Method and device for evaluating InSAR observability of complex scene based on three-dimensional geometric unit

Abstract

The present application relates to synthetic aperture radar interferometry technical field, especially in three-dimensional geometry unit based on complex scene InSAR observability evaluation method and device.It includes: obtaining space data and carrying out analysis and visibility screening;The mapping relationship of the visible unit set obtained by screening is established with radar pixels;The three-dimensional geometric unit after mapping is sampled and discretized, and the initial candidate scattering point is generated;High observation potential pixel pool is constructed;Attribution determination is carried out on high observation potential scattering point;The network is constructed with representative scattering point as node, and the arc segment model interpretation index is calculated, and the core backbone point is screened;The extended scattering point and the core backbone point are networked, the maximum value of the model interpretation index in the associated arc segment of each point is extracted as the observation quality score, and the InSAR observability evaluation atlas is generated.The present application realizes the evaluation paradigm reconstruction from "posterior statistics" to "prior quantitative prediction", and provides the decision basis with physical interpretable for complex scene InSAR monitoring scheme design.

Description

Technical Field

[0001] This invention relates to the field of synthetic aperture radar interferometry technology, and in particular to a method and apparatus for assessing the observability of complex scenes using InSAR based on three-dimensional geometric units. Background Technology

[0002] In the observability assessment of Interferometric Synthetic Aperture Radar (InSAR), a "complex scene" specifically refers to a region containing surface targets with significant three-dimensional geometric features that exhibit geometric distortions in radar imaging. These geometric distortions primarily include: layover, shadowing, and topographic-geometric confusion. InSAR observability assessment is a prerequisite for ensuring the reliability of deformation monitoring in complex scenes. However, existing assessment systems suffer from paradigm gaps.

[0003] Existing technologies primarily focus on spatial correlation analysis and point density statistics of InSAR result point clouds. These methods are "post-event evaluations" and cannot quantitatively predict monitoring potential before or during monitoring implementation. Specifically, the existing technical approach involves generating a 3D point cloud (geocoded) after InSAR calculation, and then assigning semantic information through spatial location matching. In this approach, 3D geometric units do not participate in the forward InSAR calculation; they only serve as a reference frame for spatial registration after calculation. The correlation between the evaluation results and semantic information is achieved through spatial coordinate matching after calculation, which is a "post-event correlation" and lacks embedded geometric semantic support. Furthermore, the observability metric of existing technologies is the density statistics of the InSAR result point cloud, which cannot characterize the model's interpretability of the observation data and lacks physical interpretability.

[0004] Therefore, establishing a physically interpretable quantitative assessment method for InSAR observability in complex scenarios, and reconstructing the assessment paradigm from "post-event statistics" to "a priori quantitative prediction," transforming three-dimensional geometric units from "post-event registration references" to "a priori computational constraints," is of great significance for improving the reliability of InSAR monitoring in complex environments. Summary of the Invention

[0005] To address the technical problem that existing technologies cannot accurately predict InSAR monitoring potential, and the urgent need to establish a new evaluation paradigm with embedded three-dimensional geometric semantics, this invention provides a method and apparatus for evaluating the observability of InSAR in complex scenes based on three-dimensional geometric units. The technical solution is as follows: On the one hand, a method for assessing the observability of complex scenes using InSAR based on three-dimensional geometric units is provided, characterized by the following: S1. Obtain object space data of the area to be evaluated and parse it into a set of three-dimensional geometric units composed of surface targets and terrain; perform visibility filtering on the set of three-dimensional geometric units based on radar imaging geometric relationships to obtain a set of visible units; S2. Establish the mapping relationship between the set of visible units and radar pixels. Based on the radar image resolution, perform sampling discretization in the radar image coordinate system to generate initial candidate scattering points. S3. Construct a high-observation-potential pixel pool and classify the initial candidate scattering points into high-observation-potential scattering points and the first type of remainder; S4. Determine the true object space unit to which the dominant scattering point belongs, and divide the high observation potential scattering points into representative scattering points and second-class remainders; S5. Construct a network with representative scattering points as nodes and calculate the explanatory power index of the arc segment model. Select the core backbone points from the representative scattering points and classify the remaining representative scattering points into the third category of residual terms. S6. The first, second and third categories of residual terms are used as extended scattering points and core backbone points to construct a network, and the observability evaluation spectrum is output based on the maximum index values ​​of all connected arc segments and associated arc segments.

[0006] Optionally, S2 includes: Based on the geometric projection model, the set of visible units is projected onto the radar image coordinate system, radar image features are extracted and matched with the object-side model projection, and the device geometric deviation vector between the object-side and the image-side is calculated. The device uses geometric deviation vectors to perform offset compensation and align the mapping between three-dimensional geometric units and radar pixels. Based on the radar image resolution, the mapped three-dimensional geometric units are sampled and discretized in the radar image coordinate system to generate initial candidate scattering points that match the spatial distribution of radar pixels and carry object semantic information.

[0007] Optionally, calculating the device geometric deviation vector between the object side and the image side includes: Extract significant geometric features or local intensity extremum features from radar images; A search window is established in the image side, and the object-side projection and image-side observation are matched using normalized cross-correlation or intensity centroid matching methods.

[0008] Optionally, S3 includes: Statistical stability indices are extracted from time-series SAR images to construct a high-potential pixel pool. The initial candidate scattering points that fall into the high observation potential pixel pool are identified as high observation potential scattering points, and the remaining initial candidate scattering points are classified as the first type of remainder. Among them, statistical stability indicators include at least one of amplitude deviation index, temporal phase correlation coefficient, or signal-to-noise ratio.

[0009] Optionally, S4 includes: For high observation potential scattering points, if multiple three-dimensional geometric units in the pixel where the point is located have overlapping projections, the true object space unit to which the dominant scattering point in the pixel belongs is determined based on the semantic consistency prior of the geometric units. High observation potential scattering points belonging to the actual object space unit, as well as high observation potential scattering points located in non-overlapping pixels, are collectively referred to as representative scattering points. High observation potential scattering points belonging to non-real object space units are classified as the second type of remainder; Among them, at least one of the following dimensions is calculated to perform semantic consistency prior based on geometric units: signal statistics domain index, spatial coverage domain index, phase evolution domain index, and topological association domain index. An evaluation function is constructed based on an index of at least one dimension, and the three-dimensional geometric unit corresponding to the optimal value of the evaluation function is determined as the unit to which the real object belongs.

