Method and System for Constructing Multi-Attribute Spatial Feature Block Model of Non-ferrous Metal Mines

Abstract

This application provides a method and system for constructing a multi-attribute spatial feature block model of non-ferrous metal deposits, relating to the field of geological modeling technology. This application acquires remote sensing images, first coordinate information, contour morphology information, element content data and second coordinate information of surface sampling points for non-ferrous metal deposits; determines the surface cover type of different ore body regions in the remote sensing images; and obtains texture information sets and spectral information sets based on this type; calibrates the first and second coordinate information to construct an ore body framework; forms a multi-attribute dataset; forms a multi-attribute feature set based on the ore body framework; and employs a multi-scale attention model based on geological knowledge constraints to construct a spatial block model of non-ferrous metal deposits. This achieves high-precision, multi-attribute collaborative automatic construction of a three-dimensional spatial block model of non-ferrous metal deposits, improving the accuracy of ore body boundary delineation and the continuity of attribute distribution.

Description

Technical Field

[0001] This application relates to the field of geological modeling technology, and in particular to a method and system for constructing multi-attribute spatial feature block models of non-ferrous metal ores. Background Technology

[0002] During the non-ferrous metal ore prospecting phase, there is an urgent need for a modeling method that can efficiently integrate multi-source surface observation data with shallow spatial geometric information to support subsequent resource potential assessment and exploration deployment.

[0003] To address the aforementioned needs, existing mainstream solutions have attempted to combine hyperspectral remote sensing and ground geochemical sampling data, perform spatial overlay analysis through a geographic information system platform, and use classical geostatistical methods such as Kriging or inverse distance weighting to spatially interpolate elemental content, thereby generating a preliminary three-dimensional attribute block model.

[0004] However, existing solutions have obvious limitations in practical applications. Their utilization of remote sensing information is limited to the level of surface feature extraction, and in the process of ore body geometric modeling, they mainly rely on manual drawing or simplified geometric assumptions, resulting in a large deviation between the block framework and the actual spatial structure of the ore body. Summary of the Invention

[0005] The purpose of this application is to provide a method and system for constructing multi-attribute spatial feature block models of non-ferrous metal ores, in order to solve the problem that existing technologies lack the ability to explore the deep nonlinear correlation between spectrum, texture and element content.

[0006] Firstly, this application provides a method for constructing a multi-attribute spatial feature block model of non-ferrous metal ores, including:

[0007] Acquire remote sensing images, first coordinate information, outline morphology information of non-ferrous metal mines, as well as element content data and second coordinate information of surface sampling points;

[0008] The surface cover type of different ore body areas in the remote sensing image is determined. Based on the surface cover type of different ore body areas, texture analysis and spectral analysis are performed on the ore body areas in the remote sensing image to obtain texture information set and spectral information set.

[0009] The first coordinate information and the second coordinate information are calibrated to obtain calibrated first coordinate information and calibrated second coordinate information. Based on the calibrated first coordinate information and contour morphology information, the ore body framework is constructed.

[0010] The spectral information set and texture information set are matched with the first coordinate information after calibration, and the element content data is matched with the second coordinate information after calibration to form a multi-attribute dataset.

[0011] Based on the ore body framework, multi-source information association processing is performed on multi-attribute datasets to form multi-attribute feature sets;

[0012] A multi-scale attention model based on geological knowledge constraints is used to complete the discrete attribute data in the multi-attribute feature set in order to construct a spatial block model of non-ferrous metal mines.

[0013] Optionally, the surface cover type of different ore body areas in the remote sensing image is determined. Based on the surface cover type of different ore body areas, texture analysis and spectral analysis are performed on the ore body areas in the remote sensing image to obtain texture information sets and spectral information sets, including:

[0014] Based on the geological boundary data of non-ferrous metal mines and the pixel grayscale values ​​of remote sensing images, multiple ore body regions were identified.

[0015] Based on the remote sensing reflectance, physical structure parameters and color characteristic parameters of each ore body area, the land cover type of each ore body area is determined.

[0016] Select feature dimensions and adaptation parameter values ​​that match the land cover type from the preset feature dimension library to determine the texture extraction rules for the land cover type.

[0017] Based on the texture extraction rules for different surface cover types, texture information sets are extracted from remote sensing images;

[0018] Based on the mineral spectral properties of each land cover type, target wavelength ranges are extracted from remote sensing images as target spectral bands. The target spectral bands of all land cover types are then integrated to form a spectral information set.

[0019] Optionally, feature dimensions and adaptation parameter values ​​matching the land cover type are selected from a preset feature dimension library to determine the texture extraction rules for the land cover type, including:

[0020] Select the feature dimension and matching parameter value that are suitable for each land cover type from the preset feature dimension library. When the land cover type is oxidized zone, the matching feature dimensions are roughness dimension and continuity dimension. When the land cover type is primary zone, the matching feature dimension is roughness dimension. When the land cover type is weathered rock layer, the matching feature dimensions are continuity dimension and directionality dimension.

