of internal weakness in the tree trunk Characteristics of water and salt stress in the root zone and characteristics of cumulative environmental stress Constructing the output concealment decay feature set , .

[0015] This invention also provides a dynamic assessment system for garden health based on multi-source information fusion, comprising: The original sample set construction module is used to acquire multi-source original monitoring data of target trees during the target assessment period. Based on the multi-source original monitoring data, it uses a unified object mapping and unified time benchmark reconstruction mechanism to perform the multi-source original sample set construction task and outputs a unified standard original sample set. ; Mechanism-consistent preprocessing module, used for processing based on a unified standard original sample set. The preprocessing task is performed using a mechanism-consistent preprocessing mechanism, and the output is a mechanism-consistent preprocessed dataset. ; Latent decay feature building module for mechanistic consistent preprocessing of datasets. A feature construction task is performed using a multi-domain coupling-based mechanism for constructing latent decay features, and the latent decay features are output. ; The health evolution dynamic assessment module is used for assessment based on latent frailty characteristics. The dynamic evaluation task is performed using an adjacent time series transition discrimination mechanism, and the health evolution status is output. ; The early warning judgment and response generation module is used to determine the health evolution status. The system generates graded early warning judgments and maintenance and treatment suggestions, and outputs a set of target tree health early warning results. .

[0016] The present invention also provides a garden health dynamic assessment device based on multi-source information fusion, comprising: a memory, a processor, and a garden health dynamic assessment program based on multi-source information fusion stored in the memory and executable on the processor. When the garden health dynamic assessment program based on multi-source information fusion is executed by the processor, it implements a garden health dynamic assessment method based on multi-source information fusion.

[0017] The present invention also provides a computer program product, including a dynamic assessment program for garden health based on multi-source information fusion, wherein the dynamic assessment program for garden health based on multi-source information fusion implements the dynamic assessment method for garden health based on multi-source information fusion when executed by a processor.

[0018] The beneficial effects of this invention are as follows: By introducing a multi-source information fusion mechanism of UAV multispectral canopy imagery, trunk stress wave and resistivity tomography detection, soil water and salt parameters and micrometeorological data, and combining it with a unified time reference and mechanism-consistent preprocessing strategy, this invention achieves collaborative expression of different modal data in the same physical semantic space. It can effectively eliminate data deviations caused by maintenance intervention and differences in collection frequency. Compared with existing technical solutions that rely on only a single data source or simple weighted fusion, this invention significantly improves the accuracy and stability of health status assessment of garden trees.

[0019] This invention constructs a multi-domain coupled latent degradation feature model based on appearance features, internal structural features, root zone water and salt features, and environmental stress features. Combined with an adjacent time series transition discrimination mechanism, it realizes dynamic identification and early warning of the process of trees changing from "normal appearance" to "internal degradation". Compared with the existing technology that relies on appearance abnormalities or static judgment at a single moment, it can effectively avoid the problem of delayed identification of latent degradation and improve the foresight and precision of garden maintenance decisions. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the 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.

[0021] Figure 1 This is a flowchart illustrating the first embodiment of a dynamic assessment method for garden health based on multi-source information fusion according to the present invention.

[0022] Figure 2 This is a schematic diagram of an equipment for a dynamic assessment method of garden health based on multi-source information fusion according to the present invention. Detailed Implementation

[0023] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] Example 1: As Figure 1 The diagram shown is a flowchart of the first embodiment of the dynamic assessment method for garden health based on multi-source information fusion of the present invention. The first embodiment of the dynamic assessment method for garden health based on multi-source information fusion of the present invention is presented.