[0010] Optionally, S5 also includes: When multiple geometric units are detected to have projected overlap in the same pixel region, the proportion of non-overlapping pixels for each unit is calculated. Confidence assessment is performed based on the prior verification results of the semantic consistency of geometric units in non-overlapping pixels to eliminate false unit attributions in overlapping regions.

[0011] Optionally, S6 includes: The first type of remainder, the second type of remainder, and the third type of remainder are denoted as extended scattering points. The extended scattering points and the adjacent core backbone points are spatially networked to form associated arc segments, and their arc segment model interpretation index is calculated. The maximum value of the model interpretation index in all arc segments associated with each core backbone point and each extended scattering point is extracted and used as the observation quality score of each point. This score is then mapped back to the object space to generate the InSAR observability assessment map of the complex scene. Optionally, the extended scattering points are spatially networked with the adjacent core backbone points, including: For each extended scattering point, search for core backbone points within a preset spatial range; If a core backbone point is found, an associated arc segment is established between the extended scattering point and the core backbone point, and the model interpretability index of the arc segment is calculated.

[0012] On the other hand, a device for assessing the observability of complex scenes based on three-dimensional geometric units is provided. This device is applied to the method for assessing the observability of complex scenes based on three-dimensional geometric units in InSAR. The device includes: The geometric unit processing module is used to acquire object space data of the area to be evaluated and parse it into a set of three-dimensional geometric units composed of surface targets and terrain; based on the radar imaging geometric relationship, the three-dimensional geometric unit set is filtered for visibility to obtain a set of visible units; The mapping and sampling module is used to establish the mapping relationship between the set of visible units and radar pixels. Based on the radar image resolution, it performs sampling discretization in the radar image coordinate system to generate initial candidate scattering points. The pixel pool construction module is used to construct a high observation potential pixel pool and classify the initial candidate scattering points into high observation potential scattering points and the first type of remainder; The attribution determination module is used to determine the true object space attribution unit of the dominant scattering point and to divide the high observation potential scattering points into representative scattering points and second-class remainders. The core backbone network construction module is used to construct a network with representative scattering points as nodes and calculate the explanatory power index of the arc segment model. It selects core backbone points from the representative scattering points and classifies the remaining representative scattering points into the third category of residual terms. The hierarchical full evaluation module is used to construct a network of the first, second and third category remainders as extended scattering points and core backbone points, and outputs an observability evaluation map based on the maximum index values ​​of all connected arc segments and associated arc segments.

[0013] On the other hand, a complex scene InSAR observability assessment device based on three-dimensional geometric units is provided. The complex scene InSAR observability assessment device based on three-dimensional geometric units includes: a processor; a memory, wherein computer-readable instructions are stored in the memory, and when the computer-readable instructions are executed by the processor, any one of the methods described above for complex scene InSAR observability assessment based on three-dimensional geometric units is implemented.

[0014] On the other hand, a computer-readable storage medium is provided, wherein at least one instruction is stored therein, the at least one instruction being loaded and executed by a processor to implement any of the above-described methods for assessing the observability of complex scenes based on three-dimensional geometric units in InSAR.

[0015] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: 1. This invention realizes a paradigm reconstruction of evaluation from "post-event statistics" to "a priori quantitative prediction". By constructing three-dimensional geometric units, this invention embeds the prior geometric depth of the object into the InSAR forward calculation process. Based on the geometric unit morphology, imaging geometry and visibility analysis, it can scientifically predict the monitoring potential of complex scenes before interferometric processing, providing a decision-making basis for monitoring scheme design.

[0016] 2. The screening mechanism based on physical scattering mechanisms in this invention ensures the reliability of the evaluation source. This invention constructs a high-potential pixel pool to pre-select regions with low signal-to-noise ratios or complex scattering mechanisms based on pixel statistical characteristics. Furthermore, it combines this with the single dominant scattering assumption to perform preliminary classification of initial candidate points, laying a solid data foundation for subsequent high-precision evaluation.

[0017] 3. The geometric unit semantic consistency prior proposed in this invention enables accurate determination of scattering points in overlapping regions. Addressing the common layering phenomenon in complex scenes, this invention comprehensively evaluates four dimensions: signal statistics, projection connectivity, deformation homology, and spatial topology. This effectively resolves disputes over the attribution of overlapping projection regions and significantly reduces false alarms caused by incorrect attribution.

[0018] 4. The hierarchical evaluation architecture proposed in this invention achieves robust extension of the evaluation scope. This invention constructs a stable evaluation backbone network by selecting highly reliable core backbone points, and uses these as "anchor points" to radiate evaluation capabilities to extended scattering points through associated arc segments. This robust extension mechanism significantly improves the coverage completeness of the observable area in complex scenarios while ensuring overall evaluation robustness.

[0019] 5. The model interpretability index proposed in this invention provides a quantitative score with deep physical interpretability. Unlike traditional evaluation methods that rely solely on point cloud density, the maximum value of the model interpretability index among all arc segments associated with each point directly reflects the degree of fit between the object's geometric location and the radar phase observation sequence. This can quantitatively characterize the potential accuracy and reliability of InSAR measurement results from a physical mechanism perspective, and the resulting evaluation map has greater engineering guidance value. Attached Figure Description

[0020] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0021] Figure 1 A flowchart of an InSAR observability assessment method for complex scenes based on three-dimensional geometric units provided in an embodiment of the present invention; Figure 2 This is a schematic diagram illustrating the principle of object-image spatial mapping and pixel-level sampling discretization provided in an embodiment of the present invention; Figure 3 A flowchart illustrating the construction of a network for hierarchical screening and observability evaluation of scattering points, as provided in an embodiment of the present invention. Figure 4 A block diagram of an InSAR observability assessment device for complex scenes based on three-dimensional geometric units, provided for an embodiment of the present invention; Figure 5 This is a schematic diagram of the structure of the electronic device provided in an embodiment of the present invention. Detailed Implementation

[0022] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0023] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0024] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0025] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0026] This invention provides a method for assessing the observability of complex scenes using InSAR based on three-dimensional geometric units. This method can be implemented using a device for assessing the observability of complex scenes using InSAR based on three-dimensional geometric units, which can be a terminal or a server. Figure 1 The flowchart shown is for an InSAR observability assessment method for complex scenes based on three-dimensional geometric units. The processing flow of this method may include the following steps: S1. Obtain object space data of the area to be evaluated and parse it into a set of three-dimensional geometric units composed of surface targets and terrain; perform visibility screening on the set of three-dimensional geometric units based on radar imaging geometric relationships, remove occluded units, and obtain a set of visible units.

