Aeromagnetic false anomaly elimination and hierarchical constraint reconstruction method based on multi-source remote sensing cooperation

Abstract

The embodiment of the invention provides a multi-source remote sensing collaborative aeromagnetic false anomaly elimination and hierarchical constraint reconstruction method, which relates to the technical field of geophysical exploration, and comprises the following steps: performing interference anomaly capture on multi-temporal aeromagnetic observation data to position a plurality of candidate aeromagnetic anomaly regions, collecting a plurality of multi-source remote sensing data sets to carry out false anomaly multi-source cooperative identification on the plurality of candidate aeromagnetic anomaly regions to generate a false anomaly mask region; and performing multi-scale hierarchical division on the false anomaly mask region to obtain a multi-scale blank region, and iteratively executing multi-scale cooperative constraint reconstruction and outputting purified aeromagnetic data. The problems that humanity interference false anomalies in aeromagnetic data are difficult to accurately recognize and remove, magnetic field reconstruction is distorted after anomalies are removed, and geological structure interpretation precision and mineral resource exploration reliability are affected are solved. The technical effects of thoroughly eliminating aeromagnetic false anomaly interference, obtaining reliable aeromagnetic data which completely retains real magnetic anomaly characteristics generated by a deep geologic structure and providing a high-quality data basis for geophysical exploration are achieved.

Description

Technical Field

[0001] This invention relates to the field of geophysical exploration technology, and in particular to a method for eliminating false aeromagnetic anomalies and reconstructing hierarchical constraints using multi-source remote sensing collaboration. Background Technology

[0002] Existing aeromagnetic false anomaly processing technologies mainly rely on a single remote sensing data source or human experience in the identification stage, which has obvious limitations in capturing surface interference sources. Specifically, multispectral images can only identify the shape of surface facilities but cannot confirm the metal material properties. Although hyperspectral data can analyze the spectral characteristics of materials, it is limited by spatial resolution and cannot accurately locate small interference sources. Furthermore, synthetic aperture radar is susceptible to interference from surface roughness due to the high reflectivity of metals, resulting in false alarms. This leads to insufficient reliability in determining the spatial correlation between interference sources and aeromagnetic anomalies, often resulting in missed detections or incorrect rejections of false anomalies.

[0003] During the data reconstruction phase, a globally uniform interpolation or fitting strategy was used for the blank areas formed after removal. No differentiated processing was implemented based on the blank scale, morphological complexity, and geological background. This resulted in low efficiency for small-scale blank repair due to excessive algorithm complexity, and insufficient accuracy for large-scale area reconstruction due to the lack of hierarchical constraints. In particular, when there are multiple discrete blank areas, the traditional serial processing mode cannot utilize similar features to achieve parallel computing, which significantly restricts the processing speed of massive amounts of data. At the same time, in complex tectonic areas, the failure to effectively separate regional field components led to the introduction of artificial high-frequency noise or distortion of the real geological trend into the reconstructed magnetic field, which disrupted spatial continuity and reduced the reliability of geological interpretation.

[0004] In summary, existing technologies suffer from the problem that it is difficult to accurately identify and remove false anomalies caused by human interference in aeromagnetic data, and the magnetic field reconstruction is distorted after anomaly removal, which affects the accuracy of geological structure interpretation and the reliability of mineral resource exploration.

[0005] It should be noted that the information disclosed in this background section is intended only to enhance the understanding of the overall background of the invention and should not be construed as an admission or in any way implying that the information constitutes prior art known to those skilled in the art. Summary of the Invention

[0006] To address the aforementioned shortcomings or improvement needs of existing technologies, this invention provides a multi-source remote sensing collaborative method for aeromagnetic false anomaly removal and hierarchical constraint reconstruction. This method solves the problems of difficulty in accurately identifying and removing false anomalies caused by human interference in aeromagnetic data, and the resulting distortion of the magnetic field reconstruction after anomaly removal, which affects the accuracy of geological structure interpretation and the reliability of mineral resource exploration. It achieves the technical effect of completely eliminating aeromagnetic false anomaly interference and obtaining reliable aeromagnetic data that fully preserves the true magnetic anomaly characteristics generated by deep geological structures, providing a high-quality data foundation for geophysical exploration. The specific technical solution is as follows:

[0007] This invention provides a method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction, the method comprising:

[0008] Interference anomalies are captured from multi-temporal aeromagnetic observation data to locate multiple candidate aeromagnetic anomaly regions. Based on the spatial boundaries of these candidate regions, multi-modal remote sensing satellite data is acquired to obtain multiple multi-source remote sensing datasets. False anomaly mask regions are generated by multi-source collaborative identification of these candidate regions based on the multi-source remote sensing datasets. Based on differences in anomaly scale and geological background, the false anomaly mask regions are divided into multi-scale hierarchical regions, resulting in small-scale regular blank areas, mesoscale continuous blank areas, and large-scale structurally complex blank areas. Multi-scale collaborative constraint reconstruction is iteratively performed on these small-scale regular blank areas, mesoscale continuous blank areas, and large-scale structurally complex blank areas to output purified aeromagnetic data.

[0009] In one implementation, multi-scale collaborative constraint reconstruction is iteratively performed on the small-scale regular blank areas, medium-scale continuous blank areas, and large-scale structurally complex blank areas to output the purified aeromagnetic data. The following processing is also performed:

[0010] After clustering and grouping the small-scale regular blank areas, directional interpolation reconstruction is performed in parallel to obtain primary reconstructed magnetic field data. The primary reconstructed magnetic field data is then integrated into the mesoscale continuous blank areas as constraint data, and adaptive interpolation repair is performed on the mesoscale continuous blank areas to obtain intermediate reconstructed magnetic field data. The intermediate reconstructed magnetic field data is then integrated into the large-scale complex blank areas as known data, and conditional constraint reconstruction is performed to output advanced reconstructed magnetic field data. The primary reconstructed magnetic field data, intermediate reconstructed magnetic field data, and advanced reconstructed magnetic field data are then subjected to layer-by-layer iterative verification and fusion to output the purified aeromagnetic data.

[0011] In one implementation, based on the multiple multi-source remote sensing datasets, false anomaly multi-source collaborative identification is performed on the multiple candidate aeromagnetic anomaly regions to generate a false anomaly mask region, and the following processing is also performed:

[0012] Multi-threaded collaborative interpretation processing is performed on the multiple multi-source remote sensing datasets to output multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflective target distribution maps. Using the multiple candidate aeromagnetic anomaly regions as reference benchmarks, the multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflective target distribution maps are spatially overlaid, and multi-evidence fusion false anomaly determination is performed to generate the false anomaly mask region.

[0013] In one implementation, using the plurality of candidate aeromagnetic anomaly regions as a reference, the plurality of multispectral target distribution maps, the plurality of hyperspectral metallic material distribution maps, and the plurality of SAR strong reflectivity target distribution maps are spatially overlaid. Multi-evidence fusion is performed to determine false anomalies, generating the false anomaly mask region. The following processing is also performed:

[0014] Based on a pre-defined spatial resolution, a first candidate aeromagnetic anomaly region is rasterized to obtain multiple analysis units. According to these multiple analysis units, multi-source evidence is extracted from a spatially superimposed first multispectral target distribution map, a first hyperspectral metallic material distribution map, and a first SAR strong reflectivity target distribution map to obtain multiple sets of interference evidence. Based on the mapping of these multiple sets of interference evidence, the false anomaly properties of the multiple analysis units are interpreted, outputting multiple aeromagnetic anomaly judgment results. A first false anomaly mask partition is separated from the first candidate aeromagnetic anomaly region based on the multiple aeromagnetic anomaly judgment results. Similarly, the false anomaly interpretation and separation of the multiple candidate aeromagnetic anomaly regions are performed to obtain multiple false anomaly mask partitions. The multiple false anomaly mask partitions are spatially fused to obtain the false anomaly mask region.