[0025] In the first embodiment, the method for dynamic assessment of garden health based on multi-source information fusion includes: Step S10: Obtain multi-source raw monitoring data of the target trees during the target assessment period. Based on the multi-source raw monitoring data, use a unified object mapping and unified time benchmark reconstruction mechanism to construct a multi-source raw sample set and output a unified standard raw sample set. ; It should be noted that the "multi-source raw monitoring data" in this step refers to the raw data set formed by different acquisition carriers, sampling frequencies, and physical detection mechanisms around the same target tree within the target assessment period. This includes UAV multispectral canopy image data, trunk stress wave detection data, trunk resistivity tomography detection data, shallow soil moisture content data, deep soil moisture content data, soil conductivity data, micrometeorological monitoring data, and maintenance operation record data. The so-called "unified object mapping" refers to binding data from different sources to the same target tree identity object, so that various types of data are no longer isolated discrete information, but form a multimodal observation unit oriented towards a single tree object around the same tree. The so-called "unified time reference reconstruction" refers to remapping data of different time granularities onto the same assessment time axis, in view of the actual situation that the frequency of UAV image acquisition is low, the frequency of micrometeorological acquisition is high, the trunk detection is mostly carried out intermittently, and the maintenance records have event-driven characteristics, thereby constructing a time-series sample set that can be called upon for subsequent unified analysis. This step is not simply about collecting and stacking data, but rather about first solving the fundamental problems of "inconsistent collection objects, inconsistent sampling frequencies, and incomparable time tags," so as to provide reliable input for subsequent consistent mechanism processing.

[0026] Understandably, by first constructing an initial sample set "oriented towards individual trees and a unified time benchmark," problems such as inconsistencies in the objects described by different data sources, time misalignments, and chaotic observation backgrounds can be avoided in the subsequent feature analysis stage. In other words, this step actually completes the "bottom-level data organization work" in the dynamic assessment of garden health, ensuring that each subsequent tree status inference is based on the same object, the same time scale, and the same assessment window. This guarantees that the canopy information, trunk internal information, root zone environmental information, and external meteorological information obtained in subsequent steps can be compared and jointly judged.

[0027] It should be understood that, compared to traditional techniques that commonly use single-inspection recording, single-modal independent analysis, or simply splicing data by date, this step does not mechanically merge data from different modalities. Instead, it eliminates the inherent fragmentation between multi-source heterogeneous information at the data organization level by reconstructing data using a unified object mapping and a unified time base. In traditional solutions, even if UAV imagery, soil moisture content, and micrometeorological information are collected separately, the lack of a unified index around the same individual tree and a unified time scale around the same evaluation window often results in only coarse-grained correlation judgments, making it difficult to support fine-grained analysis goals such as "the hidden decay changes of a certain tree at a specific time period." The setup of this step ensures that each subsequent processing step is based on the same data semantics, making it more suitable for continuously tracking the chronic degradation process of trees.

[0028] Step S20: Based on the unified standard original sample set The preprocessing task is performed using a mechanism-consistent preprocessing mechanism, and the output is a mechanism-consistent preprocessed dataset. ; It should be noted that the "mechanism-consistent preprocessing" in this step means that instead of treating different modal data as numerical inputs and normalizing them uniformly, it combines the physical meaning, acquisition method and actual disturbance characteristics of various types of data, and performs quality verification, physical quantity restoration, anomaly suppression and maintenance intervention correction on them respectively, so as to finally obtain a mechanism-consistent preprocessed dataset that can enter the unified feature analysis stage. Specifically, for UAV multispectral canopy imagery data, radiometric correction, geometric correction, and canopy region segmentation are required to eliminate the influence of flight altitude variations, solar angle variations, and background interference on canopy reflection characteristics. For trunk stress wave detection data, anomalous time difference points in the propagation path must be identified and significantly deviated samples removed to avoid false anomalies caused by probe coupling instability or local surface defects. For trunk resistivity tomography data, contact anomaly removal and temperature drift compensation are required to reduce the influence of electrode contact status and environmental temperature fluctuations on resistivity distribution results. For soil water and salt data and micrometeorological data, anomalous peak removal and missing value imputation are required to form a continuous and stable root zone environmental sequence. For maintenance operation record data, event structure coding is required so that subsequent actions such as pruning, irrigation, fertilization, and pest and disease control can be incorporated as intervention factors into the analysis framework. The final output of this step is not uninterpreted raw monitoring values, but a preprocessed dataset that can truly characterize the physiological and structural state of trees.