[0021] Based on the characteristic dimensions of each land cover type, a texture analysis method for each land cover type is determined. The texture analysis method for the roughness dimension is to statistically analyze the gray value difference within a pixel block of a certain size; the texture analysis method for the continuity dimension is to statistically analyze the percentage of pixels that meet the continuity length threshold; and the texture analysis method for the directionality dimension is to statistically analyze the percentage of texture directions that meet the angular range.

[0022] The feature dimensions, adaptation parameter values, and texture analysis methods of each land cover type are integrated to form texture extraction rules for each land cover type.

[0023] Optionally, the first coordinate information and the second coordinate information are calibrated to obtain calibrated first coordinate information and calibrated second coordinate information. Based on the calibrated first coordinate information and contour morphology information, a orebody framework is constructed, including:

[0024] Multiple evenly distributed and immovable natural outcrops or artificial markers were selected as measurement marks within the exploration area of ​​the non-ferrous metal mine, and the standard coordinate data of each measurement mark were obtained.

[0025] Calculate the first deviation value between the first coordinate at each measurement mark in the first coordinate information and the corresponding standard coordinate data, and calibrate the first coordinate information according to the first deviation value to obtain the calibrated first coordinate information;

[0026] Calculate the second deviation value between the second coordinate at each measurement mark and the corresponding standard coordinate data in the second coordinate information, and calibrate the second coordinate information according to the second deviation value to obtain the calibrated second coordinate information;

[0027] Based on the calibrated first coordinate information, the exploration area is divided into multiple grid cells, and the grid coordinates of each grid cell are obtained;

[0028] The terrain undulation data and ore body boundary data in the contour morphology information are associated and bound with the corresponding grid cells to obtain multiple target grid cells;

[0029] Based on the grid coordinates, all target grid cells are arranged in three-dimensional space to obtain the ore body framework.

[0030] Optionally, the spectral information set and texture information set are matched with the calibrated first coordinate information, and the elemental content data is matched with the calibrated second coordinate information to form a multi-attribute dataset, including:

[0031] The first image range of each target spectral band in the spectral information set is obtained in the remote sensing image, and all coordinate data in the first image range of the calibrated first coordinate information are matched with the corresponding target spectral band to obtain multiple spectral data with coordinates.

[0032] The second image range of each texture data group in the texture information set is obtained in the remote sensing image, and all coordinate data in the second image range of the calibrated first coordinate information are matched with the corresponding texture data group to obtain multiple texture data with coordinates.

[0033] Obtain the sampling point positions of the element content data, and match the coordinate data corresponding to the sampling point positions in the calibrated second coordinate information with the element content data to obtain the element content data with coordinates;

[0034] The coordinate-based spectral data, coordinate-based texture data, and coordinate-based element content data are integrated to form a multi-attribute dataset.

[0035] Optionally, based on the ore body framework, multi-source information association processing is performed on the multi-attribute dataset to form a multi-attribute feature set, including:

[0036] Based on the ore body framework, multi-attribute data are integrated into coordinate-bound spectral data, coordinate-bound texture data, and coordinate-bound element content data within the target coordinate range of each target grid cell, forming a related data group for each target grid cell.

[0037] Based on a preset spectral range, the coordinate-bearing spectral data in each associated data group are categorized to determine the target spectral range for each associated data group.

[0038] Based on the preset texture structure type, the texture data with coordinates in each associated data group are classified to determine the target structure type of each associated data group;

[0039] Based on the preset element content level, the coordinate element content data in each associated data group are classified to determine the target content level of each associated data group.

[0040] The target spectral range, target structure type, and target content level corresponding to all target grid units are combined to form a multi-attribute feature set.

[0041] Optionally, a multi-scale attention model based on geological knowledge constraints is used to complete the discrete attribute data in the multi-attribute feature set, in order to construct a spatial block model of non-ferrous metal deposits, including:

[0042] Determine whether there is missing data in each multi-attribute feature group in the multi-attribute feature set. If there is missing data, the multi-attribute feature group is taken as the target feature group. If there is no missing data, the multi-attribute feature group is taken as the complete feature group.

[0043] Based on the spatial distribution scale of non-ferrous metal ore bodies, the target feature group is divided by scale to obtain the first feature subset, the second feature subset, and the third feature subset.

[0044] A multi-scale attention model constrained by geological knowledge is used to process the first feature subset, the second feature subset, and the third feature subset respectively to obtain the associated feature regions corresponding to the target feature group;

[0045] Attribute data that are not missing and meet the geological constraints in the associated feature region are taken as valid attribute data. Based on the valid attribute data, the complete data of discrete attribute data is calculated and then added to the target feature group to obtain the complete feature group.

[0046] All complete feature groups and corresponding target grid cells in the ore body framework are associated and bound to form a spatial block model of non-ferrous metal ore.

[0047] Secondly, this application provides a system for constructing multi-attribute spatial feature block models of non-ferrous metal ores, including:

[0048] The acquisition module is used to acquire remote sensing images of non-ferrous metal mines, first coordinate information, outline morphology information, element content data and second coordinate information of surface sampling points;

[0049] The analysis module is used to determine the surface cover type of different ore body areas in remote sensing images. Based on the surface cover type of different ore body areas, texture analysis and spectral analysis are performed on the ore body areas in remote sensing images to obtain texture information sets and spectral information sets.