[0015] In one implementation, after clustering the small-scale regular blank areas, directional interpolation reconstruction is performed in parallel to obtain primary reconstructed magnetic field data, and the following processing is also performed:

[0016] Based on spatial proximity, the small-scale regular blank areas are clustered to form multiple blank sub-region processing groups. According to the outer boundaries of these multiple blank sub-region processing groups, the multi-temporal aeromagnetic observation data is segmented into multiple constraint boundary condition groups. Based on the aeromagnetic survey line directions of the multi-temporal aeromagnetic observation data and the multiple regional geological structure orientation groups of the multiple blank sub-region processing groups, multiple dominant reconstruction direction groups are defined. Using the multiple constraint boundary condition groups as interpolation reconstruction boundary constraints, spatial interpolation reconstruction of directional constraints is performed in parallel on the multiple blank sub-region processing groups according to the multiple dominant reconstruction direction groups, resulting in multiple local reconstructed magnetic field data groups. These multiple local reconstructed magnetic field data groups are seamlessly overlaid onto the multi-temporal aeromagnetic observation data to obtain the primary reconstructed magnetic field data.

[0017] In one implementation, after incorporating the primary reconstructed magnetic field data as constraint data into the mesoscale continuous blank area, adaptive interpolation repair of the mesoscale continuous blank area is performed to obtain intermediate reconstructed magnetic field data, and the following processing is further performed:

[0018] The primary reconstructed magnetic field data is projected onto the mesoscale continuous blank area, and spatially adjacent data is extracted to obtain a local constraint dataset. Trend analysis is performed on the local constraint dataset to extract an adaptive trend surface. Using the adaptive trend surface as a baseline constraint, parameter-adaptive spatial interpolation repair is performed on the mesoscale continuous blank area to obtain repaired magnetic field data. The repaired magnetic field data is integrated into the primary reconstructed magnetic field data to obtain the intermediate reconstructed magnetic field data.

[0019] In one implementation, after integrating the intermediate-level reconstructed magnetic field data as known data into the large-scale structurally complex blank area, conditional constraint reconstruction is performed to output high-level reconstructed magnetic field data, and the following processing is also performed:

[0020] The peripheral real observation data of the large-scale tectonically complex blank area is extracted from the multi-temporal aeromagnetic observation data; the peripheral real observation data and intermediate reconstructed magnetic field data are integrated to construct a known data constraint body; frequency domain filtering is performed on the known data constraint body to extract the long-wavelength regional background magnetic field component as the main control trend surface; regional geological distribution information and gravity anomaly data are introduced to construct geological model constraints and geophysical collaborative constraints respectively; using the main control trend surface as a framework, the geological model constraints and geophysical collaborative constraints are integrated to perform iterative reconstruction optimization under conditional constraints on the large-scale tectonically complex blank area, and conditional constraint reconstructed magnetic field data is output; the conditional constraint reconstructed magnetic field data is integrated into the intermediate reconstructed magnetic field data to output the advanced reconstructed magnetic field data.

[0021] In one implementation, interference anomaly capture is performed on multi-temporal aeromagnetic observation data to locate multiple candidate aeromagnetic anomaly regions, and the following processing is also performed:

[0022] After acquiring multi-temporal raw observation data of the target survey area, gridded correction processing is performed to obtain the multi-temporal aeromagnetic observation data. Based on regional background magnetic field simulation, the regional background magnetic field is separated from the multi-temporal aeromagnetic observation data to obtain local magnetic anomaly data. Based on the local spatial variation characteristics of the local magnetic anomaly data, an anomaly capture threshold is set. According to the anomaly capture threshold, significant anomaly region extraction based on statistical threshold segmentation is performed on the local magnetic anomaly data to obtain local anomaly regions. After spatially aggregating the local anomaly regions, boundary regularization processing is performed to form the multiple candidate aeromagnetic anomaly regions.

[0023] In one implementation, high-resolution multispectral imagery, hyperspectral remote sensing data, and SAR data are extracted from a first multi-source remote sensing dataset, and parallel collaborative interpretation is performed.

[0024] S1: After performing land cover classification processing on the high-resolution multispectral image, human target identification and extraction are performed based on the classification results to generate a first multispectral target distribution map; S2: By performing land material spectral feature matching on the hyperspectral remote sensing data, strong magnetic metal materials are identified, and a first hyperspectral metal material distribution map is output; S3: The SAR data is progressively preprocessed with radiometric calibration, multi-view processing, and geocoding to obtain SAR images. Then, high backscattering target detection is performed on the SAR images to extract strong metal reflectors and generate a first SAR strong reflective target distribution map.

[0025] Beneficial effects of the embodiments of the present invention:

[0026] The solution provided in this invention involves capturing interference anomalies in multi-temporal aeromagnetic observation data to locate multiple candidate aeromagnetic anomaly regions. Based on the spatial boundaries of these candidate regions, multi-modal remote sensing satellite data is acquired to obtain multiple multi-source remote sensing datasets. False anomaly multi-source collaborative identification is performed on these datasets to generate false anomaly mask areas. Based on differences in anomaly scale and geological background, the false anomaly mask areas are divided into multi-scale hierarchical regions, resulting in small-scale regular blank areas, mesoscale continuous blank areas, and large-scale structurally complex blank areas. Iterative multi-scale collaborative constraint reconstruction is performed on these small-scale regular blank areas, mesoscale continuous blank areas, and large-scale structurally complex blank areas to output purified aeromagnetic data. This achieves the technical effect of completely eliminating aeromagnetic false anomaly interference and obtaining reliable aeromagnetic data that fully preserves the true magnetic anomaly characteristics generated by deep geological structures, providing a high-quality data foundation for geophysical exploration. Of course, implementing any product or method of this invention does not necessarily require achieving all of the advantages described above simultaneously. Attached Figure Description

[0027] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0028] Figure 1 The flowchart of the multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction method provided by the present invention is shown.

[0029] Figure 2 The diagram illustrates the process of locating candidate aeromagnetic anomaly regions in the multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction method provided by the present invention. Detailed Implementation

[0030] To facilitate understanding of the present invention, a more complete description of the invention will be given below with reference to the accompanying drawings, which illustrate preferred embodiments of the invention. However, the invention can be implemented in many different forms and is not limited to the embodiments described herein; rather, these embodiments are provided to enable a more thorough and complete understanding of the disclosure of the invention.

[0031] Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0032] In the description of this invention, it should be understood that the terms "center," "longitudinal," "lateral," "length," "width," "thickness," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," "outer," "clockwise," and "counterclockwise," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0033] Unless otherwise expressly stated, throughout the specification and claims, the term "comprising" or its variations such as "including" or "comprises" shall be understood to include the stated elements or components without excluding other elements or other components.

