A natural resource space monitoring system and method based on multi-source remote sensing fusion

Abstract

This invention relates to the field of multi-source remote sensing fusion technology, and in particular provides a spatial monitoring system and method for natural resources based on multi-source remote sensing fusion. The system includes a heterogeneous feature collaborative extraction subsystem, a land cover type fine identification subsystem, and a natural resource dynamic monitoring subsystem. The method involves acquiring multiple remote sensing data covering the same area, extracting the geometric edges, spectral absorption indices, and radar polarization decomposition parameters of land cover features, performing spatial registration and fusion to generate a collaborative feature map containing land cover material, structure, and polarization characteristics; inputting the collaborative feature map into a pre-constructed land cover feature reference library to output a fine land cover type map; and time-series overlaying of fine land cover type maps acquired at different times to generate a natural resource dynamic monitoring and early warning map, displaying the changing trends and risk levels of various land cover types. This invention improves the accuracy, efficiency, and decision support capabilities of natural resource monitoring.

Description

Technical Field

[0001] This invention relates to the field of multi-source remote sensing fusion technology, and in particular to a spatial monitoring system and method for natural resources based on multi-source remote sensing fusion. Background Technology

[0002] With the increasing urgency of global natural resource management needs, traditional monitoring methods are no longer sufficient to meet the requirements of high-precision, dynamic, and comprehensive spatial monitoring. Currently, natural resource monitoring mainly relies on single remote sensing data sources or manual field surveys, which suffers from limited spatiotemporal resolution, lagging data updates, and insufficient integration of multi-dimensional information. Especially in areas such as forest cover monitoring, land use change analysis, and dynamic water resource assessment, how to achieve intelligent fusion and collaborative interpretation of multi-scale, multi-temporal, and multi-modal remote sensing data has become a key technological bottleneck for improving the effectiveness of natural resource management.

[0003] Prior art 1, Chinese Patent Application No. 202511543521.6 discloses a remote sensing image processing method and system for natural resource management, including the following steps: acquiring multi-source remote sensing image data and auxiliary data of the target natural resource management area, wherein the multi-source remote sensing image data are optical remote sensing images and SAR remote sensing images of the target natural resource management area at different times, and performing standardized preprocessing on the multi-source remote sensing image data. While multi-source remote sensing image processing fully incorporates the specific characteristics of natural resources such as cultivated land, forest land, and water bodies in the target area, and through targeted feature extraction and adaptation fusion, it accurately preserves the key features of various land features in multi-source remote sensing images. This enables more accurate land feature identification in natural resource management, avoids misjudgments caused by feature loss, and provides reliable image processing results and more accurate image support for basic operations such as cultivated land boundary delineation, forest coverage identification, and watershed delineation. However, the lack of a multi-scale spatial-spectral joint transformation operator prevents the simultaneous fusion of geometric edges, spectral absorption index, and radar polarization decomposition parameters at multiple scales. Consequently, the feature maps do not comprehensively describe the essential attributes of natural resources, such as rock mineral composition, forest vertical structure, and soil dielectric properties, limiting the refinement of classification results at the material and structural attribute levels.

[0004] Prior art two, Chinese patent application number 202511457138.9, discloses a method and system for multi-task information extraction from remote sensing images based on a sustainable learning model. The method includes: collecting multi-source remote sensing data of a natural resource scene to be identified; constructing a multi-task sustainable learning model based on a dynamic feature fusion mechanism and an adaptive task topology network; and extracting multi-task category information from remote sensing images based on the multi-source remote sensing data and the multi-task sustainable learning model to achieve the identification of the natural resource scene to be identified. Although the construction of a dynamic feature fusion mechanism and an adaptive task topology network enables the autonomous evolution and knowledge accumulation of the multi-task identification model for remote sensing images, as well as the intelligent understanding and identification of natural scenes, it lacks feature interpretability based on physical mechanisms and has not established a reference library of ground feature characteristics directly related to measured electromagnetic wave response data. It is difficult to quantitatively correlate remote sensing features with specific ground material, such as tree species, rock type, and structural parameters, such as tree height and canopy closure. Furthermore, it is impossible to correct the physical consistency of isolated pixels through feature matching and spatial context optimization, resulting in insufficient robustness and physical reliability of the classification results in complex natural resource scenes.

[0005] Prior art three, Chinese patent application number 202510047021.7, discloses a method, system, medium, and equipment for dynamic monitoring of natural resources based on multi-source data fusion, including: acquiring satellite remote sensing data, aerial remote sensing data, and ground monitoring data; fusing the satellite remote sensing data and aerial remote sensing data to obtain fused data; integrating the fused data with ground monitoring data to obtain comprehensive monitoring data; classifying land features based on deep learning to obtain land feature classification results, which include cultivated land, construction land, water bodies, forest land, and grassland; and optimizing the land feature classification results based on a spatiotemporal logical consistency algorithm to obtain optimized classification results and land feature change information. Although the fusion of multi-source data and optimization of natural resource classification, combined with deep learning and spatiotemporal logical consistency algorithms, have enabled dynamic monitoring of natural resources and improved monitoring accuracy, providing reliable data support for natural resource management and environmental protection, they have not achieved refined land cover classification that integrates material and structural attributes. Furthermore, the dynamic monitoring process has not incorporated topographic slope and meteorological observation data for stability analysis and early warning grading. Specifically, it cannot output classification results with clear attributes such as mature deciduous broad-leaved forest, material: lignocellulose, structure: average tree height 20 meters, canopy closure 0.8, nor has it constructed a stability evaluation model for change areas and an early warning level classification mechanism, thus limiting its application depth in natural resource risk early warning and decision support.

[0006] Current technologies 1, 2, and 3 fail to address how to achieve refined identification of natural resource land cover types at the material and structural attribute levels through collaborative feature extraction of physical mechanisms from multi-source remote sensing data, matching with measured land cover feature databases, and spatial context optimization. Furthermore, they fail to address the challenge of constructing stability analysis and early warning classification models by combining topographic and meteorological data, thus forming an integrated technical system from feature fusion and land cover identification to dynamic risk monitoring. This would improve the physical interpretability, classification refinement, and dynamic early warning capabilities of natural resource spatial monitoring. Therefore, this invention provides a natural resource spatial monitoring system and method based on multi-source remote sensing fusion. Summary of the Invention

[0007] To achieve the above objectives, the present invention adopts the following technical solution:

[0008] In one aspect, the present invention provides a spatial monitoring system for natural resources based on multi-source remote sensing fusion, comprising:

[0009] The heterogeneous feature collaborative extraction subsystem is configured to acquire multiple remote sensing data covering the same area, including high spatial resolution panchromatic band data, multispectral reflectance data, synthetic aperture radar backscattering data, and lidar point cloud data. It performs geometrical fine correction and radiometric normalization on these multiple remote sensing data to eliminate sensor differences and atmospheric effects. By constructing a multi-scale spatial-spectral joint transformation operator, it decomposes the geometrically corrected and radiometrically normalized remote sensing data at multiple spatial scales, extracting the geometric edges, spectral absorption indices, and radar polarization decomposition parameters of ground features. Spatial registration and fusion are then performed to generate a collaborative feature map containing ground feature material, structure, and polarization characteristics.

[0010] The fine-grained land cover type identification subsystem is configured to input the collaborative feature map into a pre-built land cover feature reference library. The land cover feature reference library consists of electromagnetic wave response data of typical land cover obtained from field measurements and laboratory measurements, covering feature vectors of different tree species, rocks and minerals, and soil textures. Through feature matching, each pixel in the collaborative feature map is compared with the feature vector in the land cover feature reference library, the similarity is calculated, and the land cover type is determined according to the maximum likelihood principle. At the same time, spatial context information is used to correct isolated pixels, and a fine-grained land cover type map is output, in which each type of land cover corresponds to a clear material and structural attribute.

[0011] The natural resource dynamic monitoring subsystem is configured to overlay detailed land cover type maps acquired at different time phases, extract areas where types have changed, and calculate the area and boundaries of the changed areas; combine topographic slope data and meteorological observation data to perform stability analysis on the changed areas, identify areas of concentrated change through spatial clustering, and classify the areas of concentrated change into early warning levels according to the rate of change and the degree of potential impact, generating a natural resource dynamic monitoring early warning map to show the changing trends and risk levels of various land cover types.

[0012] Another aspect of the present invention provides a spatial monitoring method for natural resources based on multi-source remote sensing fusion, which includes the following steps:

[0013] Step 1: Acquire multiple remote sensing data covering the same area, including high spatial resolution panchromatic band data, multispectral reflectance data, synthetic aperture radar backscattering data, and lidar point cloud data; perform geometrical fine correction and radiometric normalization on the multiple remote sensing data to eliminate sensor differences and atmospheric effects; construct a multi-scale spatial-spectral joint transformation operator to decompose the geometrically corrected and radiometrically normalized remote sensing data at multiple spatial scales, extract the geometric edges, spectral absorption index, and radar polarization decomposition parameters of ground objects, perform spatial registration and fusion, and generate a collaborative feature map containing ground object material, structure, and polarization characteristics;

[0014] Step 2: Input the collaborative feature map into a pre-built land cover feature reference library. The land cover feature reference library consists of electromagnetic wave response data of typical land cover obtained from field measurements and laboratory measurements, covering feature vectors of different tree species, rocks and minerals, and soil textures. Through feature matching, each pixel in the collaborative feature map is compared with the feature vector in the land cover feature reference library, the similarity is calculated, and the land cover type is determined according to the maximum likelihood principle. At the same time, spatial context information is used to correct isolated pixels, and a detailed land cover type map is output. Each type of land cover in the map corresponds to a clear material and structural attribute.

[0015] Step 3: Overlay the detailed land cover type maps acquired at different time phases to extract areas where types have changed, and calculate the area and boundaries of these areas. Combine topographic slope data and meteorological observation data to conduct stability analysis on the changed areas. Identify areas of concentrated change using spatial clustering methods, and classify these areas into early warning levels based on the rate of change and the degree of potential impact. Generate a dynamic monitoring and early warning map of natural resources to show the changing trends and risk levels of various land cover types.