[0050] The calibration module is used to calibrate the first coordinate information and the second coordinate information to obtain the calibrated first coordinate information and the calibrated second coordinate information. Based on the calibrated first coordinate information and the contour morphology information, the ore body framework is constructed.

[0051] The matching module is used to match the spectral information set and texture information set with the calibrated first coordinate information, and to match the element content data with the calibrated second coordinate information to form a multi-attribute dataset.

[0052] The association module is used to perform multi-source information association processing on multi-attribute datasets based on the ore body framework to form a multi-attribute feature set;

[0053] The module is used to construct a spatial block model of non-ferrous metal mines by using a multi-scale attention model based on geological knowledge constraints to complete the discrete attribute data in the multi-attribute feature set.

[0054] Thirdly, this application provides an electronic device, comprising:

[0055] Memory, used to store computer programs;

[0056] A processor is used to implement the steps of the method for constructing a multi-attribute spatial feature block model of non-ferrous metal ores as described in the first aspect above when executing a computer program.

[0057] Fourthly, this application provides a computer-readable storage medium storing a computer program, which, when executed by a processor, can implement the steps of the method for constructing a multi-attribute spatial feature block model of non-ferrous metal ores as described in the first aspect above.

[0058] The method for constructing a multi-attribute spatial feature block model of non-ferrous metal mines provided in this application achieves unified acquisition and structured organization of multi-source heterogeneous data by acquiring remote sensing images, first coordinate information, contour morphology information, element content data and second coordinate information of surface sampling points of non-ferrous metal mines. It breaks through the limitations of traditional remote sensing interpretation that relies solely on empirical interpretation or general band combinations, and enhances the ability to identify and distinguish mineralization indication information. It ensures that the geometric shape of the ore body is consistent with the actual spatial location, avoiding model distortion caused by coordinate misalignment. It achieves accurate alignment of remote sensing derived attributes and field sampling attributes under a unified spatial benchmark. It overcomes the problem of attribute and geometry disconnect in existing methods, making the attribute distribution more consistent with the actual ore body distribution pattern. It alleviates the uncertainty of attribute interpolation in sparsely sampled areas and improves the overall quality of the block model in terms of spatial continuity and attribute rationality.

[0059] Furthermore, multiple ore body regions are divided by combining geological boundary data and pixel grayscale values ​​of remote sensing images, and the land cover type is determined based on the remote sensing reflectance, physical structure parameters, and color feature parameters of each region. Then, texture extraction rules matching the land cover type are called from the preset feature dimension library, and texture information sets are extracted from the remote sensing images accordingly. At the same time, target spectral bands are selected and integrated to form spectral information sets based on the mineral spectral attributes corresponding to each land cover type.

[0060] It solves the shortcomings of shallow feature representation, strong generalization but weak specificity in traditional modeling, and provides key data support for building multi-attribute block models with high recognition and high geological consistency. Attached Figure Description

[0061] To more clearly illustrate the technical solutions in the embodiments of this application 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 some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0062] Figure 1 A flowchart illustrating the method for constructing a multi-attribute spatial feature block model of non-ferrous metal ores provided in this application embodiment;

[0063] Figure 2 A schematic diagram illustrating a process for obtaining a texture information set and a spectral information set, provided in an embodiment of this application;

[0064] Figure 3 A schematic diagram of the structure of the multi-attribute spatial feature block model construction system for non-ferrous metal ores provided in the embodiments of this application. Detailed Implementation

[0065] The current technology has shortcomings in terms of spatial structure fidelity and attribute distribution rationality in the non-ferrous metal ore prospecting stage due to the shallow use of remote sensing information, reliance on manual drawing or simplified assumptions for ore body geometry, and lack of effective mining of deep correlations between spectrum, texture and element content.

[0066] This scheme acquires remote sensing imagery, surface sampling element content, and corresponding coordinates and contour morphology information. Then, based on the differences in surface cover types in the ore body area, it conducts texture analysis and spectral interpretation to generate geologically significant texture and spectral information sets. A unified spatial benchmark is established through coordinate calibration, and a geometric framework conforming to the actual ore body distribution is constructed by combining contour morphology information. On this basis, accurate matching of remote sensing-derived attributes and sampling attributes under unified coordinates is achieved, forming a structurally consistent multi-attribute dataset. Furthermore, multi-source information is correlated based on the ore body framework to generate a multi-attribute feature set coupling space and attributes. Finally, a multi-scale attention mechanism integrating geological prior knowledge is introduced to intelligently complete sparse or missing attributes, thereby constructing a geometrically accurate, attribute-continuous, and geologically consistent multi-attribute spatial feature block model of non-ferrous metal ore, breaking through the technical bottlenecks of existing methods in terms of data fusion depth, geometric realism, and attribute interpolation reliability.

[0067] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0068] The core of this application is to provide a method for constructing a multi-attribute spatial feature block model of non-ferrous metal ores. A flowchart of one specific implementation method is shown below. Figure 1 As shown, the method includes:

[0069] Step 101: Obtain remote sensing images, first coordinate information, outline morphology information, element content data and second coordinate information of non-ferrous metal mines, as well as surface sampling points.