[0034] The present invention provides a multi-source remote sensing collaborative method for the removal of false anomalies in aeromagnetic data and hierarchical constraint reconstruction, which is used to solve the problem that it is difficult to accurately identify and remove false anomalies caused by human interference in aeromagnetic data, and that the magnetic field reconstruction is distorted after the anomaly removal, which affects the accuracy of geological structure interpretation and the reliability of mineral resource exploration.

[0035] Example: See Figure 1 The flowchart of the multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction method provided in this embodiment of the invention includes:

[0036] A100: Captures interference anomalies in multi-temporal aeromagnetic observation data to locate multiple candidate aeromagnetic anomaly regions.

[0037] In one implementation, see Figure 2 As shown, interference anomaly capture is performed on multi-temporal aeromagnetic observation data to locate multiple candidate aeromagnetic anomaly regions. Step A100 may further include:

[0038] A110: After acquiring the multi-temporal raw observation data of the target survey area, perform gridding correction processing to obtain the multi-temporal aeromagnetic observation data.

[0039] A120: Based on the simulation of the regional background magnetic field, the regional background magnetic field is separated from the multi-temporal aeromagnetic observation data to obtain local magnetic anomaly data.

[0040] A130: Based on the local spatial variation characteristics of the local magnetic anomaly data, set the anomaly capture threshold.

[0041] A140: Based on the anomaly capture threshold, perform significant anomaly region extraction based on statistical threshold segmentation on the local magnetic anomaly data to obtain the local anomaly region.

[0042] A150: After spatially aggregating the local anomalous regions, the multiple candidate aeromagnetic anomalous regions are formed by performing boundary regularization.

[0043] The multi-temporal aeromagnetic observation data is a collection of aeromagnetic measurement data acquired in the target survey area through different voyages and at different times. Utilizing its time dimension information helps to distinguish stable geological fields from potentially changing artificial interference signals.

[0044] This embodiment captures interference anomalies in the multi-temporal aeromagnetic observation data to identify local magnetic field distortion regions that are significantly different from the surrounding background in terms of amplitude, gradient, or morphology from massive background values, thereby obtaining the multiple candidate aeromagnetic anomaly regions.

[0045] Specifically, in this embodiment, unprocessed multi-temporal raw observation data of the target survey area is collected from a data warehouse or measurement unit. The multi-temporal raw observation data includes magnetic field strength, coordinates and auxiliary information recorded along the flight trajectory of different flights. Its spatial distribution is irregular and there are systematic measurement errors.

[0046] Since aeromagnetic data is usually collected continuously along the survey line and the data is irregularly distributed in space, this embodiment uses mathematical methods such as Kriging interpolation to interpolate discrete measurement point data into regular grid nodes under a unified geographic coordinate system to generate a continuously covered spatial raster. At the same time, it corrects non-geological field distortions caused by instrument drift, diurnal variation interference, and flight altitude fluctuations, and finally forms the multi-temporal aeromagnetic observation data that is spatially aligned, formatted, and noise-controlled, providing a benchmark data base for subsequent analysis.

[0047] It should be understood that the Earth's magnetic field includes a long-wave background component generated by the interaction of the Earth's core magnetic field and regional tectonic activity. This type of large-scale field signal can mask the anomalous features of shallow targets. Therefore, this embodiment constructs a mathematical model by simulating the regional background magnetic field and uses trend surface analysis or low-pass filtering algorithms to quantitatively estimate the long-wave background field component formed by the combined interaction of the Earth's core magnetic field and regional geological structures.

[0048] Subsequently, the background field component is extracted from the multi-temporal aeromagnetic observation data through numerical calculations, generating local magnetic anomaly data in a regular grid format dominated by medium and short wave signals. The local magnetic anomaly data centrally presents magnetic field distortion caused by local magnetic bodies or human interference sources, constituting the core input for anomaly capture.

[0049] Based on the preset abnormal field value threshold, the grid cells are traversed, and connected cells with field values ​​exceeding the threshold are clustered into independent spatial patches. The output set of local abnormal regions has clear boundaries and intensity attributes, but due to threshold sensitivity, there may be fragmentation or jagged boundary phenomena, which require subsequent optimization processing.

[0050] By setting a spatial distance tolerance, spatially correlated fragmented regions within the local anomaly areas are merged using morphological dilation or clustering algorithms to form physically complete anomaly units. The boundary regularization process employs techniques such as convex hull generation, polygon simplification, and Bézier curve smoothing to eliminate irregular geometric features, ensuring the contours conform to typical artificial target shapes. The resulting candidate aeromagnetic anomaly regions possess both spatial continuity and morphological rationality, allowing for direct correlation with multi-source remote sensing data for collaborative identification.

[0051] This embodiment obtains candidate aeromagnetic anomaly regions that combine spatial continuity and morphological rationality through a multi-level data cleaning and morphological optimization process. While achieving the technical effect of efficiently eliminating measurement noise interference and accurately capturing potential human interference targets, it indirectly provides high-confidence spatial target area input for subsequent multi-source remote sensing collaborative identification processes.

[0052] A200: Based on the spatial boundaries of the multiple candidate aeromagnetic anomaly regions, multimodal remote sensing satellite data is acquired to obtain multiple multi-source remote sensing datasets. The data types of each multi-source remote sensing dataset include high-resolution multispectral images, hyperspectral remote sensing data, and SAR data.

[0053] A300: Based on the multiple multi-source remote sensing datasets, perform multi-source collaborative identification of false anomalies in the multiple candidate aeromagnetic anomaly regions to generate a false anomaly mask region.

[0054] In one implementation, based on the multiple multi-source remote sensing datasets, multi-source collaborative identification of false anomalies is performed on the multiple candidate aeromagnetic anomaly regions to generate a false anomaly mask region. Step A300 may further include:

[0055] A310: Perform multi-threaded collaborative interpretation processing on the multiple multi-source remote sensing datasets to output multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflectivity target distribution maps.

[0056] A320: Using the multiple candidate aeromagnetic anomaly regions as reference benchmarks, spatially overlay the multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflective target distribution maps, perform multi-evidence fusion false anomaly determination, and generate the false anomaly mask region.

[0057] In one implementation, high-resolution multispectral imagery, hyperspectral remote sensing data, and SAR data are extracted from a first multi-source remote sensing dataset, and parallel collaborative interpretation is performed.

[0058] S1: After performing land cover classification processing on the high-resolution multispectral image, human target identification and extraction are performed based on the classification results to generate a first multispectral target distribution map.

[0059] S2: By performing surface material spectral feature matching on the hyperspectral remote sensing data, strongly magnetic metal materials are identified, and a first hyperspectral metal material distribution map is output.

[0060] S3: Perform radiometric calibration, multi-view processing and geocoding preprocessing on the SAR data to obtain SAR images. Then, perform high backscattering target detection on the SAR images to extract strong metallic reflectors and generate a first SAR strong reflective target distribution map.

[0061] In one implementation, using the multiple candidate aeromagnetic anomaly regions as a reference, the multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflectivity target distribution maps are spatially overlaid, and multi-evidence fusion false anomaly determination is performed to generate the false anomaly mask region. Step A320 may further include:

[0062] A321: Based on the preset spatial resolution, the first candidate aeromagnetic anomaly region is rasterized to obtain multiple analysis units.