[0029] Understandably, by transforming different modal data from "collected values" into "interpretable, comparable, and interconnected mechanistic representation values," the impact of spurious fluctuations caused by equipment errors, environmental disturbances, and maintenance interventions on subsequent judgment results is significantly reduced. In other words, this step not only improves data quality but, more importantly, enhances the "consistency of physiological meaning" and "engineering comparability" of the data. This ensures that the canopy features, trunk internal features, root zone water and salt characteristics, and environmental stress characteristics used in subsequent steps truly have the foundation to jointly participate in the identification of occult weakness. For example, in the monitoring of camphor trees in a scenic area, the target tree, numbered T-023, underwent a heavy pruning operation in early June. Within five days after the pruning, drone imagery showed a significant decrease in canopy coverage; simultaneously, due to artificial irrigation, the shallow soil moisture content increased rapidly in a short period; and when the temperature was high, the trunk resistivity tomography data also showed temperature-driven local resistivity changes. If a traditional simple normalization method were used, it would simultaneously be judged as "canopy degradation, soil improvement, and trunk abnormality," leading to inconsistent conclusions. In this step, pruning events are encoded as short-term apparent inhibitory intervention factors, irrigation events are encoded as short-term root zone humidity increase factors, and trunk resistance data are compensated for temperature drift before participating in subsequent analysis. The resulting mechanism-consistent preprocessed data can better reflect the true state of the tree, thereby avoiding misjudgments caused by human operation and environmental noise.

[0030] Step S30: Preprocess the dataset based on mechanism consistency A feature construction task is performed using a multi-domain coupling-based mechanism for constructing latent decay features, and the latent decay features are output. ; It should be noted that the "mechanism for constructing latent decay features based on multi-domain coupling" in this step refers to building a multi-domain linked feature system across the apparent domain, internal structure domain, root zone water and salt domain, and environmental stress domain, based on a preprocessed dataset with consistent mechanisms. Instead of analyzing single modal features in the canopy, trunk, root zone, or environment in isolation, this system focuses on the specific identification target of latent decay in trees. Specifically, the apparent domain mainly describes the deviation of the canopy appearance and vegetation index from the baseline, characterizing the visible state of the tree surface; the internal structure domain mainly describes the trunk stress wave propagation and resistivity tomography response characteristics, characterizing the internal vascular tissue, xylem density, and potential decay tendency; the root zone water and salt domain mainly describes the moisture matching relationship between shallow and deep soil layers, the degree of salt accumulation, and the differences between upper and lower layers, characterizing the level of continuous stress in the root zone; and the environmental stress domain mainly describes the cumulative effects of external stresses such as high temperature, low humidity, and continuous low rainfall, characterizing the external growth background of the target tree. Building upon this foundation, this step further constructs "latent decay triggering features," which are composite features specifically describing "insignificant external deviations but a continuously increasing degree of internal structural abnormalities." These features are used to capture chronic degenerative states that are difficult to identify using traditional appearance-dominated approaches. The resulting latent decay features are not general health indicators, but rather a set of target features specifically designed for scenarios where "appearance is normal but internal decay is present."

[0031] It should be understood that, compared to traditional techniques that extract health characteristics solely from apparent indicators such as canopy color, leaf area index, and tree tilt, or infer structural risk based on a single trunk inspection result, this step emphasizes the coupling relationship between multiple domains rather than the single-modal feature itself. Traditional solutions often only answer "whether the appearance looks abnormal" or "whether a particular trunk inspection was abnormal," but struggle to answer "whether there are already signs of sustainable degradation internally before the appearance deteriorates." This step significantly enhances the sensitivity to early chronic weakness by unifying apparent information, internal information, root zone information, and environmental information into the framework for constructing latent weakness features, making it particularly suitable for trees that are located in areas with long-term visitor activity, around paved surfaces, or in areas with complex drainage conditions. For example, in a municipal road greenbelt, a London plane tree showed only slight fluctuations in its canopy vegetation index over two consecutive assessment periods, remaining near the average level for the same road section, making it virtually impossible to identify any abnormalities from its appearance. However, its trunk stress wave propagation characteristics showed a gradual deterioration in propagation consistency, resistivity tomography results showed enhanced local conductivity, and the shallow soil moisture content was consistently low while the deep soil moisture content was high, indicating the coexistence of surface compaction and underlying waterlogging. Furthermore, the continuous high temperature and low humidity exacerbated transpiration pressure. In traditional methods, these signals might be considered as slight fluctuations and ignored; however, in this step, these cross-domain features are coupled to construct a hidden decay characteristic, thus enabling the identification that the tree is already in an early decline stage where "the appearance is still acceptable, but the internal and root zone conditions are continuously deteriorating."