[0027] In one feasible implementation, the three-dimensional geometric unit includes at least one of a surface target geometric unit and a terrain geometric unit; The geometric unit of the surface target has independent geometric attributes or object-space semantic attributes; the geometric attributes include at least one of geometric normal vector, surface curvature or three-dimensional spatial orientation; the object-space semantic attributes include at least one of surface target type, material attribute or functional attribute. The topographic geometric unit has topographic geometric attributes or land cover attributes; the topographic geometric attributes include at least one of slope, aspect, topographic curvature or roughness; the land cover attributes include at least one of vegetation cover type, soil type, water body or snow cover.

[0028] In one feasible implementation, the surface target includes at least one of the following: Macro-level building complexes include residential buildings, commercial center buildings, and industrial plant buildings; Linear transportation infrastructure, including elevated roads, bridges, and rail transit facilities; Separate engineering structures, including power transmission towers, communication base stations, and large storage tanks; Fine geometric components, including building facades, roof structures, and ancillary facilities.

[0029] In one feasible implementation, the terrain geometric units are generated based on the analysis of a Digital Elevation Model (DEM) or a Digital Surface Model (DSM); the terrain includes at least one of the following: Natural topography and landforms, including mountains, hills, river valleys, alluvial plains, and steep cliffs; Artificial modification of terrain, including road cuts, embankments, dams, landfill areas, and open-pit mines; Land cover types include vegetated areas, bare soil areas, water bodies, snow and ice covered areas, and deserts.

[0030] S2. Establish the mapping relationship between the set of visible units and radar pixels. Based on the radar image resolution, perform sampling discretization in the radar image coordinate system to generate initial candidate scattering points. In one possible implementation, S2 includes: Based on the geometric projection model, the visible unit set is projected onto the radar image coordinate system, radar image features are extracted and matched with the object-side model projection, and the device geometric deviation vector between the object-side and the image-side is calculated. The device uses geometric deviation vectors to perform offset compensation and align the mapping between three-dimensional geometric units and radar pixels. Based on the radar image resolution, the mapped three-dimensional geometric units are sampled and discretized in the radar image coordinate system to generate initial candidate scattering points that match the spatial distribution of radar pixels and carry object semantic information.

[0031] Specifically, the sampling discretization process is performed in the radar image coordinate system: for each radar pixel, based on the relative positional relationship between the pixel coordinates of the pixel and the pixel coordinates of the vertex after mapping of the three-dimensional geometric unit within the pixel, the object coordinates of the sampling point of the pixel on the three-dimensional geometric unit are calculated using the spatial interpolation method, and used as the initial candidate scattering point.

[0032] In one feasible implementation, the spatial interpolation method includes bilinear interpolation or barycentric coordinate interpolation. When the projection area of ​​the three-dimensional geometric unit in the radar image coordinate system is a quadrilateral, the bilinear interpolation method is used to calculate the object coordinates of the sampling point; when the projection area is a triangle, the barycentric coordinate interpolation method is used to calculate the object coordinates of the sampling point.

[0033] The initial candidate scattering point carries object-space semantic information from its corresponding three-dimensional geometric unit, including geometric unit identifier and unit index, thereby establishing the object-space attribution mapping relationship between the initial candidate scattering point and the three-dimensional geometric unit.

[0034] In one feasible implementation, calculating the device geometric deviation vector between the object side and the image side includes: Extract significant geometric features or local intensity extremum features from radar images; A search window is established in the image side, and the object-side projection and image-side observation are matched using normalized cross-correlation or intensity centroid matching methods.

[0035] In one feasible implementation, the geometric projection model is the Range-Doppler geometric model.

[0036] In one feasible implementation, the radar image features include: significant geometric features or local intensity extremum features. The matching method can be either the normalized cross-correlation (NCC) method or the intensity centroid matching method.

[0037] in accordance with Figure 2 The object-image spatial mapping and pixel-level sampling discretization principle shown is based on the visible unit set constructed by three-dimensional geometric units, which is projected onto the radar image coordinate system along the radar line of sight according to the Range-Doppler geometric projection model. For each radar pixel, when the ground area corresponding to the pixel is covered by at least one projection unit, pixel-level sampling discretization is performed using methods such as bilinear interpolation according to the relative position of the pixel coordinates and the vertex of the projection unit covering it. Finally, initial candidate scattering points that accurately match the radar pixels in spatial distribution and fully carry the semantic information of the object are generated.

[0038] S3. Construct a high-observation-potential pixel pool and classify the initial candidate scattering points into high-observation-potential scattering points and the first type of remainder; In one possible implementation, S3 includes: Statistical stability indices are extracted from time-series SAR images to construct a high-potential pixel pool. The initial candidate scattering points that fall into the high observation potential pixel pool are identified as high observation potential scattering points, and the remaining initial candidate scattering points are classified as the first type of remainder. Among them, statistical stability indicators include at least one of amplitude deviation index, temporal phase correlation coefficient, or signal-to-noise ratio.

[0039] In one feasible implementation, the amplitude deviation index is the ratio of the standard deviation of the time series amplitude to the mean of the time series amplitude; The temporal phase correlation coefficient is the residual phase temporal coherence after removing the spatially correlated phase component; The signal-to-noise ratio (SNR) is calculated based on the energy ratio of the target window to the clutter region, and characterizes the phase stability of the target pixel.

[0040] In one feasible implementation, the statistical assumption of physical scattering characteristics refers to the single dominant scattering assumption, which states that only a single dominant scattering point exists within the pixel spatial resolution range. Based on this assumption, a pixel is considered to satisfy the condition of a single dominant scattering mechanism only when its temporal signal exhibits stable statistical characteristics.