[0063] A322: Based on the multiple analysis units, multi-source evidence is extracted from the spatially superimposed first multispectral target distribution map, first hyperspectral metal material distribution map, and first SAR strong reflectivity target distribution map to obtain multiple sets of interference evidence.

[0064] A323: Based on the mapping of the multiple sets of interference evidence, interpret the false anomaly properties of the multiple analysis units and output multiple aeromagnetic anomaly judgment results.

[0065] A324: Based on the multiple aeromagnetic anomaly judgment results, separate the first false anomaly mask partition from the first candidate aeromagnetic anomaly region.

[0066] A325: By analogy, the false anomaly interpretation and separation of the multiple candidate aeromagnetic anomaly regions are performed to obtain multiple false anomaly mask partitions.

[0067] A325: Spatial fusion of the multiple pseudo-anomaly mask partitions to obtain the pseudo-anomaly mask region.

[0068] This embodiment first initiates multi-threaded collaborative interpretation on the acquired multi-source remote sensing dataset containing multi-spectral, hyperspectral, and SAR information. Each thread independently processes one data type. Specifically, the multispectral thread performs land cover classification and target identification, and outputs a spatial distribution map of labeled human targets; the hyperspectral thread focuses on material spectral matching to generate a distribution map of strongly magnetic metal materials; and the SAR thread extracts the location of metal reflectors through backscattering features.

[0069] Finally, three sets of spatial datasets are output simultaneously: multispectral target distribution map, hyperspectral metal material distribution map, and SAR strong reflectivity target distribution map, forming the basic evidence layer for subsequent fusion and identification.

[0070] Since the specific data processing process of performing multi-threaded collaborative interpretation on multiple multi-source remote sensing datasets to output multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflective target distribution maps is reusable, this embodiment takes the interpretation of the first multi-source remote sensing dataset of the first candidate aeromagnetic anomaly region to obtain the first multispectral target distribution map, the first hyperspectral metallic material distribution map, and the first SAR strong reflective target distribution map as an example to elaborate on the technical solution in detail.

[0071] Specifically, using a mature supervised classification algorithm, the high-resolution multispectral image is first processed for land cover classification, dividing the image into land cover categories such as vegetation, water bodies, and buildings to form a basic classification layer.

[0072] Based on this classification result, refined target identification is performed on the building category areas. Specifically, the factory outline is extracted through edge detection, and regular geometric features such as rectangular roofs and linear power transmission lines are combined to distinguish industrial facilities from natural landscapes. Then, texture analysis is used to filter out vegetation camouflage targets. Finally, a first multispectral target distribution map is generated, which marks the spatial location of human targets such as power transmission corridors and factory clusters. The first multispectral target distribution map is stored in the form of vector surface for subsequent multi-source fusion stage to verify the spatial correlation between aeromagnetic anomalies and human targets.

[0073] Radiometric correction and atmospheric compensation are performed on hyperspectral remote sensing data to obtain the true surface reflectance spectral curve. Then, spectral feature matching is performed on each pixel in the spectral range of 400-2500nm. Specifically, the matching involves calculating the spectral angular distance or information divergence of each pixel and comparing it with the standard spectral library of strongly magnetic metals such as iron and stainless steel. Finally, a set similarity threshold is used to screen out several pixels whose spectral matching degree exceeds the limit.

[0074] A binary mask grid is constructed based on the selected pixels, and a first hyperspectral metal material distribution map is output. The location of strongly magnetic metal materials is directly marked, providing material evidence for determining whether aeromagnetic anomalies are caused by metal facilities. In the binary mask grid, the matching pixel value is 1, and otherwise it is 0.

[0075] The raw SAR data (synthetic aperture radar data) is sequentially subjected to radiometric calibration to convert digital values ​​to backscattering coefficients, multi-look processing to suppress speckle noise, and geocoding to align with the geographic coordinate system, generating spatially accurate intensity images.

[0076] It should be understood that, due to the significant high backscattering characteristics of SAR for metallic targets, this embodiment performs high backscattering target detection on the image. Specifically, an adaptive threshold segmentation algorithm such as CFAR is used, combined with local window statistical features, to extract pixels with strong reflections significantly higher than the background. Here, strong reflection pixels are those caused by the strong backscattering characteristics of metallic targets and whose scattering intensity exceeds three times the standard deviation of the local background mean. Then, morphological closing operations are used to connect neighboring pixels to form a cluster of metallic targets. After contour tracing, polygonal vector boundaries are generated, and finally, a first SAR strong reflection target distribution map is output, which identifies the spatial distribution of strong metallic reflectors such as large machinery and rail facilities. The first SAR strong reflection target distribution map verifies the existence of human interference through metallic reflection characteristics and improves the multi-source evidence chain.

[0077] Based on this, a refined spatial subdivision is performed on each first candidate aeromagnetic anomaly region. According to a preset spatial resolution, such as a 10-meter grid, the vector boundary of the first candidate aeromagnetic anomaly region is converted into a regular raster array, with each grid serving as an independent analysis unit. This rasterization operation discretizes continuous anomaly regions into quantifiable statistical primitives, ensuring that multi-source evidence extraction is performed at a uniform scale, avoiding feature dilution caused by large-area statistics, and laying the foundation for unit-level fusion judgment.

[0078] Within the gridded analysis unit framework, multi-source interference evidence is extracted. For each analysis unit, the presence status of human targets, such as factory / transmission line coverage, is read from the overlaid first multispectral target distribution map. The presence markers of metal materials are obtained from the first hyperspectral metal material distribution map. The presence markers of strong reflective targets are extracted from the first SAR reflector distribution map. The output results are the evidence triples corresponding to each unit, resulting in multiple sets of interference evidence corresponding to the multiple analysis units to support the determination of unit-level anomaly properties.

[0079] This embodiment pre-defines multi-source evidence fusion judgment rules, specifically including the following process: First, multi-source evidence data is extracted from each analysis unit. Multispectral evidence is obtained by calculating the proportion of human-made targets covering the unit's total area, such as factories and power transmission lines. Hyperspectral evidence uses a binary existence marker: a value of 1 is recorded when a strongly magnetic metallic material with matching spectra exists within the unit, otherwise 0. SAR evidence is comprehensively judged by combining intensity and polarization characteristics: a value of 1 is recorded when a strongly reflective target within the unit simultaneously satisfies a backscattering coefficient σ° > -5dB and a cross-polarization ratio HV / VH > 2, otherwise 0. Based on this, the spatial overlap between the aeromagnetic anomaly area and the combined multi-source evidence area (human-made targets + metallic material + SAR reflector) is calculated. The overlap is the percentage of the overlapping area to the anomaly area. The final classification is as follows: if a cell simultaneously meets all four conditions—human target coverage ≥ 50%, presence of hyperspectral metallic material, presence of SAR strong reflective targets, and spatial overlap ≥ 70%—it is marked as a confirmed false anomaly; if any three of these conditions are met, it is marked as a suspected false anomaly; all other cases are marked as true anomalies. This rule comprehensively covers the judgment logic of spatial location matching, material property verification, and reflective feature coordination, ensuring that interference targets highly correlated with aeromagnetic anomalies in terms of spatial scale and physical properties are accurately identified.