[0032] Step S40: Based on the feature of concealed decay The dynamic evaluation task is performed using an adjacent time series transition discrimination mechanism, and the health evolution status is output. ; It should be noted that the "adjacent temporal transition discrimination mechanism" in this step refers to not directly giving a static health conclusion based on the feature values ​​of a single moment, but rather jointly comparing the latent weakening characteristics of the current assessment moment with the corresponding characteristics of the adjacent previous assessment moment. By analyzing the direction, magnitude, and continuity of feature changes, it identifies whether the target tree is transitioning from a state of mild physiological imbalance to a state of structural weakening. In other words, this step focuses not only on "whether there is an abnormality at present," but also on "whether the abnormality is accumulating continuously" and "whether the change has formed a turning trend from weak to strong." In practice, a comprehensive weakening characterization can be constructed based on the multi-domain coupled characteristics of the current moment, and then combined with the corresponding results of the previous moment to construct a health evolution increment, thereby forming a transition discrimination result. The output health evolution status is not a simple health score, but a temporal judgment result that includes the current state, the direction of change, and the rate of change.

[0033] Understandably, by introducing a logic for judging adjacent temporal transitions, the assessment of garden health is elevated from a "static point judgment" to a "dynamic process judgment." This is particularly important for identifying the hidden weakness of trees, because many chronic deteriorations are not manifested through absolute outliers at a single moment, but rather through gradual, continuous small shifts over multiple assessment periods. This step utilizes this continuity between adjacent time periods to identify trees that are not obvious at a single point but show a significant risk accumulation trend over time, thereby improving early warning capabilities. For example, in the lakeside walkway area of ​​a large park, the canopy appearance of a willow tree was relatively normal in both the April and May assessment periods, and ordinary manual inspections did not reveal any obvious problems; however, its internal structural characteristics showed slight abnormalities in April, which worsened further in May, while the difference in moisture between the upper and lower layers of the root zone continued to widen, and the intensity of environmental stress also accumulated. If judged solely based on the characteristics of a single moment in May, it may not have reached the traditional anomaly threshold; however, by adopting the adjacent time series transition discrimination mechanism in this step, it can be found that the tree has been developing in an unfavorable direction for two consecutive cycles, thus identifying it as a healthy evolutionary state transitioning from mild imbalance to moderate weakness, and providing a basis for the next step of early warning output.

[0034] Step S50: Based on the health evolution state The system generates graded early warning judgments and maintenance and treatment suggestions, and outputs a set of target tree health early warning results. .

[0035] It should be noted that the "health evolution status" in this step refers to the information set output from the previous step, including the current degree of hidden weakness of the target tree, the trend of changes in adjacent time periods, and the risk of state transformation; "graded early warning judgment" refers to classifying the target tree into different levels such as normal attention, mild warning, moderate warning, and high-risk warning based on the health evolution status; "maintenance treatment suggestion generation" refers to outputting corresponding on-site verification suggestions, root zone adjustment suggestions, trunk re-measurement suggestions, support and reinforcement suggestions, or key maintenance suggestions for different warning levels. The final output set of target tree health early warning results not only includes the warning level conclusion, but also the core state basis that triggered the conclusion and the corresponding maintenance suggestions, thus forming a treatment result that can be directly used by landscape managers.