[0041] Statistical stability indicators include at least one of the following: amplitude deviation index, temporal phase correlation coefficient, or signal-to-noise ratio.

[0042] The amplitude deviation index is the ratio of the standard deviation of the temporal amplitude to the mean of the temporal amplitude. The temporal phase correlation coefficient is based on the StaMPS (Stanford Method for Persistent Scatterers) algorithm, calculating temporal coherence by removing the spatially correlated phase components from the residual phase. The signal-to-noise ratio (SNR) index is calculated based on the energy ratio of the target window to the clutter region, characterizing the phase stability of the target pixel.

[0043] The high observation potential pixel pool is constructed by screening pixels whose statistical stability index is lower than a preset threshold. The preset threshold is determined according to the type of statistical stability index selected: when using the amplitude deviation index, the preset threshold is 0.6; when using the time-series phase correlation coefficient, the preset threshold is 0.5; and when using the signal-to-noise ratio index, the preset threshold is 3dB.

[0044] The physical meaning of the first type of remainder is: initial candidate scattering points that are eliminated because they do not meet the single dominant scattering assumption. Specifically, when there are multiple scattering points or complex scattering mechanisms within a pixel, the temporal signal stability of that pixel is low and cannot be reliably attributed to a single three-dimensional geometric unit. Therefore, its corresponding candidate scattering points are classified into the first type of remainder and do not participate in the subsequent high-precision evaluation process.

[0045] S4. Determine the true object space unit to which the dominant scattering point belongs, and divide the high observation potential scattering points into representative scattering points and second-class remainders; In one possible implementation, S4 includes: For high observation potential scattering points, if multiple three-dimensional geometric units in the pixel where the point is located have overlapping projections, the true object space unit to which the dominant scattering point in the pixel belongs is determined based on the semantic consistency prior of the geometric units. High observation potential scattering points belonging to the actual object space unit, as well as high observation potential scattering points located in non-overlapping pixels, are collectively referred to as representative scattering points. High observation potential scattering points belonging to non-real object space units are classified as the second type of remainder; Among them, at least one of the following dimensions is calculated to perform semantic consistency prior based on geometric units: signal statistics domain index, spatial coverage domain index, phase evolution domain index, and topological association domain index. An evaluation function is constructed based on an index of at least one dimension, and the three-dimensional geometric unit corresponding to the optimal value of the evaluation function is determined as the unit to which the real object belongs.

[0046] In one feasible implementation, the signal statistical domain index evaluates the degree of matching between the statistical characteristics of pixel signals and the physical scattering characteristics of three-dimensional geometric units. Spatial coverage index: evaluates the connectivity rationality of the image-side projection distribution of three-dimensional geometric units; Phase evolution domain index: evaluates the homogeneity of the physical origin of deformation at scattering points within the same three-dimensional geometric unit; Topological association domain index: evaluates the continuity of the semantic logic of the object space between adjacent cells; An evaluation function is constructed based on an index of at least one dimension, and the three-dimensional geometric unit corresponding to the optimal value of the evaluation function is determined as the unit to which the real object belongs.

[0047] In one feasible implementation, the evaluation function can be in the form of a weighted linear combination, a product fusion, or a Bayesian inference framework. Preferably, when using a weighted linear combination, the weights of each dimension index can be adaptively adjusted according to the scene type: for densely built-up areas, the weight of the spatial coverage domain index is appropriately increased; for areas with active deformation, the weight of the phase evolution domain index is appropriately increased.

[0048] The physical meaning of the second type of remainder is: high-potential scattering points that are eliminated because they cannot pass the geometric unit semantic consistency determination. Specifically, when multiple geometric units have projective overlap in the same pixel region, if a unique true object-side unit cannot be determined based on the geometric unit semantic consistency prior, the scattering point is classified into the second type of remainder to avoid evaluation errors caused by false classification.

[0049] In one feasible implementation, when multiple geometric units are detected to have projective overlap in the same pixel region, a supplementary evaluation of the overlay region can be performed: calculating the proportion of non-overlapping pixels for each geometric unit; and conducting a confidence assessment based on the prior verification results of the semantic consistency of the geometric units in the non-overlapping pixels, thereby eliminating false unit assignments in the overlay region. The confidence assessment method is as follows: if the prior verification results of the semantic consistency of the geometric units in the non-overlapping pixel region are also high, it indicates that the confidence of the three-dimensional geometric unit is high.

[0050] S5. Construct a network with representative scattering points as nodes and calculate the explanatory power index of the arc segment model. Select the core backbone points from the representative scattering points and classify the remaining representative scattering points into the third category of residual terms. In one possible implementation, S5 further includes: When multiple geometric units are detected to have projected overlap in the same pixel region, the proportion of non-overlapping pixels for each unit is calculated. Confidence assessment is performed based on the prior verification results of the semantic consistency of geometric units in non-overlapping pixels to eliminate false unit attributions in overlapping regions.

[0051] In one feasible implementation, a spatial network is constructed using representative scattering points as nodes to form connecting arc segments; temporal phase modeling is performed on the connecting arc segments to calculate the arc segment model interpretability index; based on a preset arc segment model interpretability threshold, core backbone points are selected from the representative scattering points, and representative scattering points that fail the threshold selection are classified as third-category remainder items.

[0052] In one feasible implementation, spatial mesh construction methods include, but are not limited to, Delaunay triangulation, k-nearest neighbor connections, fully connected mesh construction, or distance threshold connections. Those skilled in the art can select a suitable mesh construction method based on scene density and computational complexity.

[0053] Temporal phase modeling employs a parametric model to fit the temporal phase observations at both ends of the connecting arc segment. The parametric model includes at least one of the following parameters: elevation residual, linear deformation rate, and thermal expansion coefficient. The arc segment model interpretation index characterizes the fitting quality of the temporal phase model to the observed phase, with a value ranging from [0,1]. A larger value indicates a higher degree of interpretation of the observed data by the temporal phase model. The calculation methods for the arc segment model interpretation index include, but are not limited to, beamforming algorithms, periodogram algorithms, or spectral decomposition algorithms. Specific calculation formulas are described in Example 2.