[0080] The multi-source evidence fusion judgment rule is used to traverse the multiple sets of interfering evidence, interpret the false anomaly properties of the multiple analysis units, and output the judgment results of multiple aeromagnetic anomalies at the unit level.

[0081] Based on the aforementioned multiple aeromagnetic anomaly assessment results, the false anomaly partitions are reconstructed. Specifically, analysis units marked as confirmed false anomalies and suspected anomalies are aggregated into connected regions. Fragmentation is eliminated and jagged boundaries are smoothed through polygon buffer fusion and simplification, generating a first false anomaly mask partition corresponding to the first candidate anomaly region. This first false anomaly mask partition retains the original anomaly boundaries but internally marks areas that need to be removed, completing the transformation from discrete units to complete spatial objects and providing a precise target area for subsequent reconstruction.

[0082] By analogy, multi-threaded collaborative interpretation processing is performed on the multiple multi-source remote sensing datasets to output multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflective target distribution maps. Based on the multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflective target distribution maps, steps A321-A324 are executed cyclically to separate false anomalies in the multiple candidate aeromagnetic anomaly regions, resulting in multiple false anomaly mask partitions.

[0083] Finally, through spatial fusion operations such as vector merging or raster maximum value synthesis, the multiple false anomaly mask partitions are integrated to form a globally continuous false anomaly mask region, which serves as the input for aeromagnetic data reconstruction, ensuring that no interfering targets are missed.

[0084] Furthermore, a polygon simplification algorithm is used to eliminate redundant nodes, and irregular protrusions are smoothed by Bézier curves, so that the outline of the false anomaly mask area conforms to the characteristics of natural geological boundaries. The final output standard false anomaly mask area can directly guide the repair of aeromagnetic data and the restoration of background fields, ensuring the reliability of subsequent geological interpretation.

[0085] A400: Based on the differences in the scale and geological background of the anomalous area, the pseudo-anomaly mask area is divided into multi-scale hierarchical divisions to obtain small-scale regular blank areas, medium-scale continuous blank areas, and large-scale structurally complex blank areas.

[0086] Based on the differences in spatial scale characteristics within the pseudo-anomaly mask area and the complexity of the surrounding geological background, this embodiment performs multi-level zoning operations. First, it identifies blank areas with small areas and regular boundary geometry, such as rectangular factory buildings or power transmission tower bases, and classifies them as small-scale regular blank areas. Second, it divides medium-sized blank areas with continuous spatial distribution but relatively simple geological structures, such as linear transportation corridors. Finally, it defines wide-area blank areas covering large industrial clusters or located in complex tectonic backgrounds such as fault zones, forming large-scale tectonically complex blank areas.

[0087] The multi-scale hierarchical partitioning in this embodiment enables fine classification of abnormal regions, providing targeted input for subsequent hierarchical reconstruction strategies and ensuring that blank areas with different features are matched with appropriate reconstruction algorithms.

[0088] A500: Iteratively perform multi-scale collaborative constraint reconstruction on the small-scale regular blank areas, medium-scale continuous blank areas, and large-scale structurally complex blank areas to output purified aeromagnetic data.

[0089] In one implementation, multi-scale collaborative constraint reconstruction is iteratively performed on the small-scale regular blank areas, the medium-scale continuous blank areas, and the large-scale structurally complex blank areas to output the purified aeromagnetic data. Step A500 may further include:

[0090] A510: After clustering and grouping the small-scale regular blank areas, directional interpolation reconstruction is performed in parallel to obtain primary reconstructed magnetic field data.

[0091] A520: After incorporating the primary reconstructed magnetic field data as constraint data into the mesoscale continuous blank area, adaptive interpolation repair of the mesoscale continuous blank area is performed to obtain the intermediate reconstructed magnetic field data.

[0092] A530: After integrating the intermediate-level reconstructed magnetic field data as known data into the large-scale structural complex blank area, conditional constraint reconstruction is performed, and advanced reconstructed magnetic field data is output.

[0093] A540: Perform layer-by-layer iterative verification and fusion on the primary reconstructed magnetic field data, intermediate reconstructed magnetic field data, and advanced reconstructed magnetic field data, and output the purified aeromagnetic data.

[0094] In one implementation, after clustering the small-scale regular blank areas, directional interpolation reconstruction is performed in parallel to obtain primary reconstructed magnetic field data. Step A510 may further include:

[0095] A511: Cluster the small-scale regular blank areas based on spatial proximity to form multiple blank sub-region processing groups.

[0096] A512: Based on the outer boundaries of the multiple blank sub-region processing groups, the real observation data is segmented in the multi-temporal aeromagnetic observation data to obtain multiple constraint boundary condition groups.

[0097] A513: Based on the aeromagnetic survey line direction of the multi-temporal aeromagnetic observation data and the multiple regional geological structure orientation groups of the multiple blank sub-region processing groups, define multiple dominant reconstruction direction groups.

[0098] A514: Using the multiple sets of constraint boundary conditions as interpolation to reconstruct boundary constraints, and performing spatial interpolation reconstruction of directional constraints in parallel on the multiple blank sub-region processing groups according to the multiple dominant reconstruction direction groups, to obtain multiple sets of local reconstructed magnetic field data.

[0099] A515: Seamlessly overlay the multiple local reconstructed magnetic field data sets onto the multi-temporal aeromagnetic observation data to obtain the primary reconstructed magnetic field data.

[0100] In one implementation, after incorporating the primary reconstructed magnetic field data as constraint data into the mesoscale continuous blank area, adaptive interpolation repair of the mesoscale continuous blank area is performed to obtain the intermediate reconstructed magnetic field data. Step A520 may further include:

[0101] A521: Project the primary reconstructed magnetic field data onto the mesoscale continuous blank area, extract spatially adjacent data, and obtain a locally constrained dataset.

[0102] A522: Perform trend analysis on the local constrained dataset to extract an adaptive trend surface.

[0103] A523: Using the adaptive trend surface as a baseline constraint, perform parameter-adaptive spatial interpolation repair on the mesoscale continuous blank area to obtain repaired magnetic field data.

[0104] A524: Integrate the repaired magnetic field data into the primary reconstructed magnetic field data to obtain the intermediate reconstructed magnetic field data.

[0105] In one implementation, after integrating the intermediate-level reconstructed magnetic field data as known data into the large-scale structurally complex blank area, conditional constraint reconstruction is performed to output high-level reconstructed magnetic field data. Step A530 may further include:

[0106] A531: Extract the peripheral real observation data of the large-scale structurally complex blank area from the multi-temporal aeromagnetic observation data.

[0107] A532: Integrate the aforementioned real observation data from the periphery and the intermediate reconstructed magnetic field data to construct a known data constraint body.

[0108] A533: By performing frequency domain filtering on the known data constraint, the background magnetic field components in the long-wavelength region are extracted as the main control trend surface.

[0109] A534: Introduce regional geological distribution information and gravity anomaly data to construct geological model constraints and geophysical co-constraints respectively.

[0110] A535: Using the main control trend surface as a framework, and incorporating the geological model constraints and geophysical collaborative constraints, iterative reconstruction optimization under conditional constraints is performed on the large-scale tectonic complex blank area, and conditional constraint reconstructed magnetic field data is output.