[0036] For example, in the monitoring of an ancient tree grove in a scenic area, a ginkgo tree numbered T-041 showed a continuous increase in internal structural weakness and aggravated water and salt stress in the root zone over three consecutive assessment cycles, but the canopy appearance did not change significantly. After this step, the target tree was placed in a moderate warning state, and maintenance recommendations were automatically generated: "Prioritize trunk retesting, check root zone compaction and local water accumulation, and shorten the next reassessment cycle." If its health status continues to deteriorate in the subsequent two assessment cycles, it will be further upgraded to a high-risk warning, and recommendations will be output: "On-site manual verification, and if necessary, support reinforcement and key isolation management." Thus, this step not only outputs warning results but also transforms the identification of hidden weakness into clearly defined and operational management actions, thereby improving the overall effectiveness of health monitoring and maintenance of garden trees.

[0037] Example 2: Furthermore, the present invention provides a dynamic assessment system for garden health based on multi-source information fusion, employing a dynamic assessment method for garden health based on multi-source information fusion as described in the above embodiments, which can solve the technical problem of dynamic assessment of garden health based on multi-source information fusion. The beneficial effects of the dynamic assessment system for garden health based on multi-source information fusion provided by the present invention are the same as those of the dynamic assessment method for garden health based on multi-source information fusion provided in the above embodiments, and other technical features of the dynamic assessment system for garden health based on multi-source information fusion are the same as those disclosed in the methods of the above embodiments, and will not be repeated here.

[0038] Example 3: This invention provides a dynamic assessment device for garden health based on multi-source information fusion. Please refer to... Figure 2A dynamic assessment device for garden health based on multi-source information fusion includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to execute the dynamic assessment method for garden health based on multi-source information fusion described in Embodiment 1 above. The dynamic assessment device for garden health based on multi-source information fusion in this embodiment of the invention may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Description), PMPs (Portable Media Players), and in-vehicle terminals (e.g., in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers. This dynamic assessment device for garden health based on multi-source information fusion is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the invention. A multi-source information fusion-based dynamic assessment device for garden health may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory 1002 or a program loaded from a storage device 1003 into a random access memory 1004. The random access memory 1004 also stores various programs and data required for the operation of the multi-source information fusion-based dynamic assessment device for garden health. The processing unit 1001, the read-only memory 1002, and the random access memory 1004 are interconnected via a bus 1005. An I / O interface 1006 is also connected to the bus. Typically, the following systems can be connected to the I / O interface 1006: input devices 1007 including, for example, a touchscreen, touchpad, keyboard, mouse, image sensor, microphone, accelerometer, gyroscope, etc.; output devices 1008 including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices 1003 including, for example, magnetic tape, hard disk, etc.; and communication devices 1009. The communication device 1009 allows a multi-source information fusion-based dynamic assessment device for garden health to communicate wirelessly or wiredly with other devices to exchange data. While the figure shows one multi-source information fusion-based dynamic assessment device for garden health with various systems, it should be understood that implementation or possession of all the systems shown is not required. More or fewer systems may be implemented alternatively.

[0039] Example 4: This invention also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the above-described method for dynamic assessment of garden health based on multi-source information fusion. The computer program product provided by this invention can solve the technical problem of dynamic assessment of garden health based on multi-source information fusion. Compared with the prior art, the beneficial effects of the computer program product provided by this invention are the same as those of the method for dynamic assessment of garden health based on multi-source information fusion provided in the above embodiments, and will not be repeated here.

[0040] In particular, according to the embodiments disclosed in this invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of this invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this invention.

[0041] It should be understood that the various parts disclosed in this invention can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics may be combined in any suitable manner in one or more embodiments or examples.

[0042] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for dynamic assessment of garden health based on multi-source information fusion, characterized in that, The methods include: Step S10: Obtain multi-source raw monitoring data of the target trees during the target assessment period. Based on the multi-source raw monitoring data, use a unified object mapping and unified time benchmark reconstruction mechanism to construct a multi-source raw sample set and output a unified standard raw sample set. ; Step S20: Based on the unified standard original sample set The preprocessing task is performed using a mechanism-consistent preprocessing mechanism, and the output is a mechanism-consistent preprocessed dataset. ; Step S30: Preprocess the dataset based on mechanism consistency A feature construction task is performed using a multi-domain coupling-based mechanism for constructing latent decay features, and the latent decay features are output. ; Step S40: Based on the feature of concealed decay The dynamic evaluation task is performed using an adjacent time series transition discrimination mechanism, and the health evolution status is output. ; Step S50: Based on the health evolution state The system generates graded early warning judgments and maintenance and treatment suggestions, and outputs a set of target tree health early warning results. .