[0070] In this step, non-ferrous metal ore refers to mineral deposits containing non-ferrous metal elements such as copper, lead, zinc, and aluminum that have mining value. Remote sensing imagery refers to image data reflecting the characteristics of surface and underground ore bodies acquired through satellite or aerial remote sensing equipment. The first coordinate information refers to the surface location coordinate data corresponding to the remote sensing imagery.

[0071] Outline morphology information refers to the topographic relief and ore body outline data of the non-ferrous metal mine obtained through measurement. Surface sampling points refer to the locations set on the surface of the non-ferrous metal mine for collecting ore samples. Elemental content data refers to the content data of various elements in the ore obtained after testing the ore samples from the surface sampling points. Secondary coordinate information refers to the location coordinate data corresponding to the surface sampling points.

[0072] In this embodiment, the target non-ferrous metal mine is first identified. Remote sensing images of the mine can be acquired using satellite or aerial remote sensing equipment. Simultaneously, the first coordinate information corresponding to the remote sensing image can be obtained using the positioning module built into the remote sensing equipment. The terrain undulation and ore body outline of the mine can also be measured using topographic surveying equipment to obtain outline morphology information. Then, multiple surface sampling points are set up on the surface of the mine, and ore samples are collected from each surface sampling point. The ore samples are then tested using elemental detection equipment to obtain elemental content data. At the same time, the position of each surface sampling point is measured using positioning equipment to obtain second coordinate information.

[0073] For example, firstly, a type A non-ferrous metal ore is selected as the exploration target. In order to obtain the basic data of the ore comprehensively, a type B airborne remote sensing equipment equipped with a hyperspectral imager and a positioning module can be used to collect remote sensing data of the exploration area of ​​the type A non-ferrous metal ore along the preset exploration route. During the flight, the equipment synchronously records the imaging position of the remote sensing image. After coordinate transformation, the first coordinate information corresponding to the remote sensing image is generated. The image resolution and coordinate accuracy both meet the requirements of ore body feature analysis.

[0074] Next, a C-type terrain laser scanner and a 3D terrain modeling system can be used to perform a comprehensive scan and measurement of the surface topography and ore body outcrop contour of Class A non-ferrous metal mines. Through point cloud data processing and contour fitting, contour morphology information reflecting the topographic undulation and boundary morphology of the ore body can be extracted, including ore body elevation change data and planar contour coordinate data.

[0075] Subsequently, a 100m × 100m sampling grid can be drawn on the surface of the Class A non-ferrous metal mine. Surface sampling points are set at each grid node and at locations with abnormal ore body characteristics. Ore samples are collected from each sampling point using a geological hammer and sampling drill. After the samples are sent to the laboratory, the content of non-ferrous metals such as copper, zinc, and lead in the samples is detected by a D-type portable elemental analyzer to generate corresponding elemental content data.

[0076] Simultaneously, an E-type high-precision GNSS positioning instrument can be used to perform on-site positioning of each surface sampling point, recording the spatial coordinates such as latitude, longitude, and elevation of the sampling point. After coordinate calibration, the second coordinate information corresponding to each sampling point is generated. Ultimately, remote sensing images, first coordinate information, and outline morphology information of Class A non-ferrous metal mines, as well as element content data and corresponding second coordinate information of 30 surface sampling points, were successfully acquired.

[0077] Step 102: Determine the surface cover type of different ore body areas in the remote sensing image. Based on the surface cover type of different ore body areas, perform texture analysis and spectral analysis on the ore body areas in the remote sensing image to obtain texture information set and spectral information set.

[0078] In this step, the ore body region refers to the area of ​​non-ferrous metal ore with specific geological characteristics identified from remote sensing imagery. Surface cover type refers to the type of material covering the surface of the ore body region, including oxidation zones, primary zones, weathered rock layers, etc.

[0079] Step 103: Calibrate the first coordinate information and the second coordinate information to obtain calibrated first coordinate information and calibrated second coordinate information. Based on the calibrated first coordinate information and contour morphology information, construct the ore body framework.

[0080] Step 104: Match the spectral information set and texture information set with the calibrated first coordinate information, and match the element content data with the calibrated second coordinate information to form a multi-attribute dataset.

[0081] In step 105: Based on the ore body framework, multi-source information association processing is performed on the multi-attribute dataset to form a multi-attribute feature set.

[0082] Step 106: Using a multi-scale attention model based on geological knowledge constraints, data completion processing is performed on the discrete attribute data in the multi-attribute feature set to construct a spatial block model of non-ferrous metal mines.

[0083] In this step, geological knowledge constraints refer to constraint rules formulated based on existing, verified knowledge such as geological theories, metallogenic regularities, and regional geological background. Multi-scale attention models are models capable of focusing on and processing data features at different scales. Discrete attribute data refers to attribute data where multi-attribute features are concentrated but discontinuously distributed, and may contain missing or scattered attributes.