[0016] This invention achieves systematic monitoring of the spatial distribution and dynamic changes of natural resources by integrating multi-source remote sensing data. The heterogeneous feature collaborative extraction subsystem integrates multi-scale spatial-spectral joint transformation and multi-source data registration to effectively extract complementary features of land features in geometric, spectral, and polarimetric dimensions, overcoming the limitations of single data sources and generating a collaborative feature map with high information density, laying the foundation for refined identification. The land feature type refined identification subsystem performs feature matching and spatial context optimization based on a measured electromagnetic wave response database, significantly improving the accuracy of land feature classification and the ability to interpret physical attributes, outputting a refined land feature type map that combines material and structural attributes. The natural resource dynamic monitoring subsystem, through time-series overlay and multi-source data fusion analysis, achieves quantitative extraction, stability assessment, and risk classification of natural resource changes, forming spatially clear dynamic monitoring and early warning results. The collaborative work of these subsystems ultimately constructs an integrated technology chain from feature fusion and type identification to dynamic assessment, improving the accuracy, efficiency, and decision support capabilities of natural resource monitoring. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings:

[0018] Figure 1 This is a block diagram of the natural resource spatial monitoring system based on multi-source remote sensing fusion provided in Embodiment 1 of the present invention;

[0019] Figure 2 This is a schematic diagram of the spatial monitoring system for natural resources based on multi-source remote sensing fusion provided in Embodiment 1 of the present invention.

[0020] Figure 3 This is a block diagram of the heterogeneous feature collaborative extraction subsystem provided in Embodiment 2 of the present invention;

[0021] Figure 4 This is a block diagram of the fine-grained land cover type identification subsystem provided in Embodiment 6 of the present invention;

[0022] Figure 5 This is a block diagram of the natural resource dynamic monitoring subsystem provided in Embodiment 8 of the present invention;

[0023] Figure 6 This is a flowchart of the spatial monitoring method for natural resources based on multi-source remote sensing fusion provided in Embodiment 14 of the present invention;

[0024] Figure 7 A block diagram of the electronic device provided by the present invention. Detailed Implementation

[0025] The technical solutions of the present invention will now be described with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments.

[0026] Hereinafter, the terms "first," "second," etc., are used for descriptive convenience only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined as "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "multiple" means two or more.

[0027] In this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, a connection can be a fixed mechanical connection, a detachable mechanical connection, or an integral part; or, a connection can be a direct connection or an indirect connection through an intermediate medium. Furthermore, unless otherwise explicitly specified and limited, the term "coupling" should be interpreted broadly. For example, coupling can be a direct electrical connection, such as physical contact and electrical conduction between two components; it can also be understood as an electrical connection between different components in a circuit structure through physical lines capable of transmitting electrical signals, such as copper foil or wires on a printed circuit board (PCB), to transmit electrical signals; or, coupling can be an indirect electrical connection between two components through an intermediate medium; or, coupling can be an electrical connection between two components in a non-contact manner, such as an electrical connection between two components using capacitive coupling to transmit electrical signals.

[0028] In this embodiment of the invention, directional terms such as up, down, left, and right may be defined relative to the orientation of the components shown in the accompanying drawings. It should be understood that these directional terms can be relative concepts, used for relative description and clarification, and can change accordingly depending on the orientation of the components in the accompanying drawings.

[0029] Example 1: As Figure 1 As shown, this embodiment of the invention provides a spatial monitoring system for natural resources based on multi-source remote sensing fusion, comprising:

[0030] The heterogeneous feature collaborative extraction subsystem is configured to acquire multiple remote sensing data covering the same area, including high spatial resolution panchromatic band data, multispectral reflectance data, synthetic aperture radar backscattering data, and lidar point cloud data. It performs geometrical fine correction and radiometric normalization on these multiple remote sensing data to eliminate sensor differences and atmospheric effects. By constructing a multi-scale spatial-spectral joint transformation operator, it decomposes the geometrically corrected and radiometrically normalized remote sensing data at multiple spatial scales, extracting the geometric edges, spectral absorption indices, and radar polarization decomposition parameters of ground features. Spatial registration and fusion are then performed to generate a collaborative feature map containing ground feature material, structure, and polarization characteristics.

[0031] The operator expressions involved in constructing the multi-scale spatial-spectral joint transformation operator are as follows:

[0032]

[0033] In the formula, the left side Indicates spatial location ( ), spectral band and scale parameters The generated collaborative feature values ​​are dimensionless fusion feature strengths; the first term on the right. The scale is The Gaussian kernel function has the dimensionless spatial action weights; the second term Indicates the input image The second Laplacian derivative in the spatial domain, with dimensions representing the rate of change of the image grayscale gradient, is consistent with the unit of image intensity; the third term... For direction The joint spatial-spectral transform kernel function on the above, with dimensions of spatial direction modulation coefficients (dimensionless); symbol This represents the convolution operation. Represented at different scales and direction The summation of all terms; the overall dimension after combination is the image intensity unit. This expression is constructed based on the principle of multi-scale geometric analysis and spatial-spectral information synergy, with the input image... This represents multi-source remote sensing data after geometric correction and radiometric normalization, where ( ) represents the spatial location of ground features, Corresponding spectral band or radar polarization channel. Gaussian kernel function. At different scales The image is smoothed to form a scale-space representation, which is used to separate the structural information of ground features at different scales. Laplacian operator. Extracting regions of abrupt grayscale changes in an image, i.e., the geometric edges and texture details of ground features. Spatial-spectral joint transform kernel function. In direction The extracted edge information is modulated, and the kernel function is composed of the spectral absorption index and radar polarization decomposition parameters, enabling the operator to simultaneously respond to the spectral and polarization scattering characteristics of ground features. Through multi-scale decomposition, the operator extracts the overall structure of ground features at a coarse scale, such as rock bedding and forest boundaries, and captures the fine textures of ground features at a fine scale, such as tree canopy morphology and mineral veins. Through directional modulation, the operator can distinguish ground feature boundaries with different orientations, such as the orientation of mine pit edges and the extension direction of water body shorelines. The Laplace term is sensitive to the geometric edges of ground features, and the kernel function integrates the spectral absorption index reflecting the material properties of the ground feature and the radar polarization decomposition parameters reflecting the structure of the ground feature. Therefore, the collaborative feature value output by the operator simultaneously contains the geometric morphology information, spectral absorption characteristics, and polarization scattering mechanism of the ground feature, ultimately generating a collaborative feature map for fine-grained identification of ground feature types.

[0034] The fine-grained land cover type identification subsystem is configured to input the collaborative feature map into a pre-built land cover feature reference library. The land cover feature reference library consists of electromagnetic wave response data of typical land cover obtained from field measurements and laboratory measurements, covering feature vectors of different tree species, rocks and minerals, and soil textures. Through feature matching, each pixel in the collaborative feature map is compared with the feature vector in the land cover feature reference library, the similarity is calculated, and the land cover type is determined according to the maximum likelihood principle. At the same time, spatial context information is used to correct isolated pixels, and a fine-grained land cover type map is output, in which each type of land cover corresponds to a clear material and structural attribute.

[0035] The natural resource dynamic monitoring subsystem is configured to overlay detailed land cover type maps acquired at different time phases, extract areas where types have changed, and calculate the area and boundaries of the changed areas; combine topographic slope data and meteorological observation data to perform stability analysis on the changed areas, identify areas of concentrated change through spatial clustering, and classify the areas of concentrated change into early warning levels according to the rate of change and the degree of potential impact, generating a natural resource dynamic monitoring early warning map to show the changing trends and risk levels of various land cover types.

[0036] In this context, "same area" refers to geographically overlapping units, such as a complete watershed, mining area, or forest area. This ensures accurate registration of multi-source remote sensing data within the region, allowing extracted features to collectively reflect the true spatial distribution and attributes of the region's natural resources, such as forests, minerals, and water bodies. High spatial resolution panchromatic band data provides extremely high spatial detail of ground features in a single band, such as the boundaries of tree canopies and the orientation of rock joints. This is used to accurately delineate the outlines of ground features in natural resource monitoring and to assist in identifying resource utilization boundaries or traces of destruction. Multispectral reflectance data records the spectral reflectance of ground features in multiple narrow bands, such as the strong reflectance of vegetation in the near-infrared band and the absorption characteristics of soil in the visible light band. This is used to distinguish different vegetation types, soil categories, and mineral compositions, forming the basis for natural resource classification. Synthetic Aperture Radar (SAR) backscattering data acquires the scattering intensity of ground features to microwave pulses. It is sensitive to the dielectric constant and surface roughness of ground features; for example, volume scattering of forests reflects tree density, and backscattering of bare soil reflects soil moisture. It can penetrate clouds and acquire subsurface information, making it suitable for dynamic monitoring of natural resources in cloudy and rainy areas. Lidar point cloud data acquires the three-dimensional coordinates of ground features through laser ranging, generating dense point clouds that accurately describe topographic relief, vegetation height, and vertical structure. For example, it can be used to estimate timber volume and monitor the depth of mining damage, providing direct evidence for the three-dimensional quantification of natural resources. Geometric edges of ground features refer to their linear or arc-shaped boundaries in space, such as forest boundaries, mine edges, and water shorelines. Extracted from remote sensing data through multi-scale decomposition, they are used to accurately delineate the distribution range and boundaries of changing areas of natural resources. Spectral absorption indices are quantitative indicators calculated based on specific absorption depths and locations in the spectral curves of ground features. For example, the absorption characteristics of iron ions in rocks at 0.9 micrometers and the absorption valley of chlorophyll in vegetation at 0.68 micrometers are used to identify mineral types and vegetation health. Radar polarization decomposition parameters decompose radar echoes into components based on different scattering mechanisms, such as surface scattering, secondary scattering, and volume scattering. For example, forest areas have high volume scattering components, while building areas exhibit significant secondary scattering. This is used to distinguish land cover structure types and assist in natural resource classification. Land cover material, structure, and polarization characteristics are three types of information integrated into the collaborative feature map: material refers to the physical composition of land cover, such as granite and broadleaf forest timber; structure refers to the spatial morphology of land cover, such as tree height stratification and rock bedding; polarization characteristics refer to the scattering pattern of land cover interacting with electromagnetic waves. Together, these three constitute a comprehensive description of the essence of natural resources. Different tree species, rock minerals, and soil textures are specific categories in the land cover feature reference database, such as Masson pine, Chinese fir, granite, limestone, sandy loam, and clay loam. These types directly correspond to the basic objects in natural resource surveys. Material and structural attributes are the attribute information attached to each category in a detailed land cover type map. For example, if a certain area is identified as a mature deciduous broad-leaved forest, its material attribute is lignocellulose, and its structural attributes are an average tree height of 20 meters and a canopy closure of 0.8, providing a quantitative basis for dynamic analysis.Changes in land cover type refer to the transformation of land cover types in the same area from one to another at different times. For example, deforestation turns forests into bare land, and lake shrinkage turns lakes into mudflats. These changes directly reflect the dynamics of the quantity and quality of natural resources and are the core content of monitoring. Topographic slope data and meteorological observation data describe the steepness of the land surface and changes in meteorological elements, respectively: slope data, derived from digital elevation models, affects the risk of soil erosion and vegetation growth conditions; meteorological data, including precipitation and temperature, is used to analyze the stability of the changed area. For example, newly exposed bare land after heavy rainfall may trigger landslides. The rate of change and the degree of potential impact refer to the speed of land cover type changes, such as the annual reduction in forest area and its potential damage to ecosystem service functions, such as declining water conservation and loss of biodiversity. Based on this, early warning levels are classified, and dynamic monitoring and early warning maps are generated to provide a basis for decision-making in natural resource protection.