[0054] In one feasible implementation, the preset arc segment model interpretability threshold is set to 0.7. This threshold was determined through the following experiment: measured data from typical urban and mountainous city scenarios were selected, and candidate thresholds within the range of [0.5, 0.9] were traversed. The connectivity of the network formed by the core backbone points and the stability of the arc segment phase residuals were used as evaluation indicators. When the threshold is set to 0.7, the network connectivity reaches over 85%, and the standard deviation of the arc segment phase residuals is less than 3mm, achieving a better balance between strict screening and network coverage.

[0055] The physical meaning of the third type of remainder is: in spatial network construction, isolated points cannot establish reliable topological connections with the core backbone network due to the extremely low interpretability of the temporal phase model of the connecting arc segments. Classifying them into the third type of remainder aims to prevent phase noise from propagating to the global network through low-quality arc segments, ensuring the robustness of the overall evaluation results.

[0056] S6. The first, second and third categories of residual terms are used as extended scattering points and core backbone points to construct a network, and the observability evaluation spectrum is output based on the maximum index values ​​of all connected arc segments and associated arc segments.

[0057] In one feasible implementation, S6 includes: The first type of remainder, the second type of remainder, and the third type of remainder are denoted as extended scattering points. The extended scattering points and the adjacent core backbone points are spatially networked to form associated arc segments, and their arc segment model interpretation index is calculated. The maximum value of the model interpretation index in all arc segments associated with each core backbone point and each extended scattering point is extracted and used as the observation quality score of each point. This score is then mapped back to the object space to generate an InSAR observability assessment map of complex scenes. In one feasible implementation, spatial networking is constructed between the extended scattering points and adjacent core backbone points, including: For each extended scattering point, search for core backbone points within a preset spatial range; If a core backbone point is found, an associated arc segment is established between the extended scattering point and the core backbone point, and the model interpretability index of the arc segment is calculated.

[0058] In one feasible implementation, the preset spatial range is preferably a spherical neighborhood with a radius of 50-200 meters centered on the extended scattering point, and the specific value can be adaptively determined according to the scene density and radar image resolution.

[0059] If no core backbone point is found (such as an isolated scattering point in an extremely remote area), the extended scattering point is marked as "isolated and not evaluated" and will not participate in the observation quality score calculation, but will be presented as an independent identifier in the evaluation map.

[0060] In one feasible implementation, the observation quality score is calculated as follows: For each core backbone point and extended scattering point, the maximum value of the model interpretation index among all connecting arc segments and associated arc segments connected to it is extracted as the observation quality score for that point. The observation quality score ranges from [0,1], with a larger value indicating higher InSAR observability for that object location.

[0061] In one feasible implementation, the InSAR observability assessment map is a spatial distribution map with geographic coordinates as a reference frame. It maps the observation quality scores of each scattering point back to the object space using color coding, visually displaying the spatial distribution of InSAR monitoring reliability in various regions within a complex scene. The assessment map can be presented using methods such as rasterized rendering, contour plotting, or pseudo-color overlay.

[0062] In one feasible implementation, the model interpretability index is used to characterize the degree to which the time-series phase model interprets the observed phase, including the following implementation methods: 1. Beamforming Algorithm Model Explanation Metrics The calculation formula is:

[0063] in, As a guide vector, its expression is:

[0064] For the model phase, its expression is:

[0065] in, Elevation residual (unit: meters) The linear deformation rate is expressed in meters per year. Vertical baseline (unit: meters). The time baseline is in days. Radar wavelength (unit: meters); This is the timing phase observation vector. This indicates the conjugate transpose. express Norm.

[0066] The explanatory power index of the arc segment model is the normalized amplitude ratio, which characterizes the degree of matching between the phase observation value at that point and the parameter model. The value range is [0,1]. The larger the value, the higher the degree of explanation of the mathematical model for the observation data.

[0067] Optional implementation: In some application scenarios, time-series phase modeling may only include elevation residuals. or simultaneously includes elevation residuals and linear deformation rate For heat-sensitive targets (such as bridges and railway tracks), the coefficient of thermal expansion can also be introduced. At this time, the guide vector is .

[0068] 2. Periodogram Algorithm

[0069] The formula for calculating the model explanatory power index is:

[0070] in, The number of time-series SAR images used in the calculation; The imaginary unit; For the first The observed phase of an image at the pixel location; For the first solution based on the parametric model The theoretical model phase of the image is expressed as the aforementioned model phase definition: ; This indicates the modulo operation.

[0071] The explanatory power index of the arc model is normalized phase coherence, which characterizes the consistency between the observed phase and the model phase. The value range is [0,1]. The larger the value, the higher the degree of explanation of the mathematical model for the observed data.

[0072] 3. Algorithm Selection

[0073] Those skilled in the art can choose appropriate methods for calculating model interpretability indices based on specific application scenarios, and all of these fall within the protection scope of this invention.

[0074] In one feasible implementation, "complex scene" refers to an observation area where severe geometric distortion or multi-source scattering superposition occurs due to the complex object geometry and radar side-looking imaging characteristics. Typical scenes include, but are not limited to: 1. High-density urban areas: For example, where the building density is greater than 50% or the street width is less than twice the height of adjacent buildings. In such scenarios, the visibility filtering in step S1 can effectively identify large shadow areas, while the semantic consistency determination in step S4 is specifically designed to address severe overlay distortion caused by high-rise buildings.

[0075] 2. Mountainous cities: Characterized by the coupling of undulating terrain (e.g., greater than 100 meters) and high-rise building heights. In this scenario, the three-dimensional geometric units constructed by this invention through step S1 can distinguish between terrain slope and building facades, solving the problem of "top-ground confusion".

[0076] 3. Large-scale infrastructure clusters: These include complex linear structures such as bridges, elevated roads, and rail transit. In such scenarios, the hierarchical evaluation in step S6 provides robust topological constraints for the complex linear structures through core backbone points.

[0077] Furthermore, this embodiment uses the following indicators to quantify the complexity of the scenario, which is used to assist in the robustness analysis of the evaluation results: Three-dimensional geometric complexity ( ): Defined as the number of three-dimensional geometric units per unit area. The calculation formula is as follows: The higher this metric, the more densely packed the logical branches that require attribution decisions in step S5.