[0084] This application embodiment achieves unified acquisition and structured organization of multi-source heterogeneous data by acquiring remote sensing images of non-ferrous metal mines, first coordinate information, contour morphology information, element content data and second coordinate information of surface sampling points; ensures that the geometry of the ore body is consistent with the actual spatial location, and avoids model distortion caused by coordinate misalignment; overcomes the problem of attribute and geometry being disconnected in existing methods, making the attribute distribution more consistent with the actual ore body distribution pattern; reduces the uncertainty of attribute interpolation in sparse sampling areas, and improves the overall quality of the block model in terms of spatial continuity and attribute rationality.

[0085] This application provides a specific embodiment, such as Figure 2 As shown, step 102 involves determining the surface cover type of different ore body areas in the remote sensing image. Based on the surface cover type of different ore body areas, texture analysis and spectral analysis are performed on the ore body areas in the remote sensing image to obtain texture information sets and spectral information sets. Specifically, this includes the following steps:

[0086] Step 201: Based on the geological boundary data of non-ferrous metal mines and the pixel grayscale values ​​of remote sensing images, determine multiple ore body regions.

[0087] In this step, geological boundary data refers to spatial data representing the location, extent, and geological boundary attributes of non-ferrous metal mines. Pixel grayscale value refers to the grayscale value corresponding to each pixel in the remote sensing image, used to reflect the brightness of objects on the Earth's surface.

[0088] In this embodiment, geological boundary data of the non-ferrous metal ore is first retrieved, which clearly defines the geological boundary range of the ore body. Simultaneously, the pixel grayscale value of each pixel in the remote sensing image is extracted. Next, the geological boundary data and the pixel grayscale values ​​of the remote sensing image are spatially overlaid and analyzed. By calculating the ore body region division threshold based on the pixel grayscale values, regions with grayscale values ​​in different intervals and located within the geological boundary are divided.

[0089] The threshold for dividing ore body regions is equal to the difference between the maximum target gray value and the minimum target gray value, divided by the number of regions to be divided. The maximum target gray value refers to the maximum gray value of a pixel within the geological boundary in the remote sensing image, and the minimum target gray value refers to the minimum gray value of a pixel within the geological boundary in the remote sensing image. The number of regions to be divided refers to the specific number of remote sensing images within the geological boundary that are planned to be divided into ore body regions.

[0090] Step 202: Determine the land cover type for each ore body region based on the remote sensing reflectance, physical structure parameters, and color characteristic parameters of each ore body region.

[0091] In this step, remote sensing reflectance refers to the proportion of remote sensing electromagnetic waves reflected by surface objects in the ore body area. Physical structure parameters refer to parameters related to the physical characteristics of the surface of the ore body area, such as material composition, arrangement, and density. Color characteristic parameters refer to color-related parameters such as hue, color gradation, and color saturation of the surface of the ore body area in the remote sensing image.

[0092] In this embodiment, firstly, based on each ore body area, the remote sensing reflectance of each area is extracted using the spectral analysis function of remote sensing images. Then, the physical structure parameters of each area are obtained using surface structure detection equipment, and the color feature parameters of each area are extracted using image color analysis tools. These three types of parameters are then compared with a preset land cover type parameter library, which includes standard values ​​for remote sensing reflectance, physical structure parameters, and color feature parameters corresponding to types such as oxidized zone, primary zone, and weathered rock layer.

[0093] By calculating the similarity between the actual parameters and standard parameters of the ore body region, the specific calculation process can be achieved by first dividing the absolute difference between the actual parameter value and the standard parameter value by the standard parameter value, and then subtracting the quotient from the value 1. Then, based on the type corresponding to the standard value with the highest similarity, the surface cover type of each ore body region is determined.

[0094] Step 203: Select feature dimensions and adaptation parameter values ​​that match the land cover type from the preset feature dimension library to determine the texture extraction rules for the land cover type.

[0095] In this step, the preset feature dimension library refers to a pre-established database that includes various texture analysis dimensions and their corresponding parameter values. The dimensions in the library include roughness, continuity, and directionality. Texture extraction rules refer to the specific rules used to extract texture information of ore body areas from remote sensing images.

[0096] Step 204: Extract texture information sets from remote sensing images according to the texture extraction rules for each surface cover type.

[0097] In this embodiment, firstly, for each land cover type, the corresponding texture extraction rules are retrieved. Then, according to the feature dimensions and adaptation parameter values ​​in the rules, texture information is extracted from the corresponding mineral body area in the remote sensing image. For example, for the roughness dimension, the difference in gray values ​​within a specified size pixel block is statistically analyzed to obtain roughness texture information; for the continuity dimension, the proportion of pixels that meet the continuity length threshold is statistically analyzed to obtain continuity texture information; and for the directionality dimension, the proportion of texture directions that meet the angular range is statistically analyzed to obtain directional texture information.

[0098] Next, the texture information of all extracted ore body areas is classified and organized according to the surface cover type. Finally, the organized texture information is integrated to form a unified texture information set.

[0099] Step 205: Based on the mineral spectral attributes of each land cover type, extract the target wavelength range from the remote sensing image as the target spectral band, and integrate the target spectral bands of all land cover types to form a spectral information set.