[0111] A536: Integrate the conditionally constrained reconstructed magnetic field data into the intermediate reconstructed magnetic field data, and output the advanced reconstructed magnetic field data.

[0112] Specifically, based on a preset spatial proximity threshold, such as a 50-meter Euclidean distance, spatial clustering analysis is performed on scattered small-scale regular blank areas. Independent blank areas that are adjacent in location or have a distance less than the threshold, such as multiple factory foundations, are aggregated into continuous spatial clusters, forming multiple blank sub-region processing groups. Each blank sub-region processing group represents a geographically continuous blank area unit, reducing the number of subsequent reconstruction tasks and ensuring the spatial continuity of the reconstruction field within the group, thus providing optimized input for parallel reconstruction.

[0113] For each blank sub-region processing group, its outer boundary buffer is extracted, such as the original measurement values ​​in the multi-temporal aeromagnetic observation data within a range of 100 meters. The real magnetic field point set corresponding to each sub-region is obtained through geospatial segmentation operation. These point sets constitute multiple constraint boundary condition sets, which serve as fixed boundary control points for interpolation reconstruction. This ensures that the reconstructed magnetic field and the surrounding measured field achieve a smooth transition in amplitude and gradient at the boundary, avoiding reconstruction distortion.

[0114] Based on the flight trajectory direction of the original aeromagnetic survey lines and the geological structural orientation of the blank sub-regions during the acquisition of multi-temporal aeromagnetic observation data, a dominant reconstruction direction group is defined for each blank sub-region processing group. Specifically, an anisotropic interpolation guiding vector is generated through direction weight allocation, so that the reconstructed field extends along the natural variation trend of the magnetic field. For example, the flight trajectory direction of the original aeromagnetic survey lines is the main survey line running due north, and the geological structural orientation of the blank sub-regions is the azimuth of the fault zone extension. The weight allocation is 0.7 for the survey line direction and 0.3 for the structural orientation.

[0115] Using the measured magnetic field point set provided by the constraint boundary condition set as fixed interpolation anchor points, directional spatial interpolation reconstruction is performed independently for each blank sub-region processing group. Specifically, based on the preset dominant reconstruction direction group parameters, the spatial correlation model is dynamically adjusted using the anisotropic kriging algorithm. Larger range parameters are set along the direction of the aeromagnetic survey line and the geological structure, so that the reconstructed field exhibits a gentle and gradual change in that direction. Smaller range parameters are set perpendicular to the dominant direction to retain reasonable magnetic field gradient changes.

[0116] By synchronously processing the multiple blank sub-region processing groups through a parallel computing architecture, the reconstruction efficiency of massive small-scale blank regions is improved, and multiple local reconstruction magnetic field data groups that precisely match the geometry of the multiple blank sub-region processing groups are generated. This achieves the physical consistency restoration of the background magnetic field within small-scale regular regions, providing basic field data for subsequent reconstruction stages.

[0117] Based on the spatial location of the multiple blank sub-region processing groups, the multiple local reconstructed magnetic field data groups are precisely covered by the corresponding blank sub-regions according to geographical coordinates, directly replacing the null or interference marker values ​​in the original data. Then, the seams are eliminated by weighted smoothing of the boundary measurement points to ensure a continuous transition of magnetic field strength and gradient between the reconstructed area and the measured area. Finally, the integrated primary reconstructed magnetic field data is output. The primary reconstructed magnetic field data fully retains the original data format, coordinate reference and unreconstructed area values, while achieving seamless filling of small-scale regular blank areas to form the pre-constrained background field required for the mesoscale reconstruction stage.

[0118] The primary reconstructed magnetic field dataset, containing the small-scale blank area repair values ​​obtained in step A510, is spatially superimposed with the mesoscale continuous blank area. The outer buffer zone of the blank area, such as all magnetic field measurement points within a 200-meter radius, including both original observation points and reconstructed points, is then extracted. A local constraint dataset covering the blank area boundary is constructed using geospatial queries. This local constraint dataset serves as the boundary anchor point for interpolation repair, ensuring amplitude continuity and gradient continuity between the reconstructed field and the surrounding magnetic field in the transition zone.

[0119] Moving window trend analysis was performed on the local constrained dataset. A second-order polynomial surface fitting algorithm was used to calculate the background magnetic field variation trend window by window. The results of each window were then fused by weighted averaging to generate a global adaptive trend surface. The adaptive trend surface represents the spatial distribution law of the low-frequency background field in the mesoscale continuous blank area and is used as the baseline constraint framework for interpolation repair, guiding the reconstructed field to conform to the macroscopic morphology of the regional magnetic field.

[0120] Using the adaptive trend surface as the benchmark framework for background magnetic field changes, the interpolation parameters are dynamically configured based on the local magnetic field gradient characteristics at different locations within the mesoscale continuous blank area. Specifically, the spline tension coefficient is increased to 0.7 in the high gradient transition zone, such as the anomalous boundary region, to suppress oscillation distortion, while the tension coefficient is reduced to 0.3 in the low gradient stable region to preserve subtle change characteristics.

[0121] A directionally constrained variable tension spline algorithm is adopted, and the main interpolation axis is set along the direction of the aeromagnetic survey line to make the reconstructed field conform to the magnetic field variation law of the flight trajectory. At the same time, the normal constraint force is adjusted with reference to the geological structure to avoid unreasonable field value jumps across fault zones.

[0122] By iteratively solving the magnetic field distribution that satisfies the trend surface framework and boundary continuity conditions, spatially coordinated repair magnetic field data is generated, realizing the natural transition and restoration of the background field in the blank area, and ensuring that the reconstruction results have a continuous transition in amplitude, gradient and spatial trend.

[0123] The repaired magnetic field data is seamlessly integrated into the primary reconstructed magnetic field data according to spatial coordinates. Specifically, the original null values ​​in the mesoscale blank areas are first replaced, and then bilinear interpolation is performed to smooth the transition in overlapping areas such as the boundary between small and mesoscale areas to eliminate data seams and ensure the continuity of magnetic field gradients. The final output of the intermediate reconstructed magnetic field data completely retains the primary reconstruction results and adds repaired values ​​for the mesoscale blank areas, forming a continuous magnetic field dataset covering small and medium-scale regions, providing a high-precision constraint field basis for the reconstruction of large-scale complex areas.

[0124] The original measured magnetic field values ​​of the outer region of the large-scale structurally complex blank area, such as the 500-meter buffer zone, are accurately extracted from multi-temporal aeromagnetic observation data as the true observation data of the surrounding area. The extraction operation here prioritizes spatial topology analysis technology, strictly screens the original measurement point data located on the periphery of the blank area and not affected by reconstruction, and excludes any data points in the small and medium-scale areas that have been repaired, to ensure that the constraint point set is a true and undisturbed observation value.

[0125] The obtained peripheral real observation data constitute the boundary constraint set for reconstruction. Its spatial distribution density needs to be more than 1.2 times that of the original data to ensure that the reconstructed field and the real geological background are seamlessly connected at the kilometer scale and to avoid boundary distortion problems caused by constraint data contamination.

[0126] The extracted peripheral real observation data and intermediate reconstructed magnetic field data are integrated at multiple levels. Specifically, firstly, sub-meter spatial registration is achieved through coordinate transformation; secondly, adaptive gridding is used to unify discrete measurement points into a standard grid system; and finally, a known data constraint body covering a range of 1-2 kilometers outside the blank area is constructed.