[0037] The principles described in the above embodiments are referenced in the appendix. Figure 2 This embodiment integrates multi-source remote sensing data to achieve systematic monitoring of the spatial distribution and dynamic changes of natural resources. The heterogeneous feature collaborative extraction subsystem integrates multi-scale spatial-spectral joint transformation and multi-source data registration to effectively extract complementary features of land features in geometric, spectral, and polarimetric dimensions, overcoming the limitations of single data sources and generating a collaborative feature map with high information density, laying the foundation for refined identification. The land feature type refined identification subsystem performs feature matching and spatial context optimization based on a measured electromagnetic wave response database, significantly improving the accuracy of land feature classification and the ability to interpret physical attributes, outputting a refined land feature type map that combines material and structural attributes. The natural resource dynamic monitoring subsystem, through temporal overlay and multi-source data fusion analysis, achieves quantitative extraction, stability assessment, and risk classification of natural resource changes, forming spatially clear dynamic monitoring and early warning results. The collaborative work of these subsystems ultimately constructs an integrated technology chain from feature fusion and type identification to dynamic assessment, improving the accuracy, efficiency, and decision support capabilities of natural resource monitoring.

[0038] Example 2: Figure 3 As shown, based on Example 1, the heterogeneous feature collaborative extraction subsystem provided in this embodiment of the invention includes:

[0039] The displacement offset correction component is configured to spatially register the feature data layers of the extracted geometric edges, the feature data layer of the spectral absorption index, and the feature data layer of the radar polarization decomposition parameters, using the digital surface model generated from the lidar point cloud data as the elevation control benchmark, to correct the position offset due to differences in the data source.

[0040] The data fusion component is configured to fuse the three types of feature data layers after registration. The feature data layer of geometric edges serves as the spatial skeleton to determine the boundaries and contours of ground features; the feature data layer of spectral absorption index serves as the material attribute filler to give the material composition information of ground features within each contour; and the feature data layer of radar polarization decomposition parameters serves as the structural attribute supplement to describe the vertical stratification and scattering mechanism of ground features within the contour.

[0041] The output component is configured to generate a collaborative feature map in which each pixel simultaneously contains geometric edge information of ground features, material properties characterized by spectral absorption index, and structural and polarization characteristics characterized by radar polarization decomposition parameters.

[0042] In the above embodiments, this embodiment achieves the collaborative characterization of multi-dimensional attributes of natural resource features through spatial registration, fusion, and structured expression of multi-source feature data. The displacement correction component uses the lidar digital surface model as the elevation benchmark to spatially register three feature layers: geometric edges, spectral absorption index, and radar polarization decomposition parameters. This eliminates spatial positional deviations caused by factors such as sensor perspective and terrain undulations, ensuring strict correspondence between different features at the same geographical location and providing a spatial consistency basis for feature fusion. The data fusion component uses the geometric edge feature layer as the spatial skeleton to accurately delineate the boundaries and contours of features; it uses the spectral absorption index feature layer to fill in material attributes, endowing the internal regions of the contours with material composition information; and it uses the radar polarization decomposition parameter feature layer to supplement structural attributes, describing the vertical stratification and scattering mechanisms within the features. The three feature layers are superimposed under the constraints of the spatial skeleton to form a structured feature expression system. The result output component generates a collaborative feature map, where each pixel simultaneously contains geometric morphological information, spectral absorption characteristics, and polarization scattering characteristics, achieving an integrated description of the spatial distribution, material composition, and structural features of features. The synergistic effect of the technical features of each component in this embodiment transforms the heterogeneous features extracted from multi-source remote sensing data into fused features with clear physical meaning and spatial alignment, providing a feature foundation for the fine identification of land cover types that combines geometric accuracy, material differentiation, and structural description capabilities.

[0043] Example 3: Based on Example 2, the data fusion component provided in this embodiment of the invention includes:

[0044] The spatial skeleton network construction sub-component is configured to take the feature data layer of the geometric edges after displacement correction as input. The feature data layer contains discrete edge segments of ground features at different scales. The discrete edge segments have breaks and discontinuities due to sensor differences or ground feature shadows. The discrete edge segments are connected by tracing the endpoints of adjacent segments along the edge direction, connecting broken segments belonging to the same ground feature boundary into continuous curves. The connected continuous curves are then closed by identifying and closing the endpoints of unclosed curves to form closed polygonal contours. All closed polygonal contours together constitute the spatial skeleton network.

[0045] The material attribute filling sub-component is configured to overlay the spatial skeleton network with the feature data layer of the spectral absorption index after displacement correction, using each closed polygon divided by the spatial skeleton network as the basic unit; within the contour of each closed polygon, the values ​​of all pixels in the feature data layer of the spectral absorption index are extracted, the distribution characteristics of the values ​​are statistically analyzed, and the value with the highest frequency is selected as the representative value of the polygon's material attribute; the representative value of the material attribute is assigned to each pixel within the polygon, so that the entire polygon area has a consistent material attribute identifier; each closed polygon of the spatial skeleton network is filled with the ground feature material information represented by the spectral absorption index;

[0046] The structural attribute point-by-point embedding sub-component is configured to perform pixel-by-pixel alignment between the image composed of multiple pixels with already filled material attributes and the feature data layer of the radar polarimetric decomposition parameters after displacement offset correction. The feature data layer of the radar polarimetric decomposition parameters contains three independent data layers: surface scattering component, secondary scattering component, and volume scattering component. For each pixel in the image, the surface scattering component value, secondary scattering component value, and volume scattering component value corresponding to the pixel position are extracted respectively. The three component values ​​are then appended as a set of structural attribute parameters to the original geometric edge information and material attribute information of the pixel. After point-by-point embedding, each pixel finally contains three types of information: geometric edge information of the polygon boundary identifier where the pixel is located, representative value of polygon material attribute, and three radar polarimetric decomposition parameters of the pixel position, generating a complete collaborative feature map.

[0047] In the above embodiments, the data fusion component of this embodiment achieves multi-level feature integration from geometric structure to physical attributes through the collaborative processing of multi-source remote sensing data. The spatial skeleton network construction sub-component, based on the geometric edge feature data layer after displacement correction, connects broken line segments and closes them into polygonal contours to form a continuous topological framework describing the spatial distribution pattern of ground features; it can effectively overcome the edge discontinuity problem caused by sensor differences or shadows, and establish a complete foundation for the expression of ground feature boundaries. The material attribute filling sub-component superimposes the spatial skeleton network with the spectral absorption index feature data layer, statistically analyzes the distribution characteristics of internal pixel values ​​using polygons as units, and assigns the spectral absorption index value with the highest frequency as the representative value of the material attribute to the entire area; while maintaining the spatial continuity of the material, it reduces local noise interference, so that areas of the same material receive consistent material identification. The structural attribute point-by-point embedding sub-component aligns the image with the radar polarization decomposition parameter feature data layer after material filling, and adds three component values ​​of surface scattering, secondary scattering, and volume scattering to each pixel; through the pixel-by-pixel embedding operation, each pixel simultaneously contains geometric boundary identification, representative value of material attribute, and radar polarization parameters, forming multi-dimensional feature coupling.

[0048] In summary, this embodiment establishes the correspondence between geometric edges, spectral materials, and radar scattering characteristics within a unified spatial framework; improves the accuracy of regional material consistency representation through material filling under polygonal contour constraints; preserves spatial detail differences in radar scattering features through a point-by-point embedding mechanism; generates a collaborative feature map that simultaneously contains boundary geometric information, continuous material properties, and point-level structural parameters; and provides a physically interpretable multi-source fusion data foundation for land cover classification, change detection, or 3D reconstruction.

[0049] Example 4: Based on Example 3, the structural attribute point-by-point embedding sub-components provided in this embodiment of the invention include:

[0050] The pixel position correspondence processing module is configured to use the image with completed material attribute filling as the reference image, in which each pixel contains geometric edge information of polygon boundary markers and material attribute representative values; use the feature data layer of radar polarization decomposition parameters after displacement offset correction as the data layer to be registered; compare the spatial coordinates of the reference image and the data layer to be registered pixel by pixel; if coordinate offset or pixel size inconsistency is found due to data format differences, the pixel grid of the reference image is used as a template to spatially resample the three component data layers of the data layer to be registered. After resampling, the position of each pixel in the data layer to be registered coincides with the corresponding pixel position in the reference image, generating three component data layers with unified spatial reference.