[0100] In this step, mineral spectral properties refer to the spectral characteristics of minerals of different land cover types, such as absorption, reflection, and transmission of electromagnetic waves of different wavelengths. The target wavelength range refers to the range of electromagnetic waves that can reflect the spectral properties of minerals of a specific land cover type.

[0101] In this embodiment, the spectral properties of the corresponding minerals are first analyzed for each land cover type to identify the wavelength range in which the minerals of that type have characteristic responses in the remote sensing electromagnetic spectrum. Then, based on this wavelength range, the corresponding target wavelength range is extracted from the remote sensing image and used as the target spectral band. Next, the target spectral bands corresponding to all land cover types are collected and organized, removing duplicate wavelength ranges. Finally, the organized target spectral bands are integrated to form a spectral information set including all characteristic wavelength ranges.

[0102] This application embodiment realizes the systematic acquisition of texture features of ore body area. By analyzing the spectral properties of minerals, the target spectral bands are extracted and integrated into a spectral information set, which accurately captures the spectral features of ore body area. The final texture information set and spectral information set can comprehensively and accurately reflect the surface features of ore body area.

[0103] For example, remote sensing images of Class A non-ferrous metal mines and corresponding geological boundary data can be selected first. Pixel grayscale values ​​of the remote sensing images are extracted. The maximum and minimum target grayscale values ​​within the geological boundaries of the remote sensing images are determined. Then, a threshold for dividing the ore body regions is calculated based on the required number of ore body regions. According to this threshold, the remote sensing images within the geological boundaries are divided into three ore body regions: A, B, and C. Then, for these three ore body regions, remote sensing reflectance, physical structure parameters, and color feature parameters are extracted respectively. These parameters are compared with a preset land cover type parameter library, and after calculating the similarity, region A is determined to be an oxidation zone, region B to be a primary zone, and region C to be a weathered rock layer.

[0104] Subsequently, a pre-defined feature dimension library was retrieved. Roughness and continuity dimensions, along with appropriate parameter values, were matched for the oxidation zone, the primary zone, and the weathered rock layer. Continuity and directionality dimensions, along with appropriate parameter values, were then matched for the primary zone. Based on this, texture extraction rules for each type were formulated. Following these rules, texture information was extracted from the remote sensing images of the three ore body regions (A, B, and C). The extracted texture information was then integrated to obtain a texture information set.

[0105] Finally, by analyzing the mineral spectral properties of the oxidation zone, primary zone, and weathered rock layers, the target wavelength range corresponding to each type can be extracted from remote sensing images as target spectral bands. After integrating these target spectral bands, a spectral information set can be obtained.

[0106] This application provides a specific embodiment. Step 203 involves selecting feature dimensions and adaptation parameter values ​​that match the land cover type from a preset feature dimension library to determine the texture extraction rules for the land cover type. This specifically includes the following steps:

[0107] Step 211: Select the feature dimension and matching parameter value that are suitable for each land cover type from the preset feature dimension library. When the land cover type is oxidation zone, the matching feature dimensions are roughness dimension and continuity dimension. When the land cover type is primary zone, the matching feature dimension is roughness dimension. When the land cover type is weathered rock layer, the matching feature dimensions are continuity dimension and directionality dimension.

[0108] In this step, the oxidation zone refers to the area of ​​the ore body near the surface formed by weathering and oxidation. The roughness dimension is an analytical dimension used to represent the roughness of the surface texture in the ore body area. The continuity dimension is an analytical dimension used to represent the continuity of the surface texture in the ore body area.

[0109] The primary zone refers to the deep areas within the ore body that are unaffected or only slightly affected by weathering and oxidation. Weathered strata refer to the surface strata formed after rocks have undergone weathering. The directional dimension refers to the analytical dimension used to represent the directional characteristics of the surface texture of the ore body area.

[0110] In this embodiment of the application, firstly, a preset feature dimension library is retrieved, which stores the corresponding mapping relationship between different land cover types and feature dimensions and adaptation parameter values. Then, for each land cover type, the feature dimensions and adaptation parameter values ​​that are suitable for it are selected one by one from the library.

[0111] Specifically, when the land cover type is oxidized zone, the roughness dimension and continuity dimension and their corresponding adaptation parameter values ​​are selected; when the land cover type is primary zone, the roughness dimension and its corresponding adaptation parameter values ​​are selected; when the land cover type is weathered rock layer, the continuity dimension and directionality dimension and their corresponding adaptation parameter values ​​are selected. Finally, each land cover type is associated with the selected feature dimensions and adaptation parameter values ​​and recorded.

[0112] Step 212: Based on the feature dimensions of each land cover type, determine the texture analysis method for each land cover type. The texture analysis method for the roughness dimension is to statistically analyze the gray value difference within the pixel block of the size; the texture analysis method for the continuity dimension is to statistically analyze the proportion of pixels that meet the continuity length threshold; and the texture analysis method for the directionality dimension is to statistically analyze the proportion of texture direction that meets the angle range.