[0127] The known data constraint volume comprises three core data layers: original aeromagnetic observations providing basic boundary constraints, small-scale reconstructed values ​​contributing local details, and mesoscale reconstructed values ​​providing transition zone information, forming a high-confidence constraint field with a spatial density 1.8 times that of the original data. The value of the known data constraint volume lies in establishing a boundary condition system with continuously varying gradients, providing a solid mathematical and physical foundation for large-scale reconstruction.

[0128] Frequency domain filtering is performed on the integrated known data constraint volume. A two-dimensional Fourier transform is used to convert the spatial domain magnetic field data into a wavenumber domain representation. In the wavenumber domain, a low-pass filter is used to separate the regional background field from the local anomaly field. A cutoff frequency is set based on the regional tectonic scale characteristics, and low-frequency magnetic field components controlled by deep geological structures are extracted from the regional background field. An inverse Fourier transform is performed on these low-frequency magnetic field components to generate a master trend surface. This master trend surface reflects the gradually varying magnetic field background caused by macroscopic geological factors such as basement undulations. Its spatial variation gradient is consistent with the regional geological structural trend and serves as the benchmark framework field for subsequent reconstruction, ensuring that the reconstruction results conform to deep geological patterns.

[0129] Regional geological maps were introduced to analyze lithological distribution characteristics. Geological model constraints were constructed based on typical magnetic parameters of different rock types to delineate magnetic body boundaries and set magnetic susceptibility constraints. Regional gravity anomaly data were simultaneously loaded, and geophysical co-equations were established as geophysical co-constraints by analyzing the correlation between gravity and magnetic fields within tectonic units.

[0130] The two constraints together constitute a physical property constraint system. Specifically, the geological model constraint limits the spatial distribution of magnetic bodies, while the geophysical co-constraint provides the correlation rules between rock density and magnetization intensity, forming a joint constraint framework that suppresses and reconstructs multiple solutions.

[0131] Using the dominant trend surface as the benchmark framework for background magnetic field changes, and incorporating the magnetic body boundary conditions provided by the geological model constraints and the gravity and magnetic correlation established by the geophysical co-constraints, iterative reconstruction optimization is performed on the large-scale tectonically complex blank areas.

[0132] In the iterative reconstruction and optimization process, the main trend surface value is first assigned to the blank area as the initial field. Then, in each iteration, the degree of matching between the current reconstructed magnetic field and the geological boundary conditions is evaluated, and the spatial correlation between the reconstructed magnetic field and the gravity field is verified to conform to physical laws.

[0133] By dynamically adjusting the magnetic field distribution, the multi-objective function gradually converges, and finally outputs reconstructed magnetic field data that meets the conditions of regional background field trend consistency, gradient continuity across geological boundaries, and correlation between gravity and magnetic properties.

[0134] The magnetic field data generated by conditional constraint reconstruction is spatially integrated with the intermediate-scale reconstructed magnetic field data. Specifically, firstly, the original null data in the large-scale blank area is replaced; secondly, a smooth interpolation algorithm is used to fuse the data in the transition zone between the small-scale and medium-scale reconstruction areas and the large-scale reconstruction areas; and finally, the local distortion area is optimized by verifying the continuity of the magnetic field gradient.

[0135] The final output of advanced reconstructed magnetic field data forms a continuous background field covering the entire region, completely eliminating the influence of human interference, restoring the magnetic field characteristics controlled by deep geological structures, and providing a data foundation that conforms to standards for subsequent geological interpretation.

[0136] After completing the hierarchical reconstruction of primary, intermediate, and advanced magnetic field data, a layer-by-layer iterative verification and fusion process is performed. Specifically, the three-layer reconstruction results are first superimposed according to spatial coordinates. Systematic verification is then carried out in the transition areas between different reconstruction levels. In the overlapping areas of primary and intermediate reconstruction data, such as the boundary between factory foundation clusters and strip-shaped blank areas, the continuity of the magnetic field gradient is verified. In the overlapping areas of intermediate and advanced reconstruction data, such as the extension area of ​​the fault zone, the consistency of the spatial trend is verified. In complex structural areas where the three layers of data overlap, such as the area covered by industrial clusters, the coordination of the overall background field morphology is checked.

[0137] Through multiple rounds of iterative optimization, the reconstructed field values ​​of each layer are dynamically adjusted to meet three fusion criteria: first, to ensure that the difference in magnetic field intensity between adjacent reconstructed layers does not exceed the natural fluctuation range of the regional background field; second, to ensure that the rate of change of magnetic field gradient in the cross-layer transition zone conforms to the geological structural law; and third, to strictly preserve the spatial structural characteristics of the real geological anomalies.

[0138] The final output is seamless purified aeromagnetic data with a continuous background field and reasonable geological significance, completely eliminating the influence of human interference, and providing a high-quality data foundation for structural interpretation and mineral exploration.

[0139] This embodiment achieves comprehensive purification and restoration of aeromagnetic data through multi-source remote sensing collaborative identification and hierarchical reconstruction technology. It completely eliminates the interference of false aeromagnetic anomalies caused by human activities such as power transmission and transformation facilities and industrial clusters, and obtains reliable aeromagnetic data that maintains natural continuity in amplitude, gradient and spatial distribution, and fully preserves the true magnetic anomaly characteristics generated by deep geological structures. This provides a high-quality data foundation for geophysical exploration.

[0140] The above description is merely a preferred 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 scope of the technology 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.

[0141] The foregoing description of specific exemplary embodiments of the invention is for illustrative and explanatory purposes. These descriptions are not intended to limit the invention to the precise forms disclosed, and it will be apparent that many changes and variations can be made in accordance with the foregoing teachings. The exemplary embodiments were chosen and described in order to explain the specific principles of the invention and its practical application, thereby enabling those skilled in the art to implement and utilize various different exemplary embodiments of the invention, as well as various different choices and variations. The scope of the invention is intended to be defined by the claims and their equivalents.

Claims

1. A method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction, characterized in that, The method includes: Interference anomalies were captured from multi-temporal aeromagnetic observation data to locate multiple candidate aeromagnetic anomaly regions; Based on the spatial boundaries of the multiple candidate aeromagnetic anomaly regions, multimodal remote sensing satellite data is acquired to obtain multiple multi-source remote sensing datasets; Based on the multiple multi-source remote sensing datasets, the multiple candidate aeromagnetic anomaly regions are subjected to multi-source collaborative identification of false anomalies to generate a false anomaly mask region. Based on the differences in scale and geological background of the anomalous area, the pseudo-anomaly mask area is divided into multi-scale hierarchical divisions to obtain small-scale regular blank areas, medium-scale continuous blank areas, and large-scale structurally complex blank areas. Multi-scale collaborative constraint reconstruction is iteratively performed on the small-scale regular blank areas, medium-scale continuous blank areas, and large-scale structurally complex blank areas to output purified aeromagnetic data.