[0051] The local neighborhood extraction module is configured to overlay the three component data layers of the spatial reference with the reference image. For each pixel in the reference image, it reads the surface scattering component value, secondary scattering component value, and volume scattering component value from the same position in the three component data layers. After reading, a local neighborhood window containing eight adjacent pixels is constructed with the pixel as the center. The values ​​of all pixels in the window for the three components are extracted, and the local mean and standard deviation of each component are calculated. The component values ​​of the center pixel are compared with the local mean of the corresponding component. If a component value deviates from the local mean by more than three times the standard deviation, the component value is determined to be an outlier and replaced with the local mean. After the filtering process, each pixel obtains a set of surface scattering component values, secondary scattering component values, and volume scattering component values ​​that have been verified for local consistency.

[0052] The image metadata unit reconstruction module is configured to append the three component values, after outlier filtering, as a set of structural attribute parameters to the original data of the corresponding pixel in the reference image. The original data includes geometric edge information identified by the polygon boundary where the pixel is located, as well as the representative value of the polygon material attribute. The merging operation combines the geometric edge information, the representative value of the material attribute, the surface scattering component value, the secondary scattering component value, and the volume scattering component value in a fixed order to form a new image metadata unit. The data units of all pixels together constitute a collaborative feature map. Each pixel in the collaborative feature map contains three types of information: geometric edge information identified by the polygon boundary, material attribute represented by the spectral absorption index, and structural and polarization characteristics characterized by three radar polarization decomposition parameters.

[0053] In the above embodiments, this embodiment solves the geometric registration problem between multi-source data through spatial resampling, ensuring the location accuracy of attribute embedding; reduces the impact of outliers in radar polarization parameters through local statistical verification, enhancing data spatial consistency; constructs a unified data structure to achieve the organic integration of geometric, material, and polarization scattering features at the pixel level; generates multi-dimensional feature expressions that simultaneously possess boundary continuity, material homogeneity, and scattering detail differences; and provides highly reliable input data for feature fusion-based fine classification of ground features, structural inversion, or target recognition.

[0054] Example 5: Based on Example 4, the image metadata unit reconstruction module provided in this embodiment of the invention includes:

[0055] The spatial index organization submodule is configured to build a two-dimensional spatial index table based on the original row and column numbers of each pixel in the reference image. The two-dimensional spatial index table records the storage location of the data unit corresponding to each row and column. All data units are arranged in ascending order of row number and column number to form an ordered sequence of data units, which corresponds completely to the spatial grid of the reference image.

[0056] The neighborhood consistency verification submodule is configured to access each data cell row by row and column by column based on an ordered sequence of data cells. For the currently accessed data cell, it extracts the polygon boundary identifier recorded in its geometric edge information. It then queries the data cells of the cell to its east, south, and southeast corner, and extracts the polygon boundary identifiers of these three adjacent cells. The boundary identifier of the current cell is compared with the boundary identifiers of the three adjacent cells. If the boundary identifiers are inconsistent and the current cell is located on the polygon boundary, the current cell is marked as a boundary cell. A secondary verification is performed on all boundary cells. During the verification, the neighborhood range is expanded to twenty-four surrounding cells with the current cell as the center. The boundary identifier that appears most frequently in the neighborhood is counted and replaced with the boundary identifier in the geometric edge information of the current cell. After the neighborhood consistency verification, the geometric edge information in all data cells is spatially continuous and closed.

[0057] The data unit writing submodule is configured to write a complete sequence of data units that has passed neighborhood consistency verification into a raster data file. During writing, the pixel grid of the reference image is used as a template. Each row of the data file corresponds to a row of pixels in the reference image, and each column corresponds to a column of pixels in the reference image. The data unit written at each pixel position contains five data items: geometric edge information, material property representative value, surface scattering component value, secondary scattering component value, and volume scattering component value. After writing is completed, a collaborative feature map file is generated. The data units of all pixels in the collaborative feature map file together constitute a complete collaborative feature map. Each pixel in the map contains three types of information: geometric edge information identified by polygon boundaries, material properties represented by spectral absorption index, and structural and polarization characteristics characterized by three radar polarization decomposition parameters.

[0058] In the above embodiments, this embodiment maintains the original spatial relationship of multi-source feature data through two-dimensional spatial indexing; optimizes the continuity of geometric boundaries and reduces edge noise through neighborhood consistency verification; realizes standardized encapsulation of geometric, material and polarization features in the raster data structure; and generates a collaborative feature map that is spatially continuous, has clear boundaries and complete features.

[0059] Example 6: As Figure 4 As shown, based on Example 1, the detailed feature type identification subsystem provided in this embodiment of the invention includes:

[0060] The multidimensional comparison component is configured to read data units of each pixel from the collaborative feature map, convert geometric edge information into edge intensity coefficients, and combine them with material attribute representative values ​​and three radar polarization decomposition parameters to form a multidimensional feature vector for each pixel. Simultaneously, it extracts feature vectors of all known land cover types from a pre-built land cover feature reference library. These feature vectors consist of electromagnetic wave response data from field and laboratory measurements, covering different tree species, rock minerals, and soil textures. The multidimensional feature vector of each pixel is compared dimension-by-dimensionally with the feature vectors of each land cover type in the reference library. The absolute value of the difference in each dimension is calculated, and the absolute values ​​of the differences in all dimensions are summed to obtain the initial difference value between the pixel and each land cover type in the feature reference library. This initial difference value is then converted into an initial similarity score.

[0061] The preliminary type determination component is configured to retrieve the statistical distribution parameters of the feature vectors of each land cover category in the feature reference library, obtained from field measurements, including the mean and variance of each dimension; and establish a multidimensional probability density distribution of the land cover based on the statistical distribution parameters of the feature vectors. The multidimensional feature vector of each pixel is then substituted into the probability density distribution of each land cover category to calculate the likelihood value of the pixel belonging to each category; the likelihood values ​​of all categories are compared, and the land cover category with the highest likelihood value is selected as the preliminary type determination of the pixel, and the maximum likelihood value is recorded as the determination confidence level of the pixel.

[0062] The isolated cell correction component is configured to overlay the geometric edge information from the preliminary type determination map and the collaborative feature map, using the closed polygon outlines defined by the geometric edge information as spatial units. Within each closed polygon, the preliminary type determination and corresponding determination confidence of all cells are statistically analyzed, and the frequency and average confidence of each type are calculated. The type with the highest frequency and an average confidence exceeding a preset threshold is selected as the dominant type of the polygon. Isolated cells within the polygon that are inconsistent with the dominant type are identified, and their types are corrected to the dominant type of the polygon. For cells located on the polygon boundary, the dominant type and average confidence of their adjacent polygons are extracted. If the determination confidence of the boundary cell is lower than the average confidence of any dominant type in the adjacent polygons, its type is corrected to the dominant type of the adjacent polygon with a higher average confidence. After correction, a refined land cover type map is output. In the refined land cover type map, the type of each cell is spatially continuous with the surrounding land cover types, and each type of land cover corresponds to clear material and structural attributes.

[0063] In the above embodiments, this embodiment realizes a complete classification process from pixel-level multidimensional feature extraction, probabilistic type determination to spatial context optimization, and finally outputs a fine land cover type map with clear material and structural attributes, spatial continuity and conformity to the distribution law of land cover, which improves the accuracy, robustness and practicality of land cover type identification.

[0064] Example 7: Based on Example 6, the type preliminary determination component provided in this embodiment of the invention includes:

[0065] The construction and storage format conversion subcomponent is configured to retrieve the feature vector statistical distribution parameters of each type of land cover from the land cover feature reference library. The statistical distribution parameters of each type of land cover include the mean and variance of each dimension under that category. The mean values ​​of all dimensions of each type of land cover are arranged in a fixed order to form the mean vector of the land cover. The variance values ​​of all dimensions of each type of land cover are arranged in the same order to form the variance vector of the land cover. The mean vector and variance vector of each type of land cover are combined and encapsulated into a probability density parameter set, and the probability density parameter set corresponds one-to-one with the land cover category identifier. The probability density parameter sets of all categories are arranged in the category order in the land cover feature reference library to form a complete probability density parameter library.

[0066] The calculation subcomponent is configured to read the multidimensional feature vector of each pixel, with the values ​​of each dimension arranged in a fixed order; retrieve the probability density parameter set of each land cover category from the probability density parameter library, including the mean vector and variance vector of the land cover; compare the value of each dimension of the pixel's multidimensional feature vector with the value in the mean vector of the corresponding dimension, and calculate the degree of deviation of the dimension value relative to the mean; at the same time, call the value in the variance vector of the corresponding dimension to scale the degree of deviation, and combine the scaled values ​​of the deviation of all dimensions to obtain the initial attribution metric value of the pixel relative to the current land cover category; repeat the above calculation for all land cover categories, and each pixel obtains a set of initial attribution metric values, the number of which is equal to the total number of categories in the land cover feature reference library;

[0067] The likelihood value generation subcomponent is configured to construct a sequence of initial attribution measures for each pixel corresponding to all land cover categories; traverse the sequence of measures to find the maximum and minimum values; based on the maximum and minimum values, perform a linear transformation on each initial attribution measure in the sequence of measures, mapping the transformed numerical range to a continuous interval between zero and one; after the transformation, each pixel corresponds to a value between zero and one for each land cover category, which is the likelihood value of the pixel belonging to that land cover category; organize all likelihood values ​​of all pixels according to pixel location and land cover category to generate a three-dimensional likelihood value data block. The first and second dimensions of the three-dimensional likelihood value data block correspond to the spatial location of the pixel, and the third dimension corresponds to each category in the land cover feature reference library. Each spatial location stores a set of likelihood values ​​of the location pixel belonging to each type of land cover.

[0068] In the above embodiments, this embodiment realizes the complete process from feature parameter extraction, measurement calculation to probability mapping, which can efficiently complete the preliminary probability classification of multiple types of land cover based on statistical distribution parameters, and provide structured probability output for refined classification or decision fusion.