[0078] Distortion pixel ratio ( : Defined as the ratio of the number of pixels causing overlay or shadow distortion to the total number of pixels. This metric directly reflects the proportion of the first type of remainder (points that do not satisfy the single dominant scattering assumption) in the S4 step, quantifying the challenge the scene poses to traditional InSAR methods.

[0079] Semantic diversity index ( ): Calculated based on Shannon entropy of surface target type:

[0080] in For the first The frequency of distribution of surface-like targets within the region. A higher semantic diversity index indicates a more significant advantage in accuracy for feature matching using semantic consistency priors.

[0081] To clearly present the complete screening and network construction logic from the initial candidate scattering points to the final observability assessment map, Figure 3 A flowchart is provided for the step-by-step selection of scattering points and the construction of the observability evaluation network.

[0082] like Figure 3As shown, initial candidate scattering points first undergo temporal stability screening. High-potential scattering points that meet statistical thresholds such as amplitude deviation index, temporal phase correlation coefficient, or signal-to-noise ratio advance to the next level, while those that fail are classified as the first category of remainders. High-potential scattering points undergo semantic consistency verification, and their true object space affiliation is determined based on multi-dimensional indicators such as signal statistics domain, spatial coverage domain, phase evolution domain, and topological association domain. Successfully identified points are upgraded to representative scattering points, while those that cannot be uniquely identified are classified as the second category of remainders. Representative scattering points are spatially networked, and arc segment model interpretability indices are calculated. Points exceeding a preset threshold are selected as core backbone points, while the rest are classified as the third category of remainders. The three categories of remainders are merged to form extended scattering points, which are then spatially networked with the core backbone points. The arc segment model interpretability indices are calculated again, and finally, the observation quality score is generated by the maximum value of the index of the associated arc segments of each point. This score is mapped back to the object space, and an InSAR observability assessment map is output.

[0083] In this embodiment of the invention, a paradigm shift from "post-hoc statistics" to "prior quantitative prediction" in evaluation is achieved, providing a physically interpretable decision-making basis for the design of InSAR monitoring schemes in complex scenarios. By constructing three-dimensional geometric units and embedding the prior geometric depth of the object into the InSAR forward calculation process, the monitoring potential of complex scenarios can be scientifically predicted based on the geometric unit morphology, imaging geometry, and visibility analysis before interferometry processing, providing a decision-making basis for monitoring scheme design.

[0084] Figure 4 This is a block diagram of an InSAR observability assessment device 400 for complex scenes based on three-dimensional geometric units, according to an exemplary embodiment. The device 400 is used in an InSAR observability assessment method for complex scenes based on three-dimensional geometric units. (Refer to...) Figure 4 The device includes a geometric unit processing module 410, a mapping and sampling module 420, a pixel pool construction module 430, a classification decision module 440, a core backbone network construction module 450, and a hierarchical full evaluation module 460. Among them: The geometric unit processing module 410 is used to acquire object space data of the area to be evaluated and parse it into a set of three-dimensional geometric units composed of surface targets and terrain; based on the radar imaging geometric relationship, the three-dimensional geometric unit set is filtered for visibility to obtain a set of visible units; The mapping and sampling module 420 is used to establish the mapping relationship between the set of visible units and radar pixels. Based on the radar image resolution, it performs sampling discretization in the radar image coordinate system to generate initial candidate scattering points. Pixel pool construction module 430 is used to construct a high observation potential pixel pool and classify the initial candidate scattering points into high observation potential scattering points and the first type of remainder; The attribution determination module 440 is used to determine the true object space attribution unit of the dominant scattering point and to divide the high observation potential scattering points into representative scattering points and second-class remainders. The core backbone network construction module 450 is used to construct a network with representative scattering points as nodes and calculate the explanatory power index of the arc segment model. It selects core backbone points from the representative scattering points and classifies the remaining representative scattering points into the third category of residual terms. The hierarchical full evaluation module 460 is used to construct a network of the first, second and third category remainders as extended scattering points and core backbone points, and output an observability evaluation map based on the maximum index values ​​of all connected arc segments and associated arc segments.

[0085] Optionally, the mapping and sampling module 420 is also used to project the set of visible units onto the radar image coordinate system based on the geometric projection model, extract radar image features and match them with the object-side model projection, and calculate the device geometric deviation vector between the object-side and the image-side. The device uses geometric deviation vectors to perform offset compensation and align the mapping between three-dimensional geometric units and radar pixels. Based on the radar image resolution, the mapped three-dimensional geometric units are sampled and discretized in the radar image coordinate system to generate initial candidate scattering points that match the spatial distribution of radar pixels and carry object semantic information.

[0086] Optionally, calculating the device geometric deviation vector between the object side and the image side includes: Extract significant geometric features or local intensity extremum features from radar images; A search window is established in the image side, and the object-side projection and image-side observation are matched using normalized cross-correlation or intensity centroid matching methods.

[0087] Optionally, the pixel pooling module 430 is also used to extract statistical stability indices from temporal SAR images to construct a high-observation-potential pixel pool. The initial candidate scattering points that fall into the high observation potential pixel pool are identified as high observation potential scattering points, and the remaining initial candidate scattering points are classified as the first type of remainder. Among them, statistical stability indicators include at least one of amplitude deviation index, temporal phase correlation coefficient, or signal-to-noise ratio.

[0088] Optionally, the attribution determination module 440 is also used to determine the true object space attribution unit of the dominant scattering point in the pixel based on the semantic consistency prior of the geometric unit if multiple three-dimensional geometric units of the pixel where the high observation potential scattering point is located have overlapping projections. High observation potential scattering points belonging to the actual object space unit, as well as high observation potential scattering points located in non-overlapping pixels, are collectively referred to as representative scattering points. High observation potential scattering points belonging to non-real object space units are classified as the second type of remainder; Among them, at least one of the following dimensions is calculated to perform semantic consistency prior based on geometric units: signal statistics domain index, spatial coverage domain index, phase evolution domain index, and topological association domain index. An evaluation function is constructed based on an index of at least one dimension, and the three-dimensional geometric unit corresponding to the optimal value of the evaluation function is determined as the unit to which the real object belongs.