[0113] In this embodiment of the application, firstly, for each feature dimension of land cover type, the corresponding texture analysis method is determined one by one. For the roughness dimension, the texture analysis method is to statistically analyze the gray value difference within a pixel block of a specified size. The gray value difference is equal to the maximum gray value within the pixel block minus the minimum gray value within the pixel block. The maximum gray value within the pixel block refers to the maximum gray value of all pixels in a pixel block of a specified size, and the minimum gray value within the pixel block refers to the minimum gray value of all pixels in a pixel block of a specified size.

[0114] For the continuity dimension, the texture analysis method can be to statistically determine the percentage of pixels that meet the continuity length threshold. The percentage is equal to the number of pixels that meet the continuity length threshold divided by the total number of pixels in the pixel block, and then multiplied by 100%. Here, the number of pixels that meet the continuity length threshold refers to the number of pixels in a pixel block whose consecutive pixel length reaches the preset continuity length threshold, and the total number of pixels in the pixel block refers to the total number of pixels included in a pixel block of a specified size.

[0115] For the directional dimension, the texture analysis method can be to statistically determine the percentage of texture directions that conform to the angular range. The percentage of texture directions is equal to the number of texture direction pixels that conform to the angular range divided by the total number of texture direction pixels, and then multiplied by 100%. Here, the number of texture direction pixels that conform to the angular range refers to the number of pixels in a pixel block whose texture direction falls within a preset angular range, and the total number of texture direction pixels refers to the total number of pixels in a pixel block that exhibit texture direction characteristics.

[0116] Step 213: Integrate the feature dimensions, adaptation parameter values, and texture analysis methods for each land cover type to form texture extraction rules for each land cover type.

[0117] In this embodiment, firstly, for each land cover type, the corresponding feature dimensions and adaptation parameter values, as well as the corresponding texture analysis method, are retrieved. Next, these elements are organized according to a unified rule structure. First, the land cover type name is defined, then the corresponding feature dimensions, adaptation parameter values, and texture analysis methods are listed sequentially. Finally, the organized elements are combined into complete rule entries, forming a texture extraction rule specific to each land cover type.

[0118] The embodiments of this application avoid the problems of vague and inaccurate analysis results, forming the core technical barrier of this technical solution in the process of formulating texture analysis methods.

[0119] This application provides a specific embodiment. Step 103 involves calibrating the first coordinate information and the second coordinate information to obtain calibrated first coordinate information and calibrated second coordinate information. Based on the calibrated first coordinate information and contour morphology information, a ore body framework is constructed, specifically including the following steps:

[0120] Step 301: Select multiple evenly distributed and immovable natural outcrops or artificial markers from the exploration area of ​​the non-ferrous metal mine as measurement marks, and obtain the standard coordinate data of each measurement mark.

[0121] In this step, the exploration area refers to the designated geographical area within which geological exploration work for non-ferrous metal deposits is carried out. A natural outcrop refers to a portion of rock or ore body that naturally emerges from the surface within the non-ferrous metal deposit exploration area. Artificial markers refer to artificially placed markers within the non-ferrous metal deposit exploration area used for location reference.

[0122] In this embodiment of the application, the specific scope of the exploration area of ​​the non-ferrous metal mine is defined. Then, according to the principle of uniform distribution, the selection points of measurement markers are planned within the exploration area. Natural outcrops that are not easy to move or damage are selected as measurement markers. If the number or distribution of natural outcrops cannot meet the requirements, artificial markers are set at the empty locations as supplementary measurement markers. Then, high-precision positioning equipment is used to measure the spatial coordinates of each selected measurement marker to obtain the standard coordinate data corresponding to each measurement marker.

[0123] Step 302: Calculate the first deviation value between the first coordinate at each measurement mark and the corresponding standard coordinate data in the first coordinate information, and calibrate the first coordinate information according to the first deviation value to obtain the calibrated first coordinate information.

[0124] In this embodiment, the first coordinate data corresponding to each measurement marker position is first extracted from the first coordinate information. Then, the difference between the first coordinate at each measurement marker and the standard coordinate data corresponding to that measurement marker is calculated to obtain the first deviation value. Then, based on the first deviation values ​​at all measurement markers, the first coordinate information is adjusted using an overall correction method. A corresponding deviation correction amount is applied to each coordinate data in the first coordinate information to finally obtain the calibrated first coordinate information.

[0125] Step 303: Calculate the second deviation value between the second coordinate at each measurement mark and the corresponding standard coordinate data in the second coordinate information, and calibrate the second coordinate information according to the second deviation value to obtain the calibrated second coordinate information.

[0126] In this embodiment, the second coordinate data corresponding to each measurement mark position is extracted from the second coordinate information. Then, the difference between the second coordinate at each measurement mark and the standard coordinate data corresponding to that measurement mark is calculated to obtain the second deviation value. Then, based on the second deviation values ​​at all measurement marks, an overall correction operation is performed on the second coordinate information, matching the corresponding deviation correction amount for each coordinate data in the second coordinate information and completing the adjustment, thereby obtaining the calibrated second coordinate information.

[0127] Step 304: Based on the calibrated first coordinate information, divide the exploration area into multiple grid cells and obtain the grid coordinates of each grid cell.