[0069] Example 8: As Figure 5 As shown, based on Embodiment 1, the natural resource dynamic monitoring subsystem provided in this embodiment of the invention specifically includes:

[0070] The stability index calculation component is configured to overlay the extracted change area layer with a terrain slope data layer. The terrain slope data is exported from a digital elevation model, with each raster cell recording the slope value at that location. For each change area, the slope values ​​of all raster cells within its boundary are extracted, the cumulative distribution of slope values ​​is calculated, and the slope value corresponding to 90% of the cumulative distribution is taken as the representative slope of the change area. Simultaneously, the variation range of slope values ​​within the area is calculated, i.e., the difference between the maximum and minimum values. The representative slope and variation range are combined to generate the terrain influence coefficient for that area. The change area layer is then overlaid with a meteorological observation data layer, which includes precipitation and temperature grids for multiple time phases. For each change area, the cumulative precipitation sequence at different time phases is extracted, and the mean and variance of the cumulative precipitation sequence are calculated. The ratio of the mean to the variance is taken as the hydrological disturbance intensity of the change area. The terrain influence coefficient is multiplied by the hydrological disturbance intensity to obtain the stability index for each change area. The lower the stability index value, the more unstable the area.

[0071] The spatial clustering identification component is configured to take a layer of change regions containing stability indices as input, and use the geometric center point of each change region as a representative point of the change region. The coordinates of the representative point are calculated from the region boundary. A spatial adjacency graph is constructed for all representative points, connecting representative points whose mutual distance is less than a preset distance threshold, which is twice the average width of all change regions. In the spatial adjacency graph, all connected subgraphs are extracted, and each connected subgraph contains a set of mutually adjacent representative points. The change regions corresponding to all representative points in the same connected subgraph are merged into a spatial unit. When merging, the outer envelope of the region boundary is taken as the boundary of the new unit, and the median of the region stability index is taken as the stability index of the new unit. All merged spatial units are change concentration areas, and each concentration area corresponds to a set of spatially adjacent change regions with similar stability.

[0072] The warning level classification and mapping component is configured to calculate the rate of change and potential impact for each area of ​​concentrated change. The rate of change is obtained by dividing the total area change of all changed areas within the concentrated change area at different time phases by the time interval, which is determined by the difference between the acquisition dates of the two time phases. The potential impact is determined based on the ecological service function weight values ​​corresponding to different land cover types within the concentrated change area. These ecological service function weight values ​​are obtained in advance through field surveys; for example, water conservation areas and biodiversity hotspots have high weights. The component extracts the type of each pixel in the detailed land cover type map within the concentrated change area, calculates the area proportion of each type, multiplies the area proportion of each type by its ecological service function weight value, and then sums the results to obtain the final impact value. The potential impact value of the concentrated change area is calculated; the change rate value and the potential impact value are combined, and a two-dimensional plane is constructed with the change rate as the horizontal axis and the potential impact value as the vertical axis. The combined value of each concentrated area is mapped to a point on the two-dimensional plane; the warning level is divided according to the preset threshold line. The threshold line divides the plane into three regions: high, medium and low. The concentrated area falling into the high region is marked as the red warning level, the medium region is marked as the yellow warning level, and the low region is marked as the blue warning level; all concentrated change areas and their warning levels are drawn on the geographic base map. The boundaries of the concentrated change areas are filled with the color of the corresponding warning level, and the change rate value and the potential impact value are marked to generate a natural resource dynamic monitoring and early warning map.

[0073] In the above embodiments, the natural resource dynamic monitoring subsystem of this embodiment achieves a systematic process for stability assessment, spatial cluster analysis, and hierarchical early warning mapping of natural resource change areas through the collaborative work of the stability index calculation component, spatial cluster identification component, and early warning level classification and mapping component. The stability index calculation component overlays change areas with topographic slope and meteorological observation data to extract representative slope, slope change amplitude, and hydrological disturbance intensity, comprehensively generating an index reflecting the stability of regional natural conditions. The index integrates topographic and climatic factors, quantifying the degree of influence of the natural environment on the change area, providing a basic assessment indicator for analysis. Based on the stability index, the spatial cluster identification component identifies spatially adjacent change areas with similar stability by constructing a spatial adjacency graph, merging them into concentrated change areas; it aggregates discrete change areas into units with spatial continuity and stability similarity, reducing redundant analysis units, highlighting spatial distribution patterns, and providing suitable geographical units for regional-scale assessment. The warning level classification and mapping component, at the scale of concentrated change areas, combines two indicators—the rate of change and the degree of potential impact—and determines the warning level of each concentrated area through two-dimensional planar mapping and threshold classification. The rate of change reflects the intensity of dynamic change, while the degree of potential impact reflects the importance of ecosystem service functions. The combination of the two can comprehensively assess the urgency and severity of impact of concentrated change areas. Finally, the spatial location, warning level, and relevant indicator values ​​of each concentrated area are intuitively displayed through geographic mapping, forming a warning map that can be directly used for decision support.

[0074] In summary, this embodiment realizes a complete analysis chain from quantifying the stability of changing areas and spatial clustering to hierarchical early warning mapping. It can systematically identify unstable and high-impact concentrated areas in natural resource changes and provide spatially clear and hierarchically differentiated early warning information for natural resource management and protection.

[0075] Example 9: Based on Example 8, the early warning level classification and mapping component provided in this embodiment of the invention specifically includes:

[0076] A subcomponent for scale division is created and configured to obtain the sets of change rate values ​​and potential impact value sets for all concentrated change areas. The change rate value set is traversed to find its maximum and minimum values. Using the minimum value as the starting point and the maximum value as the ending point, the range of change rate values ​​is divided into ten consecutive intervals, each corresponding to one unit scale mark on the horizontal axis. Similarly, the potential impact value set is traversed to find its maximum and minimum values. Using the minimum value as the starting point and the maximum value as the ending point, the range of potential impact value sets is divided into ten consecutive intervals, each corresponding to one unit scale mark on the vertical axis. A two-dimensional coordinate system is plotted, with the change rate represented by the horizontal axis and the potential impact level by the vertical axis. The origin of the coordinate system corresponds to the intersection of the minimum change rate and the minimum potential impact level. The positive direction of the horizontal axis points in the direction of increasing change rate, and the positive direction of the vertical axis points in the direction of increasing potential impact level.

[0077] The normalization and coordinate calculation subcomponent is configured to read the rate of change and potential impact values ​​of each region of change from the data records of each region of change; substitute the rate of change value into the horizontal axis scale division rules to determine the horizontal axis interval number to which the rate of change value belongs, and take the median of the horizontal axis interval number as the projection coordinate of the region of change on the horizontal axis; substitute the potential impact value into the vertical axis scale division rules to determine the vertical axis interval number to which the potential impact value belongs, and take the median of the vertical axis interval number as the projection coordinate of the region of change on the vertical axis; combine the horizontal axis projection coordinates and the vertical axis projection coordinates to generate the coordinate points of the region of change on the two-dimensional plane, where the horizontal coordinate value of the coordinate point is the median of the interval corresponding to the rate of change, and the vertical coordinate value is the median of the interval corresponding to the potential impact value;

[0078] The geometric positioning and identification sub-component is configured to use the generated two-dimensional plane coordinate system as a base map, traverse all areas of concentrated change, and plot the coordinates of each area on the base map. During plotting, a circular symbol with a diameter of two units is drawn centered on the coordinate point, and the inside of the circular symbol is filled with a color corresponding to the warning level of that area. A text label is attached to the right of each circular symbol, containing the unique identifier of the area of ​​concentrated change, the rate of change value, and the potential impact value, with the rate of change and potential impact values ​​rounded to two decimal places. After the coordinates of all areas of concentrated change are plotted, a two-dimensional scatter plot is generated containing the combined value mapping points of all areas of concentrated change. Each point in the scatter plot represents an area of ​​concentrated change, and the spatial position of the point reflects the comprehensive characteristics of the area of ​​concentrated change in both the rate of change and the potential impact.

[0079] In the above embodiments, this embodiment realizes the simultaneous visualization of two types of indicators: the rate of change and the degree of potential impact of multiple concentrated change areas. This allows users to intuitively identify the distribution patterns, clustering trends, and differences in early warning levels of each concentrated area in terms of dual-dimensional characteristics through scatter plots, providing spatial graphical basis for comprehensive analysis.

[0080] Example 10: Based on Example 9, the normalization processing and coordinate calculation sub-component provided in this embodiment of the invention specifically includes:

[0081] The extreme value extraction and range determination module is configured to treat the change rate values ​​of all concentrated change areas as a numerical sequence. Each value in the numerical sequence corresponds to the land cover type transformation area of ​​a concentrated area per unit time. The module iterates through the numerical sequence, comparing each value with the maximum and minimum values ​​of the current record. Initially, the first value in the sequence is set as both the maximum and minimum values. The module reads values ​​one by one. If the read value is greater than the maximum value of the current record, the maximum value is replaced with the new value. If the read value is less than the minimum value of the current record, the minimum value is replaced with the new value. After traversing all values, the module obtains the maximum and minimum values ​​in the set of change rate values. The maximum value reflects the fastest rate of change in the region's natural resources, and the minimum value reflects the slowest rate. Together, they determine the overall fluctuation range of the change rate.

[0082] The extreme value extraction and range determination module is configured to treat the potential impact values ​​of all concentrated change areas as another numerical sequence. Each value in the numerical sequence corresponds to the degree of damage to ecosystem service functions caused by changes in land cover types in a concentrated area. The module iterates through the numerical sequence, using the same extreme value extraction method as the change rate values, and compares each value with the maximum and minimum values ​​of the current record. Initially, the first value in the sequence is set as both the maximum and minimum values. The module reads values ​​one by one. If the read value is greater than the maximum value of the current record, the maximum value is replaced with the read value. If the read value is less than the minimum value of the current record, the minimum value is replaced with the read value. After traversing all values, the module obtains the maximum and minimum values ​​in the potential impact value set. The maximum value reflects the maximum ecological damage that changes in regional natural resources may cause, and the minimum value reflects the minimum damage. Together, they determine the overall fluctuation range of the potential impact.