2. The method for dynamic assessment of garden health based on multi-source information fusion as described in claim 1, characterized in that, In step S10, the multi-source raw monitoring data includes UAV multispectral canopy image data, trunk stress wave detection data, trunk resistance tomography detection data, shallow soil moisture content data, deep soil moisture content data, soil electrical conductivity data, micro-meteorological monitoring data, and maintenance operation record data.

3. The method for dynamic assessment of garden health based on multi-source information fusion as described in claim 1, characterized in that, In step S10, multi-source raw monitoring data of the target trees during the target assessment period are obtained. Based on the multi-source raw monitoring data, a unified object mapping and unified time benchmark reconstruction mechanism is used to perform the multi-source raw sample set construction task, and output the unified standard raw sample set. The steps specifically include: Step S101: Obtain multi-source raw monitoring data of the target trees during the target assessment period, and assign a unique TreeID to each type of data in the multi-source raw monitoring data. Step S102: Obtain the sampling time corresponding to the unique identifier TreeID of the target tree. Collection location and sampling level labels Based on the time of data collection Collection location Sampling level labels The original multi-source sample sequences were constructed using the TreeID, a unique identifier for the target tree. Step S103: According to the preset evaluation time window The original multi-source sample sequences are subjected to uniform time resampling and time-stamp alignment to obtain a uniform time-stamped original sample set. , ;in, Indicates time A subset of drone canopy imagery data, Indicates time A subset of tree trunk stress wave data, Indicates time A subset of trunk resistance tomography data, Indicates time A subset of soil salinity data, Indicates time A subset of micro-meteorological data, Indicates time A subset of maintenance operation records.

4. The method for dynamic assessment of garden health based on multi-source information fusion as described in claim 2, characterized in that, In step S20, based on the unified standard original sample set The preprocessing task is performed using a mechanism-consistent preprocessing mechanism, and the output is a mechanism-consistent preprocessed dataset. The steps specifically include: Step S201: Analyze the original sample set with unified standards. The data for each modality are subjected to quality checks, specifically including: performing radiometric correction, geometric correction, and canopy region segmentation on UAV multispectral canopy image data; removing abnormal propagation time difference on trunk stress wave detection data; removing contact anomalies and compensating for temperature drift on trunk resistivity tomography data; removing abnormal peaks and imputing missing values ​​on soil water and salt data and micrometeorological data; and performing event structured coding on maintenance operation record data. Step S202: Perform physical quantity normalization processing on the unified standard original sample set after quality verification, and output the basic feature sequence with consistent mechanism; Step S203: Construct a maintenance intervention inhibition factor based on maintenance operation record data, and perform maintenance intervention inhibition correction on the mechanism-consistent basic feature sequences according to the maintenance intervention inhibition factor using an exponential decay weighted correction method, outputting a mechanism-consistent preprocessed dataset. .

5. The method for dynamic assessment of garden health based on multi-source information fusion as described in claim 1, characterized in that, Step S202, which involves performing physical quantity normalization on the unified standard original sample set after quality verification and outputting a consistent basic feature sequence, specifically includes: Step S021: Calculate the time for the UAV multispectral canopy image data after quality verification. Normalized Difference Vegetation Index ; Step S2022: Obtain the trunk stress wave propagation distance based on the quality-verified trunk stress wave detection data. With propagation time difference And calculate the propagation speed of stress waves in the tree trunk. ; Step S2023: Calculate the average conductivity of the trunk cross section based on the quality-verified trunk resistance tomography data. ; Step S2024: Calculate the temperature-compensated soil electrical conductivity based on the quality-verified soil electrical conductivity data. ; Step S2025: For the air temperature in the meteorological monitoring data after quality verification... and relative humidity Calculate vapor pressure deficit ; Step S2026: Based on Normalized Difference Vegetation Index Tree trunk stress wave propagation speed Average electrical conductivity of tree trunk cross section Soil electrical conductivity after temperature compensation and vapor pressure deficit Perform physical quantity normalization processing to output a basic feature sequence with consistent mechanism.