[0089] Optionally, the core backbone network construction module 450 also includes: a system geometric deviation compensation submodule, used to calculate the proportion of non-overlapping pixels of each unit when multiple geometric units are detected to have projection overlap in the same pixel area; Confidence assessment is performed based on the prior verification results of the semantic consistency of geometric units in non-overlapping pixels to eliminate false unit attributions in overlapping regions.

[0090] Optionally, the hierarchical full evaluation module 460 is also used to record the first type of remainder, the second type of remainder and the third type of remainder as extended scattering points, and to spatially construct a network of the extended scattering points and the adjacent core backbone points to form related arc segments and calculate their arc segment model interpretation index. The maximum value of the model interpretation index in all arc segments associated with each core backbone point and each extended scattering point is extracted and used as the observation quality score of each point. This score is then mapped back to the object space to generate the InSAR observability assessment map of the complex scene. Optionally, the extended scattering points are spatially networked with the adjacent core backbone points, including: For each extended scattering point, search for core backbone points within a preset spatial range; If a core backbone point is found, an associated arc segment is established between the extended scattering point and the core backbone point, and the model interpretability index of the arc segment is calculated.

[0091] Figure 5 This is a schematic diagram of the structure of an InSAR observability assessment device for complex scenes based on three-dimensional geometric units, provided in an embodiment of the present invention. Figure 5 As shown, a complex scene InSAR observability assessment device based on three-dimensional geometric units may include the above-mentioned... Figure 4 The diagram illustrates an InSAR observability assessment device for complex scenes based on three-dimensional geometric units. Optionally, the InSAR observability assessment device 410 for complex scenes based on three-dimensional geometric units may include a first processor 2001.

[0092] Optionally, a complex scene InSAR observability assessment device 410 based on three-dimensional geometric units may also include a memory 2002 and a transceiver 2003.

[0093] The first processor 2001, memory 2002, and transceiver 2003 can be connected via a communication bus.

[0094] The following is combined with Figure 5 A detailed description of the various components of an InSAR observability assessment device 410 for complex scenes based on three-dimensional geometric units is provided below: The first processor 2001 is the control center of a complex scene InSAR observability assessment device 410 based on three-dimensional geometric units. It can be a single processor or a collective term for multiple processing elements. For example, the first processor 2001 can be one or more central processing units (CPUs), application-specific integrated circuits (ASICs), or one or more integrated circuits configured to implement embodiments of the present invention, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0095] Optionally, the first processor 2001 can perform various functions of a complex scene InSAR observability assessment device 410 based on three-dimensional geometric units by running or executing software programs stored in the memory 2002 and calling data stored in the memory 2002.

[0096] In a specific implementation, as one example, the first processor 2001 may include one or more CPUs, for example... Figure 5 CPU0 and CPU1 are shown in the diagram.

[0097] In a specific implementation, as one example, a complex scene InSAR observability assessment device 410 based on three-dimensional geometric units may also include multiple processors, for example... Figure 5 The first processor 2001 and the second processor 2004 are shown in the diagram. Each of these processors can be a single-core processor or a multi-core processor. Here, a processor can refer to one or more devices, circuits, and / or processing cores used to process data (such as computer program instructions).

[0098] The memory 2002 is used to store the software program that executes the present invention, and is controlled by the first processor 2001 to execute it. The specific implementation method can be referred to the above method embodiment, and will not be repeated here.

[0099] Optionally, the memory 2002 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of carrying or storing desired program code in the form of instructions or data structures and accessible by a computer, but not limited thereto. The memory 2002 may be integrated with the first processor 2001 or may exist independently, and may be connected via an interface circuit of a complex scene InSAR observability assessment device 410 based on three-dimensional geometric units. Figure 5 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0100] The transceiver 2003 is used to communicate with network devices or with terminal devices.

[0101] Alternatively, transceiver 2003 may include a receiver and a transmitter. Figure 5 (Not shown separately). The receiver is used to implement the receiving function, and the transmitter is used to implement the transmitting function.

[0102] Optionally, the transceiver 2003 can be integrated with the first processor 2001 or exist independently, and can be connected to the interface circuit of a complex scene InSAR observability assessment device 410 based on three-dimensional geometric units. Figure 5 (Not shown in the image) is coupled to the first processor 2001, and this embodiment of the invention does not specifically limit this.

[0103] It should be noted that, Figure 5The structure of a complex scene InSAR observability assessment device 410 based on three-dimensional geometric units shown in the figure does not constitute a limitation on the router. Actual knowledge structure identification devices may include more or fewer components than shown, or combine certain components, or have different component arrangements.

[0104] Furthermore, the technical effect of a complex scene InSAR observability assessment device 410 based on three-dimensional geometric units can be referred to the technical effect of a complex scene InSAR observability assessment method based on three-dimensional geometric units described in the above method embodiments, and will not be repeated here.

[0105] It should be understood that the first processor 2001 in the embodiments of the present invention may be a central processing unit (CPU), or it may be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. The general-purpose processor may be a microprocessor or any conventional processor, etc.

[0106] It should also be understood that the memory in the embodiments of the present invention can be volatile memory or non-volatile memory, or may include both volatile and non-volatile memory. The non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory. The volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of random access memory (RAM) are available, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced synchronous DRAM (ESDRAM), synchronous linked DRAM (SLDRAM), and direct rambus RAM (DR RAM).

[0107] The above embodiments can be implemented, in whole or in part, by software, hardware (such as circuits), firmware, or any other combination thereof. When implemented using software, the above embodiments can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable sensors. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that includes one or more sets of available media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive.

[0108] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0109] It should be understood that, in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0110] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0111] 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.

[0112] In addition, 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.

[0113] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or a 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 to cause a computer device (which may be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention.