[0083] The boundary value generation module is configured to use the minimum value of the set of change rate values ​​as the starting boundary and the maximum value as the ending boundary. Between the starting and ending boundaries, ten consecutive numerical intervals of equal length are generated sequentially, with the starting boundary as the left endpoint of the first interval and the ending boundary as the right endpoint of the tenth interval. The left endpoint of each numerical interval is obtained by adding the interval number minus one and multiplying by the interval length to the starting boundary, and the right endpoint of each interval is obtained by adding the interval length to the left endpoint. The same method is used to process the set of potential influence values, with the minimum value as the starting boundary and the maximum value as the ending boundary, generating ten consecutive numerical intervals of equal length. The scale division of the horizontal axis change rate and the vertical axis potential influence is completed, with each interval corresponding to a unit scale, providing a quantitative benchmark for mapping the combined values ​​of the concentrated area to a two-dimensional plane.

[0084] In the above embodiments, this embodiment realizes the simultaneous visualization of two types of indicators: the rate of change and the degree of potential impact of multiple concentrated change areas. This allows users to intuitively identify the distribution patterns, clustering trends, and differences in early warning levels of each concentrated area in terms of dual-dimensional characteristics through scatter plots, providing spatial graphical basis for comprehensive analysis.

[0085] Example 11: Based on Example 10, the extreme value extraction and range determination module provided in this embodiment of the invention specifically includes:

[0086] The extraction and double assignment submodule is configured to arrange the sequence of change rate values ​​according to the concentration zone number order, read the first value at the beginning of the concentration zone number sequence, and write the first value into both the maximum value temporary storage area and the minimum value temporary storage area. The values ​​stored in the maximum value temporary storage area and the minimum value temporary storage area are equal, both being the change rate values ​​of the first concentration zone in the concentration zone number sequence.

[0087] The storage unit update submodule is configured to sequentially read the second value in the rate of change numerical sequence, and match the second value with the values ​​in the maximum value temporary storage area and the minimum value temporary storage area respectively. If the loaded second value is greater than the value in the maximum value temporary storage area, the loaded second value replaces the original value in the maximum value temporary storage area, so that the maximum value temporary storage area stores the new larger value. If the loaded value is less than the value in the minimum value temporary storage area, the loaded value replaces the original value in the minimum value temporary storage area, so that the minimum value temporary storage area stores the new smaller value. After the matching and updating of the second value is completed, the third value is read, and the loading, matching and updating operations are repeated until all values ​​in the rate of change numerical sequence have been processed.

[0088] The extreme value extraction and output submodule is configured to, after traversing all values ​​in the rate of change numerical sequence, store the maximum value in the maximum value temporary storage area as the maximum value in the rate of change numerical sequence, and store the minimum value in the minimum value temporary storage area as the minimum value in the rate of change numerical sequence; extract the value in the maximum value temporary storage area as the maximum rate of change, and extract the value in the minimum value temporary storage area as the minimum rate of change; the two values ​​together constitute the extreme value pair of the rate of change numerical set, and output to the extreme value extraction and range module for boundary value generation.

[0089] In the above embodiments, this embodiment achieves automated identification and extraction of the maximum and minimum values ​​in a numerical sequence of change rates. The module first initializes the extreme value temporary storage area through a dual-assignment submodule, simultaneously setting the first and second values ​​of the sequence as the initial benchmarks for both the maximum and minimum values. Subsequently, the storage unit update submodule dynamically compares and updates the candidate extreme values ​​in the temporary storage area by traversing the remaining values ​​in the sequence, ensuring that the temporary storage area always retains the actual maximum and minimum values ​​from the currently processed data after each comparison. Finally, after traversing the sequence, the extreme value extraction and output submodule extracts and outputs the determined maximum and minimum values ​​from the temporary storage area, forming extreme value pairs. This embodiment, through a step-by-step iterative comparison method, accurately locks the extreme value range of the numerical set, providing an accurate data foundation for boundary value generation, while avoiding the computational overhead of full sequence sorting or repeated scanning, thus improving the efficiency and reliability of extreme value extraction.

[0090] Example 12: Based on Example 11, the extreme value extraction and output submodule provided in this embodiment of the invention specifically includes:

[0091] The read and identification confirmation unit is configured to read the currently stored value from the maximum value temporary storage area and mark the value as a candidate value for the maximum rate of change; read the currently stored value from the minimum value temporary storage area and mark the value as a candidate value for the minimum rate of change; after reading, the values ​​in the two temporary storage areas are compared and verified with the first and last values ​​of the original rate of change value sequence to confirm that the candidate value for the maximum value is not less than any value in the sequence and the candidate value for the minimum value is not greater than any value in the sequence. After verification, the two candidate values ​​are officially confirmed as the maximum rate of change and the minimum rate of change.

[0092] The extreme value pair attribute appending unit is configured to write the confirmed maximum and minimum rate of change into the same data recording unit. In the data recording unit, an attribute label is appended to the maximum rate of change, with the label content being the upper limit of the rate of change and the corresponding concentration zone number. An attribute label is appended to the minimum rate of change, with the label content being the lower limit of the rate of change and the corresponding concentration zone number. At the same time, the difference between the maximum and minimum rate of change is calculated, and the difference is appended to the data recording unit as the rate of change range amplitude. The three together constitute a complete set of extreme value pair description data.

[0093] The encapsulation and output orientation unit is configured to encapsulate data record units containing the maximum value of the rate of change, the minimum value of the rate of change, the range of the rate of change, and their respective attribute labels, forming an extreme value pair data packet; an identifier is added to the header of the extreme value pair data packet, the identifier content being the extreme value pair of the rate of change; the encapsulated extreme value pair data packet is transmitted to the output interface of the extreme value extraction and range determination module, and the output interface is connected to the input interface of the boundary value generation module.

[0094] In the above embodiments, this embodiment transforms the raw values ​​generated during the extreme value extraction process into a structured data package containing values, attributes, and derived indicators, ensuring the integrity, interpretability, and traceability of the extreme value data; through standardized encapsulation and targeted output, it achieves seamless integration with the downstream boundary value generation module, providing input data with a unified format and complete information.

[0095] Example 13: Based on Example 12, the extreme value pair attribute appending unit provided in this embodiment of the invention specifically includes:

[0096] The creation and basic value writing sub-unit is configured to create an empty data record unit containing three independent value storage areas, labeled as the first storage area, the second storage area, and the third storage area, respectively; the maximum value of the confirmed rate of change is obtained and written to the first storage area; the minimum value of the confirmed rate of change is obtained and written to the second storage area; the first storage area and the second storage area store the maximum value of the rate of change and the minimum value of the rate of change, respectively.

[0097] The calculation and additional storage sub-unit is configured to read the maximum change rate from the first storage area and the minimum change rate from the second storage area; compare the maximum change rate with the minimum change rate, calculate the difference between the two, and use the difference as the change rate range amplitude; create a fourth storage area in the data recording unit, and write the calculated change rate range amplitude into the fourth storage area; the data recording unit already contains four storage areas, which respectively store the maximum change rate, the minimum change rate, the change rate range amplitude, and the reserved attribute label storage area;

[0098] The generation and completion of the sub-units are configured to generate attribute labels for the maximum rate of change, with the label content consisting of a fixed prefix upper limit of the rate of change and the corresponding concentration zone number. These attribute labels are then written to the label storage area adjacent to the first storage area in the data recording unit. Similarly, attribute labels are generated for the minimum rate of change, with the label content consisting of a fixed prefix lower limit of the rate of change and the corresponding concentration zone number. These attribute labels are then written to the label storage area adjacent to the second storage area in the data recording unit. Finally, attribute labels are generated for the range of rate of change, with the label content fixed as the range of rate of change fluctuation. These attribute labels are then written to the label storage area adjacent to the fourth storage area in the data recording unit. After the attribute labels are added, the values ​​and labels in the first, second, and fourth storage areas of the data recording unit together constitute a complete set of extreme value pairs describing the data, recording the upper and lower limits of the rate of change, the corresponding concentration zone source, and the overall fluctuation amplitude.

[0099] In the above embodiments, this embodiment converts discrete extreme value values ​​into data record units containing original values, derived values, and complete metadata. It not only retains the original numerical information of the extreme values ​​but also clarifies the data source of the extreme values ​​through additional concentration zone numbers and provides supplementary characteristics of the data distribution through range amplitude. The resulting descriptive data is self-interpretive, providing standardized, information-complete, and directly usable structured data objects.

[0100] Example 14: As Figure 6 As shown, based on Examples 1-13, the spatial monitoring method for natural resources based on multi-source remote sensing fusion provided in this embodiment of the invention includes the following steps:

[0101] Step S100: Acquire multiple remote sensing data covering the same area, including high spatial resolution panchromatic band data, multispectral reflectance data, synthetic aperture radar backscattering data, and lidar point cloud data; perform geometric fine correction and radiometric normalization on the multiple remote sensing data to eliminate sensor differences and atmospheric effects; construct a multi-scale spatial-spectral joint transformation operator to decompose the multiple remote sensing data after geometric fine correction and radiometric normalization at multiple spatial scales, extract the geometric edges, spectral absorption index, and radar polarization decomposition parameters of ground objects, perform spatial registration and fusion, and generate a collaborative feature map containing ground object material, structure, and polarization characteristics;

[0102] Step S200: Input the collaborative feature map into the pre-constructed land cover feature reference library. The land cover feature reference library consists of electromagnetic wave response data of typical land cover obtained from field measurements and laboratory measurements, covering feature vectors of different tree species, rock minerals and soil textures. Through feature matching, compare each pixel in the collaborative feature map with the feature vector in the land cover feature reference library, calculate the similarity and determine the land cover type according to the maximum likelihood principle. At the same time, use spatial context information to correct isolated pixels and output a detailed land cover type map. Each type of land cover in the map corresponds to a clear material and structural attribute.

[0103] Step S300: Overlay the detailed land cover type maps acquired at different time phases to extract areas where types have changed, and calculate the area and boundaries of the changed areas; combine topographic slope data and meteorological observation data to perform stability analysis on the changed areas, identify areas of concentrated change through spatial clustering, and classify the areas of concentrated change into early warning levels according to the rate of change and the degree of potential impact, and generate a dynamic monitoring and early warning map of natural resources to show the changing trends and risk levels of various land cover types.