6. The method for dynamic assessment of garden health based on multi-source information fusion as described in claim 5, characterized in that, In step S202, ;in, Indicates time Near-infrared reflectivity, Indicates time Reflectivity in the red light band; ; ;in, Indicates time The original soil electrical conductivity, Indicates time Soil temperature, Indicates the temperature compensation coefficient; ;in, Indicates temperature The saturated vapor pressure function under the given conditions; ;in, Indicates the cross section of the tree trunk. Each tomographic unit at time... The resistivity value, This indicates the total number of cross-sectional tomographic units.

7. The method for dynamic assessment of garden health based on multi-source information fusion as described in claim 5, characterized in that, In step S30, the dataset is preprocessed based on mechanism consistency. A feature construction task is performed using a multi-domain coupling-based mechanism for constructing latent decay features, and the latent decay features are output. The steps specifically include: Step S301: Preprocess the dataset based on mechanism consistency Normalized Difference Vegetation Index and the preset reference baseline Constructing apparent deviation features , ; Step S302: Preprocess the dataset based on mechanism consistency Tree trunk stress wave propagation speed Average electrical conductivity of tree trunk cross section Soil electrical conductivity after temperature compensation and vapor pressure deficit Construct external weakening features of the tree trunk respectively Characteristics of internal weakness in the tree trunk Characteristics of water and salt stress in the root zone and characteristics of cumulative environmental stress ; Step S303: Based on apparent deviation features External signs of tree trunk weakness Characteristics of internal weakness in the tree trunk Characteristics of water and salt stress in the root zone and characteristics of cumulative environmental stress Constructing the output concealment decay feature set , .

8. A dynamic assessment system for garden health based on multi-source information fusion, applied to the dynamic assessment method for garden health based on multi-source information fusion as described in any one of claims 1 to 7, characterized in that, The garden health dynamic assessment system based on multi-source information fusion includes: The original sample set construction module is used to acquire multi-source original monitoring data of target trees during the target assessment period. Based on the multi-source original monitoring data, it uses a unified object mapping and unified time benchmark reconstruction mechanism to perform the multi-source original sample set construction task and outputs a unified standard original sample set. ; Mechanism-consistent preprocessing module, used for processing based on a unified standard original sample set. The preprocessing task is performed using a mechanism-consistent preprocessing mechanism, and the output is a mechanism-consistent preprocessed dataset. ; Latent decay feature building module for mechanistic consistent preprocessing of datasets. A feature construction task is performed using a multi-domain coupling-based mechanism for constructing latent decay features, and the latent decay features are output. ; The health evolution dynamic assessment module is used for assessment based on latent frailty characteristics. The dynamic evaluation task is performed using an adjacent time series transition discrimination mechanism, and the health evolution status is output. ; The early warning judgment and response generation module is used to determine the health evolution status. The system generates graded early warning judgments and maintenance and treatment suggestions, and outputs a set of target tree health early warning results. .

9. A dynamic assessment device for garden health based on multi-source information fusion, characterized in that, The garden health dynamic assessment device based on multi-source information fusion includes: a memory, a processor, and a garden health dynamic assessment program based on multi-source information fusion stored in the memory and executable on the processor. When the garden health dynamic assessment program based on multi-source information fusion is executed by the processor, it implements a garden health dynamic assessment method based on multi-source information fusion as described in any one of claims 1 to 7.

10. A computer program product, characterized in that, The computer program product includes a dynamic assessment program for garden health based on multi-source information fusion. When the dynamic assessment program for garden health based on multi-source information fusion is executed by the processor, it implements a dynamic assessment method for garden health based on multi-source information fusion as described in any one of claims 1 to 7.

Images