[0114] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for assessing the observability of complex scenes using InSAR based on three-dimensional geometric units, characterized in that, The method includes: S1. Obtain object space data of the area to be evaluated and parse it into a set of three-dimensional geometric units composed of surface targets and terrain; perform visibility filtering on the set of three-dimensional geometric units based on radar imaging geometric relationships to obtain a set of visible units; S2. Establish the mapping relationship between the set of visible units and radar pixels. Based on the radar image resolution, perform sampling discretization in the radar image coordinate system to generate initial candidate scattering points. S3. Construct a high-observation-potential pixel pool and classify the initial candidate scattering points into high-observation-potential scattering points and the first type of remainder; S4. Determine the true object space unit to which the dominant scattering point belongs, and divide the high observation potential scattering points into representative scattering points and second-class remainders; S5. Construct a network with representative scattering points as nodes and calculate the explanatory power index of the arc segment model. Select the core backbone points from the representative scattering points and classify the remaining representative scattering points into the third category of residual terms. S6. The first, second and third categories of residual terms are used as extended scattering points and core backbone points to construct a network, and the observability evaluation spectrum is output based on the maximum index values ​​of all connected arc segments and associated arc segments.

2. The InSAR observability assessment method for complex scenes based on three-dimensional geometric units according to claim 1, characterized in that, S2 includes: Based on the geometric projection model, the visible unit set is projected onto the radar image coordinate system, radar image features are extracted and matched with the object-side model projection, and the device geometric deviation vector between the object-side and the image-side is calculated. The device uses geometric deviation vectors to perform offset compensation and align the mapping between three-dimensional geometric units and radar pixels. Based on the radar image resolution, the mapped three-dimensional geometric units are sampled and discretized in the radar image coordinate system to generate initial candidate scattering points that match the spatial distribution of radar pixels and carry object semantic information.

3. The InSAR observability assessment method for complex scenes based on three-dimensional geometric units according to claim 2, characterized in that, The device for calculating the geometric deviation vector between the object and image sides includes: Extract significant geometric features or local intensity extremum features from radar images; A search window is established in the image side, and the object-side projection and image-side observation are matched using normalized cross-correlation or intensity centroid matching methods.

4. The InSAR observability assessment method for complex scenes based on three-dimensional geometric units according to claim 3, characterized in that, S3 includes: Statistical stability indices are extracted from time-series SAR images to construct a high-potential pixel pool. The initial candidate scattering points that fall into the high observation potential pixel pool are identified as high observation potential scattering points, and the remaining initial candidate scattering points are classified as the first type of remainder. Among them, statistical stability indicators include at least one of amplitude deviation index, temporal phase correlation coefficient, or signal-to-noise ratio.

5. The InSAR observability assessment method for complex scenes based on three-dimensional geometric units according to claim 4, characterized in that, S4 includes: For high observation potential scattering points, if multiple three-dimensional geometric units in the pixel where the point is located have overlapping projections, the true object space unit to which the dominant scattering point in the pixel belongs is determined based on the semantic consistency prior of the geometric units. High observation potential scattering points belonging to the actual object space unit, as well as high observation potential scattering points located in non-overlapping pixels, are collectively referred to as representative scattering points. High observation potential scattering points belonging to non-real object space units are classified as the second type of remainder; Among them, at least one of the following dimensions is calculated to perform semantic consistency prior based on geometric units: signal statistics domain index, spatial coverage domain index, phase evolution domain index, and topological association domain index. An evaluation function is constructed based on an index of at least one dimension, and the three-dimensional geometric unit corresponding to the optimal value of the evaluation function is determined as the unit to which the real object belongs.

6. The InSAR observability assessment method for complex scenes based on three-dimensional geometric units according to claim 5, characterized in that, The S5 also includes: When multiple geometric units are detected to have projected overlap in the same pixel region, the proportion of non-overlapping pixels for each unit is calculated. Confidence assessment is performed based on the prior verification results of the semantic consistency of geometric units in non-overlapping pixels to eliminate false unit attributions in overlapping regions.

7. The InSAR observability assessment method for complex scenes based on three-dimensional geometric units according to claim 6, characterized in that, S6 includes: The first type of remainder, the second type of remainder, and the third type of remainder are denoted as extended scattering points. The extended scattering points and the adjacent core backbone points are spatially networked to form associated arc segments, and their arc segment model interpretation index is calculated. By summarizing the connecting arc segments and associated arc segments, the maximum value of the model interpretability index is extracted from all arc segments associated with each core backbone point and each extended scattering point, which is used as the observation quality score for each point. This score is then mapped back to the object space to generate the InSAR observability assessment map of the complex scene.

8. The InSAR observability assessment method for complex scenes based on three-dimensional geometric units according to claim 7, characterized in that, The step of spatially networking the extended scattering points with the adjacent core backbone points includes: For each extended scattering point, search for core backbone points within a preset spatial range; If a core backbone point is found, an associated arc segment is established between the extended scattering point and the core backbone point, and the model interpretability index of the arc segment is calculated.

9. A device for evaluating the observability of complex scenes based on three-dimensional geometric units in InSAR, wherein the device is used to implement the method for evaluating the observability of complex scenes based on three-dimensional geometric units in InSAR as described in any one of claims 1-8, characterized in that, The device includes: The geometric unit processing module is used to acquire object space data of the area to be evaluated and parse it into a set of three-dimensional geometric units composed of surface targets and terrain; based on the radar imaging geometric relationship, the three-dimensional geometric unit set is filtered for visibility to obtain a set of visible units; The mapping and sampling module is used to establish the mapping relationship between the set of visible units and radar pixels. Based on the radar image resolution, it performs sampling discretization in the radar image coordinate system to generate initial candidate scattering points. The pixel pool construction module is used to construct a high observation potential pixel pool and classify the initial candidate scattering points into high observation potential scattering points and the first type of remainder; The attribution determination module is used to determine the true object space attribution unit of the dominant scattering point and to divide the high observation potential scattering points into representative scattering points and second-class remainders. The core backbone network construction module is used to construct a network with representative scattering points as nodes and calculate the explanatory power index of the arc segment model. It selects core backbone points from the representative scattering points and classifies the remaining representative scattering points into the third category of residual terms. The hierarchical full evaluation module is used to construct a network of the first, second and third category remainders as extended scattering points and core backbone points, and outputs an observability evaluation map based on the maximum index values ​​of all connected arc segments and associated arc segments.

10. A device for assessing the observability of complex scenes using InSAR based on three-dimensional geometric units, characterized in that, The InSAR observability assessment device for complex scenes based on three-dimensional geometric units includes: A processor; a memory storing computer-readable instructions that, when executed by the processor, implement any one of the InSAR observability assessment methods for complex scenes based on three-dimensional geometric units as described in any one of claims 1-8.

Images