[0104] In the above embodiments, this embodiment achieves systematic monitoring of the spatial distribution and dynamic changes of natural resources through collaborative processing and fusion analysis of multi-source remote sensing data. In the feature extraction stage, geometrical fine correction and radiometric normalization eliminate data differences. Combined with multi-scale spatial-spectral joint transformation and multi-source data registration, geometric edges, spectral absorption indices, and polarization decomposition parameters are extracted from panchromatic, multispectral, synthetic aperture radar, and lidar data to form a collaborative feature map that integrates material, structural, and polarization characteristics. This overcomes the limitations of a single data source and enhances the distinguishability and physical interpretability of ground feature characteristics. In the ground feature identification stage, feature matching and spatial context optimization are performed based on a measured electromagnetic wave response database to improve the accuracy of ground feature type determination and the refinement of attribute description, outputting ground feature classification results with clear material and structural attributes. In the dynamic monitoring stage, change areas are extracted through temporal overlay. Stability analysis and spatial clustering are performed using topographic and meteorological data to achieve quantitative assessment and risk classification of change areas, forming monitoring results reflecting the changing trends and early warning levels of natural resources. The various steps in this embodiment collectively construct a complete technical process from data fusion, feature extraction, land cover identification to dynamic assessment, improving the accuracy, efficiency, and decision support capabilities of natural resource monitoring.

[0105] Figure 7 A block diagram of an exemplary electronic device suitable for implementing embodiments of the present invention is shown.

[0106] Electronic devices may include a central processing unit / microprocessor / main control chip; and a storage medium coupled to the central processing unit / microprocessor / main control chip, wherein computer-executable instructions are stored for performing the steps of various methods of embodiments of the present invention when executed by a processor.

[0107] The central processing unit / microprocessor / main control chip may include, but is not limited to, one or more processors or microprocessors.

[0108] Storage media may include, but are not limited to, random access memory (RAM), read-only memory (ROM), flash memory, EPROM memory, EEPROM memory, registers, and computer storage media (such as hard disks, floppy disks, solid-state drives, removable disks, CD-ROMs, DVD-ROMs, Blu-ray discs, etc.).

[0109] In addition, the electronic device may include (but is not limited to) a data bus, an input / output bus / external bus / device bus, a display, and input / output devices (e.g., keyboard, mouse, speaker, etc.).

[0110] The central processing unit / microprocessor / main control chip can communicate with external devices via wired or wireless networks (not shown) through input / output buses / external buses / device buses.

[0111] The storage medium may also store at least one computer-executable instruction for performing the steps of various functions and / or methods in the embodiments described herein when the central processing unit / microprocessor / main control chip is running.

[0112] In one embodiment, the at least one computer-executable instruction may also be compiled into or comprise a software product, wherein one or more computer-executable instructions are executed by a processor to perform the steps of the various functions and / or methods in the embodiments described herein.

[0113] In the embodiments provided by this invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some interfaces; the indirect coupling or communication connection between systems or units may be electrical, mechanical, or other forms.

[0114] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0115] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.

[0116] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for executing all or part of the steps of the methods of the various embodiments of this invention through a computer device (which may be a personal computer, server, or network device, etc.). The aforementioned storage medium includes: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, optical disks, and other media capable of storing program code.

[0117] The above embodiments are only used to illustrate the technical solutions of the present invention, and are not intended to limit it. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A spatial monitoring system for natural resources based on multi-source remote sensing fusion, characterized in that, Include: The heterogeneous feature collaborative extraction subsystem is configured to acquire multiple remote sensing data covering the same area. The processed multiple remote sensing data are decomposed at multiple spatial scales to extract the geometric edges of ground objects, spectral absorption index and radar polarization decomposition parameters, and then spatially registered and fused to generate a collaborative feature map. The fine-grained land cover type identification subsystem is configured to compare each pixel in the collaborative feature map with the feature vector in the land cover feature reference library through feature matching, calculate the similarity and determine the land cover type according to the maximum likelihood principle; at the same time, it corrects isolated pixels and outputs a fine-grained land cover type map. The natural resource dynamic monitoring subsystem is configured to overlay detailed land cover type maps acquired at different time phases, extract areas where types have changed, and calculate the area and boundaries of the changed areas; combine topographic slope data and meteorological observation data to perform stability analysis on the changed areas, identify areas of concentrated change, and classify the areas of concentrated change into early warning levels according to the rate of change and the degree of potential impact, and generate a natural resource dynamic monitoring early warning map.

2. The natural resource spatial monitoring system based on multi-source remote sensing fusion as described in claim 1, characterized in that, Heterogeneous feature collaborative extraction subsystem, including: The displacement offset correction component is configured to spatially register the feature data layers of the extracted geometric edges, the feature data layer of the spectral absorption index, and the feature data layer of the radar polarization decomposition parameters, using the digital surface model generated from the lidar point cloud data as the elevation control benchmark, to correct the position offset due to differences in the data source. The data fusion component is configured to fuse the three types of feature data layers after registration. The feature data layer of geometric edges serves as the spatial skeleton to determine the boundaries and contours of ground features; the feature data layer of spectral absorption index serves as the material attribute filler to give the material composition information of ground features within each contour; and the feature data layer of radar polarization decomposition parameters serves as the structural attribute supplement to describe the vertical stratification and scattering mechanism of ground features within the contour. The output component is configured to generate a collaborative feature map in which each pixel simultaneously contains geometric edge information of ground features, material properties characterized by spectral absorption index, and structural and polarization characteristics characterized by radar polarization decomposition parameters.

3. The spatial monitoring system for natural resources based on multi-source remote sensing fusion as described in claim 1, characterized in that, The detailed feature type identification subsystem in this step includes: The multidimensional comparison component is configured to read data units of each pixel from the collaborative feature map, convert geometric edge information into edge intensity coefficients, and combine them with material attribute representative values ​​and three radar polarization decomposition parameters to form a multidimensional feature vector for each pixel. Simultaneously, it extracts feature vectors of all known land cover types from a pre-built land cover feature reference library. The multidimensional feature vector of each pixel is compared dimension-by-dimensionally with the feature vectors of each land cover type in the reference library, the absolute value of the difference in each dimension is calculated, and the absolute values ​​of the differences in all dimensions are accumulated to obtain the initial difference value between the pixel and each land cover type in the land cover feature reference library. The initial difference value is then converted into an initial similarity score. The preliminary type determination component is configured to retrieve the statistical distribution parameters of the feature vectors of each type of land cover in the feature reference library obtained from field measurements; establish a multidimensional probability density distribution of the land cover based on the statistical distribution parameters of the feature vectors; substitute the multidimensional feature vector of each pixel into the probability density distribution of each type of land cover to calculate the likelihood value of the pixel belonging to each type of land cover; compare the likelihood values ​​of all categories, select the land cover category with the largest likelihood value as the preliminary type determination of the pixel, and record the largest likelihood value as the determination confidence of the pixel. The isolated cell correction component is configured to overlay the geometric edge information in the preliminary type map and the collaborative feature map, using the closed polygonal contours divided by the geometric edge information as spatial units. Within each closed polygon, the preliminary judgment type and corresponding judgment confidence of all pixels are counted, the frequency of occurrence and average confidence of each type are calculated, and the type with the highest frequency and average confidence exceeding the preset threshold is selected as the dominant type of the polygon. Identify isolated cells within polygons that do not match the dominant type, and correct the type of the isolated cells to match the dominant type of the polygon they belong to; for cells located on the polygon boundary, extract the dominant type and average confidence level of their adjacent polygons; after correction, output a detailed land cover type map.

4. The spatial monitoring system for natural resources based on multi-source remote sensing fusion as described in claim 1, characterized in that, The natural resources dynamic monitoring subsystem includes: The stability index calculation component is configured to overlay the extracted change area layer with the terrain slope data layer. For each change area, it extracts the slope values ​​of all raster cells within its boundary, calculates the cumulative distribution of slope values, and takes the slope value corresponding to 90% of the cumulative distribution as the representative slope of the change area. Simultaneously, it calculates the variation range of slope values ​​within the area and combines the representative slope and variation range to generate the terrain influence coefficient of the area. Then, it overlays the change area layer with the meteorological observation data layer. For each change area, it extracts the cumulative precipitation sequence at different time phases, calculates the mean and variance of the cumulative precipitation sequence, and uses the ratio of the mean to the variance as the hydrological disturbance intensity of the change area. Finally, it multiplies the terrain influence coefficient by the hydrological disturbance intensity to obtain the stability index of each change area. The spatial clustering identification component is configured to use the geometric center point of each change region as the representative point of the change region, and the coordinates of the representative point are calculated from the region boundary. A spatial adjacency graph is constructed for all representative points, connecting representative points whose mutual distance is less than a preset distance threshold, which is twice the average width of all change regions. In the spatial adjacency graph, all connected subgraphs are extracted. The change regions corresponding to all representative points in the same connected subgraph are merged into a spatial unit. When merging, the outer envelope of the region boundary is taken as the boundary of the new unit, and the median of the region stability index is taken as the stability index of the new unit. All merged spatial units are change concentration areas, and each concentration area corresponds to a set of spatially adjacent change regions with similar stability. The early warning level classification and mapping component is configured to calculate the rate of change and potential impact for each area of ​​concentrated change. The rate of change is obtained by dividing the total area change of all changed areas within the concentrated change area at different time phases by the time interval, which is determined by the difference between the acquisition dates of the two time phases. The potential impact is determined based on the ecological service function weight values ​​corresponding to the different vegetation types within the concentrated area. The rate of change and potential impact values ​​are combined, and a two-dimensional plane is constructed with the rate of change as the horizontal axis and the potential impact as the vertical axis. The combined value of each concentrated area is mapped to a point on the two-dimensional plane. Early warning levels are classified according to preset threshold lines. All areas of concentrated change and their early warning levels are plotted on a geographic base map, with the boundaries of the areas of concentrated change filled with the color corresponding to the early warning level, and the rate of change and potential impact values ​​are labeled to generate a dynamic monitoring and early warning map of natural resources.