2. The method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction as described in claim 1, characterized in that, Iteratively performing multi-scale collaborative constraint reconstruction on the small-scale regular blank areas, medium-scale continuous blank areas, and large-scale structurally complex blank areas, the output of the purified aeromagnetic data includes: After clustering and grouping the small-scale regular blank areas, directional interpolation reconstruction is performed in parallel to obtain primary reconstructed magnetic field data; After incorporating the primary reconstructed magnetic field data as constraint data into the mesoscale continuous blank area, adaptive interpolation repair of the mesoscale continuous blank area is performed to obtain the intermediate reconstructed magnetic field data. After integrating the intermediate-level reconstructed magnetic field data as known data into the large-scale structural complex blank area, conditional constraint reconstruction is performed to output advanced reconstructed magnetic field data. The primary reconstructed magnetic field data, intermediate reconstructed magnetic field data, and advanced reconstructed magnetic field data are subjected to layer-by-layer iterative verification and fusion to output the purified aeromagnetic data.

3. The method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction as described in claim 1, characterized in that, Based on the multiple multi-source remote sensing datasets, false anomaly multi-source collaborative identification is performed on the multiple candidate aeromagnetic anomaly regions to generate false anomaly mask regions, including: Multi-threaded collaborative interpretation processing is performed on the multiple multi-source remote sensing datasets to output multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflectance target distribution maps. Using the multiple candidate aeromagnetic anomaly regions as reference benchmarks, multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflective target distribution maps are spatially overlaid, and multi-evidence fusion false anomaly determination is performed to generate the false anomaly mask region.

4. The method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction as described in claim 3, characterized in that, Using the multiple candidate aeromagnetic anomaly regions as reference benchmarks, the multiple multispectral target distribution maps, multiple hyperspectral metallic material distribution maps, and multiple SAR strong reflectivity target distribution maps are spatially overlaid. Multi-evidence fusion is then performed to determine false anomalies, generating the false anomaly mask region, including: Based on the preset spatial resolution, the first candidate aeromagnetic anomaly region is rasterized to obtain multiple analysis units; Based on the multiple analysis units, multi-source evidence is extracted from the spatially superimposed first multispectral target distribution map, first hyperspectral metallic material distribution map, and first SAR strong reflective target distribution map to obtain multiple sets of interference evidence. Based on the mapping of the multiple sets of interference evidence, the false anomaly properties of the multiple analysis units are interpreted, and multiple aeromagnetic anomaly judgment results are output. Based on the multiple aeromagnetic anomaly judgment results, the first false anomaly mask partition is separated from the first candidate aeromagnetic anomaly region; By analogy, the false anomaly interpretation and separation of the multiple candidate aeromagnetic anomaly regions are performed to obtain multiple false anomaly mask partitions; The multiple pseudo-anomaly mask partitions are spatially fused to obtain the pseudo-anomaly mask region.

5. The method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction as described in claim 2, characterized in that, After clustering and grouping the small-scale regular blank areas, directional interpolation reconstruction is performed in parallel to obtain primary reconstructed magnetic field data, including: The small-scale regular blank areas are clustered based on spatial proximity to form multiple blank sub-region processing groups; Based on the outer boundaries of the multiple blank sub-region processing groups, the real observation data is segmented in the multi-temporal aeromagnetic observation data to obtain multiple sets of constraint boundary conditions. Based on the aeromagnetic survey line direction of the multi-temporal aeromagnetic observation data and the multiple regional geological structure orientation groups of the multiple blank sub-region processing groups, multiple dominant reconstruction direction groups are defined; The boundary constraints are reconstructed by interpolation using the multiple sets of constraint boundary conditions. Spatial interpolation reconstruction of directional constraints is performed in parallel on the multiple blank sub-region processing groups according to the multiple dominant reconstruction direction groups to obtain multiple sets of local reconstructed magnetic field data. The multiple sets of locally reconstructed magnetic field data are seamlessly overlaid onto the multi-temporal aeromagnetic observation data to obtain the primary reconstructed magnetic field data.

6. The method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction as described in claim 2, characterized in that, After incorporating the primary reconstructed magnetic field data as constraint data into the mesoscale continuous blank area, adaptive interpolation repair of the mesoscale continuous blank area is performed to obtain the intermediate reconstructed magnetic field data, including: The primary reconstructed magnetic field data is projected onto the mesoscale continuous blank area, and spatially adjacent data is extracted to obtain a locally constrained dataset. Perform trend analysis on the local constrained dataset to extract an adaptive trend surface; Using the adaptive trend surface as a baseline constraint, the mesoscale continuous blank area is subjected to parameter-adaptive spatial interpolation repair to obtain repaired magnetic field data; The repaired magnetic field data is integrated into the primary reconstructed magnetic field data to obtain the intermediate reconstructed magnetic field data.

7. The method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction as described in claim 2, characterized in that, After integrating the intermediate-level reconstructed magnetic field data as known data into the large-scale structurally complex blank area, conditional constraint reconstruction is performed to output high-level reconstructed magnetic field data, including: Extract the peripheral real observation data of the large-scale structurally complex blank area from the multi-temporal aeromagnetic observation data; By integrating the aforementioned real observation data from the periphery and the intermediate-level reconstructed magnetic field data, a known data constraint body is constructed; Frequency domain filtering is performed on the known data constraint to extract the background magnetic field components in the long-wave region as the main control trend surface; Regional geological distribution information and gravity anomaly data are introduced to construct geological model constraints and geophysical co-constraints, respectively; Using the main control trend surface as a framework, and incorporating the geological model constraints and geophysical co-constraints, iterative reconstruction optimization under conditional constraints is performed on the large-scale tectonically complex blank area, and conditional constraint reconstructed magnetic field data is output. The conditionally constrained reconstructed magnetic field data is integrated into the intermediate reconstructed magnetic field data, and the advanced reconstructed magnetic field data is output.

8. The method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction as described in claim 1, characterized in that, Interference anomaly capture was performed on multi-temporal aeromagnetic observation data to locate multiple candidate aeromagnetic anomaly regions, including: After acquiring the multi-temporal raw observation data of the target survey area, gridding correction processing is performed to obtain the multi-temporal aeromagnetic observation data; Based on the simulation of the regional background magnetic field, the regional background magnetic field is separated from the multi-temporal aeromagnetic observation data to obtain local magnetic anomaly data; Based on the local spatial variation characteristics of the local magnetic anomaly data, an anomaly detection threshold is set. Based on the anomaly capture threshold, significant anomaly regions are extracted from the local magnetic anomaly data using statistical threshold segmentation to obtain local anomaly regions; After spatially aggregating the local anomalous regions, the multiple candidate aeromagnetic anomaly regions are formed by performing boundary regularization.

9. The method for multi-source remote sensing collaborative aeromagnetic false anomaly removal and hierarchical constraint reconstruction as described in claim 3, characterized in that, High-resolution multispectral imagery, hyperspectral remote sensing data, and SAR data are extracted from the first multi-source remote sensing dataset, and parallel collaborative interpretation is performed. S1: After performing land cover classification processing on the high-resolution multispectral image, human target identification and extraction are performed based on the classification results to generate a first multispectral target distribution map; S2: By performing surface material spectral feature matching on the hyperspectral remote sensing data, strongly magnetic metal materials are identified, and a first hyperspectral metal material distribution map is output. S3: Perform radiometric calibration, multi-view processing and geocoding preprocessing on the SAR data to obtain SAR images. Then, perform high backscattering target detection on the SAR images to extract strong metallic reflectors and generate a first SAR strong reflective target distribution map.

Images