5. The natural resource spatial monitoring system based on multi-source remote sensing fusion as described in claim 4, characterized in that, The early warning level classification and mapping components specifically include: A sub-component for scale division is established and configured to obtain the set of change rate values ​​and the set of potential impact values ​​for all areas of concentrated change. A two-dimensional coordinate system is drawn with the change rate represented by the horizontal axis and the potential impact value represented by the vertical axis. The origin of the two-dimensional coordinate system corresponds to the intersection of the minimum change rate and the minimum potential impact value. The positive direction of the horizontal axis points to the direction of increasing change rate and the positive direction of the vertical axis points to the direction of increasing potential impact. The normalization and coordinate calculation subcomponent is configured to read the rate of change and potential impact values ​​of each region of change from the data records of each region of change; substitute the rate of change value into the horizontal axis scale division rules to determine the horizontal axis interval number to which the rate of change value belongs, and take the median of the horizontal axis interval number as the projection coordinate of the region of change on the horizontal axis; substitute the potential impact value into the vertical axis scale division rules to determine the vertical axis interval number to which the potential impact value belongs, and take the median of the vertical axis interval number as the projection coordinate of the region of change on the vertical axis; combine the horizontal axis projection coordinates and the vertical axis projection coordinates to generate the coordinate points of the region of change on the two-dimensional plane, where the horizontal coordinate value of the coordinate point is the median of the interval corresponding to the rate of change, and the vertical coordinate value is the median of the interval corresponding to the potential impact value; The geometric positioning and identification sub-component is configured to use the generated two-dimensional plane coordinate system as a base map, traverse all areas of concentrated change, and plot the coordinate points of each area of ​​concentrated change on the base map. During plotting, a circular symbol with a diameter of two units is drawn with the coordinate point as the center, and the inside of the circular symbol is filled with a color corresponding to the warning level of the area. A text label is attached to the right of each circular symbol. After the coordinate points of all areas of concentrated change are plotted, a two-dimensional scatter plot containing the combined value mapping points of all areas of concentrated change is generated.

6. The spatial monitoring system for natural resources based on multi-source remote sensing fusion as described in claim 5, characterized in that, The normalization and coordinate calculation sub-component includes: The extreme value extraction and range determination module is configured to treat the change rate values ​​of all concentrated change areas as a numerical sequence. Each value in the numerical sequence corresponds to the land cover type conversion area of ​​a concentrated area per unit time. The module iterates through the numerical sequence and compares each value with the maximum and minimum values ​​of the current record. Initially, the first value in the sequence is set as both the maximum and minimum values. After traversing all the values, the maximum and minimum values ​​in the set of change rate values ​​are obtained. The maximum value reflects the fastest speed of change in the region's natural resources, and the minimum value reflects the slowest speed. Together, they determine the overall fluctuation range of the change rate. The extreme value extraction and range determination module is configured to take the potential impact values ​​of all concentrated change areas as another numerical sequence, where each value in the numerical sequence corresponds to the degree of damage to ecosystem service functions caused by changes in land cover types in a concentrated area. Traverse the numerical sequence and use the same extreme value extraction method as the rate of change numerical value to compare each value with the maximum and minimum values ​​of the current record in turn. Initially, set the first value in the sequence as both the maximum and minimum values. Read the values ​​one by one, and after traversing all the values, obtain the maximum and minimum values ​​in the set of potential impact values. The two values ​​together determine the overall fluctuation range of the potential impact. The boundary value generation module is configured to take the minimum value of the set of change rate values ​​as the starting boundary and the maximum value as the ending boundary; between the starting boundary and the ending boundary, ten consecutive numerical intervals of equal length are generated sequentially, with the starting boundary as the left endpoint of the first interval and the ending boundary as the right endpoint of the tenth interval. The set of numerical values ​​representing potential impact is processed, with the minimum value as the starting boundary and the maximum value as the ending boundary, to generate ten consecutive numerical intervals of equal length. The scale division of the horizontal axis change rate and the vertical axis potential influence degree is completed, with each interval corresponding to a unit scale, providing a quantitative benchmark for mapping the combined values ​​of the concentrated area to a two-dimensional plane.

7. The spatial monitoring system for natural resources based on multi-source remote sensing fusion as described in claim 6, characterized in that, The extreme value extraction and range determination module includes: The extraction and double assignment submodule is configured to arrange the sequence of change rate values ​​according to the concentration zone number order, read the first value at the beginning of the concentration zone number sequence, and write the first value into both the maximum value temporary storage area and the minimum value temporary storage area. The storage unit update submodule is configured to read the second value in the rate of change value sequence in sequence, match the second value with the values ​​in the maximum value temporary storage area and the minimum value temporary storage area respectively; after the matching and updating of the second value is completed, the third value is read, and the loading, matching and updating operations are repeated until all values ​​in the rate of change value sequence have been processed. The extreme value extraction and output submodule is configured to, after traversing all values ​​in the rate of change numerical sequence, store the maximum value in the maximum value temporary storage area as the maximum value in the rate of change numerical sequence, and store the minimum value in the minimum value temporary storage area as the minimum value in the rate of change numerical sequence. Extract the values ​​in the maximum value temporary storage area as the maximum rate of change, and extract the values ​​in the minimum value temporary storage area as the minimum rate of change. The two values ​​together constitute the extreme value pair of the rate of change value set, which is output to the extreme value extraction and range for boundary value generation.

8. The spatial monitoring system for natural resources based on multi-source remote sensing fusion as described in claim 7, characterized in that, The extreme value extraction and output submodule includes: The read and mark confirmation unit is configured to read the currently stored value from the maximum value temporary storage area and mark the value as a candidate value for the maximum rate of change. Read the currently stored value from the minimum value temporary storage area and mark the value as a candidate value for the minimum rate of change; After reading, the values ​​in the two temporary storage areas are compared with the first and last values ​​of the original rate of change value sequence. Once the verification is successful, the two candidate values ​​are officially confirmed as the maximum and minimum rate of change. The extreme value pair attribute appending unit is configured to write the confirmed maximum and minimum rate of change into the same data recording unit. In the data recording unit, an attribute label is appended to the maximum rate of change, with the label content being the upper limit of the rate of change and the corresponding concentration zone number. An attribute label is appended to the minimum rate of change, with the label content being the lower limit of the rate of change and the corresponding concentration zone number. At the same time, the difference between the maximum and minimum rate of change is calculated, and the difference is appended to the data recording unit as the rate of change range amplitude. The three together constitute a complete set of extreme value pair description data. The encapsulation and output orientation unit is configured to encapsulate data record units containing the maximum value of the rate of change, the minimum value of the rate of change, the range of the rate of change, and their respective attribute labels, forming an extreme value pair data packet; an identifier is added to the header of the extreme value pair data packet, the identifier content being the extreme value pair of the rate of change; and the encapsulated extreme value pair data packet is transmitted to the output interface of the extreme value extraction and range determination module.

9. The spatial monitoring system for natural resources based on multi-source remote sensing fusion as described in claim 8, characterized in that, Extreme value pairs are attached to the attribute unit, which includes: The creation and basic value writing sub-cell is configured to create an empty data record cell, obtain the confirmed maximum rate of change and write it to the first storage area; obtain the confirmed minimum rate of change and write it to the second storage area; the first storage area and the second storage area store the maximum rate of change and the minimum rate of change, respectively; The computation and additional storage sub-unit is configured to read the maximum rate of change from the first storage area and the minimum rate of change from the second storage area; Compare the maximum rate of change with the minimum rate of change, calculate the difference between the two, and use the difference as the range of the rate of change. Create a fourth storage area in the data recording unit and write the calculated change rate range amplitude into the fourth storage area; the data recording unit already contains four storage areas, which respectively store the maximum change rate, the minimum change rate, the change rate range amplitude, and the reserved attribute label storage area. The generated and complete combined sub-units are configured to generate attribute labels for the maximum rate of change, with the label content consisting of a fixed prefix upper limit of the rate of change and the corresponding central area number. The attribute labels are written to the label storage area adjacent to the first storage area in the data recording unit. Attribute labels are generated for the minimum rate of change, with the label content consisting of a fixed prefix lower limit of the rate of change and the corresponding central area number. The attribute labels are written to the label storage area adjacent to the second storage area in the data recording unit. Attribute labels are generated for the range of the rate of change, with the label content fixed as the range of rate of change fluctuation. The attribute labels are written to the label storage area adjacent to the fourth storage area in the data recording unit. After the attribute labels are added, the values ​​and labels in the first storage area, the second storage area, and the fourth storage area of ​​the data record unit together constitute a complete set of extreme value pairs describing the data.

10. A method for spatial monitoring of natural resources based on multi-source remote sensing fusion for implementing the spatial monitoring system for natural resources based on multi-source remote sensing fusion as described in any one of claims 1 to 9, characterized in that, include: Acquire multiple remote sensing data covering the same area, including high spatial resolution panchromatic band data, multispectral reflectance data, synthetic aperture radar backscattering data, and lidar point cloud data; Geometric fine correction and radiometric normalization were performed on various remote sensing data. By constructing a multi-scale spatial-spectral joint transformation operator, various remote sensing data after geometric fine correction and radiometric normalization are decomposed at multiple spatial scales. The geometric edges, spectral absorption index and radar polarization decomposition parameters of ground objects are extracted, and spatial registration and fusion are performed to generate a collaborative feature map containing ground object material, structure and polarization characteristics. The collaborative feature map is input into a pre-built land cover feature reference library, which consists of electromagnetic wave response data of typical land cover obtained from field measurements and laboratory measurements, covering feature vectors of different tree species, rocks and minerals, and soil textures. Through feature matching, each pixel in the collaborative feature map is compared with the feature vector in the land cover feature reference library, the similarity is calculated, and the land cover type is determined according to the maximum likelihood principle. At the same time, spatial context information is used to correct isolated pixels, and a detailed land cover type map is output, in which each type of land cover corresponds to a clear material and structural attribute. By overlaying detailed land cover type maps acquired at different times, areas where types change are extracted, and the area and boundaries of these areas are calculated. Combining topographic slope data and meteorological observation data, stability analysis is performed on the changed areas. Spatial clustering methods are used to identify areas of concentrated change, and warning levels are assigned to these areas based on the rate of change and the degree of potential impact. A dynamic monitoring and warning map of natural resources is generated, showing the changing trends and risk levels of various land cover types.

Images