[0128] In this embodiment, the calibrated first coordinate information is retrieved and used as a spatial reference. The exploration area of ​​the non-ferrous metal mine is divided into multiple regular grid units according to a preset grid size and division rules. Then, each grid unit is uniquely identified, and the grid coordinates, such as vertex coordinates and center coordinates, of each grid unit are obtained through coordinate calculations.

[0129] Step 305: Associate and bind the terrain undulation data and ore body boundary data in the contour morphology information with the corresponding grid cells to obtain multiple target grid cells.

[0130] In this embodiment, terrain undulation data and ore body boundary data are first separated from the contour morphology information. Then, the terrain undulation data is spatially associated with the corresponding grid cells according to elevation information. Next, the ore body boundary data is bound to the corresponding grid cells according to spatial location range. Subsequently, grid cells that are associated with both terrain undulation data and ore body boundary data are selected, and these grid cells are defined as target grid cells.

[0131] Step 306: Arrange all target grid cells in three-dimensional space according to the grid coordinates to obtain the ore body framework.

[0132] In this embodiment, firstly, the grid coordinates of each target grid cell are retrieved. Then, based on the three-dimensional spatial coordinate parameters in the grid coordinates, combined with the actual topographic undulation patterns and ore body boundary distribution characteristics of the non-ferrous metal mine, all target grid cells are arranged in an orderly manner in three-dimensional space. During the arrangement process, the elevation changes of the topographic undulation data and the spatial range limitations of the ore body boundary data are strictly followed, and finally, an ore body framework that can accurately reflect the spatial structure of the ore body is formed.

[0133] The embodiments of this application significantly improve the spatial accuracy and rationality of constructing multi-attribute spatial feature block models for non-ferrous metal ores.

[0134] This application provides a specific embodiment. Step 104 involves matching the spectral information set and texture information set with the calibrated first coordinate information, and matching the elemental content data with the calibrated second coordinate information to form a multi-attribute dataset. This specifically includes the following steps:

[0135] Step 401: Obtain the first image range of each target spectral band in the remote sensing image from the spectral information set, and perform coordinate matching between all coordinate data in the first image range of the calibrated first coordinate information and the corresponding target spectral band to obtain multiple spectral data with coordinates.

[0136] In this step, the first image range refers to the spatial coverage area of ​​each target spectral band in the spectral information set in the remote sensing image.

[0137] In this embodiment, the spectral information set is first retrieved, and each independent target spectral band is extracted. Then, the spatial indexing function of the remote sensing image is used to locate the pixel distribution area corresponding to each target spectral band in the remote sensing image, and the spatial boundary of this area is determined as the first image range. Next, all coordinate data whose coordinate values ​​fall within the first image range are extracted from the calibrated first coordinate information. These coordinate data are then bound to the corresponding target spectral bands to complete the coordinate matching operation, ultimately obtaining multiple spectral data with coordinates.

[0138] Step 402: Obtain the second image range of each texture data group in the texture information set in the remote sensing image, and perform coordinate matching between all coordinate data in the second image range of the calibrated first coordinate information and the corresponding texture data group to obtain multiple texture data with coordinates.

[0139] In this step, a texture data set refers to a collection of texture information in a texture information set, divided according to land cover type or feature dimension. The second image extent refers to the spatial coverage area of ​​each texture data set in the remote sensing image.

[0140] In this embodiment, firstly, the texture information set is retrieved and divided into multiple independent texture data groups according to the land cover type. Then, using the spatial positioning function of remote sensing imagery, the pixel distribution area corresponding to each texture data group in the remote sensing imagery is determined, and the spatial boundary of this area is defined as the second image range. Next, all coordinate data whose coordinate values ​​fall within the range of the second imagery are extracted from the calibrated first coordinate information. These coordinate data are then associated and bound one by one with the corresponding texture data group to complete the coordinate matching operation, ultimately obtaining multiple texture data with coordinates.

[0141] Step 403: Obtain the sampling point location of the element content data, and match the coordinate data corresponding to the sampling point location in the calibrated second coordinate information with the element content data to obtain the element content data with coordinates.

[0142] In this embodiment of the application, element content data is retrieved, and the location information of the ground sampling point corresponding to each data entry is extracted. Then, from the calibrated second coordinate information, coordinate data that completely matches the spatial coordinates of these sampling points are selected. The selected coordinate data is bound one-to-one with the corresponding element content data to complete the coordinate matching operation, and finally, element content data with coordinates is obtained.

[0143] Step 404: Integrate the coordinated spectral data, coordinated texture data, and coordinated element content data to form a multi-attribute dataset.

[0144] In this embodiment, firstly, coordinate-based spectral data, coordinate-based texture data, and coordinate-based element content data are integrated. Then, according to the principle of coordinate consistency, the spectral data, texture data, and element content data corresponding to the same coordinate position are merged and organized. For cases where multiple sets of data correspond to the same coordinate, the final merging result is determined by calculating data weights.

[0145] The merging result is obtained by multiplying spectral data by its weight, texture data by its weight, and elemental content data by its weight, and then dividing the sum by the total weight. Here, spectral weight refers to the proportion of spectral data during merging, texture weight refers to the proportion of texture data during merging, elemental content weight refers to the proportion of elemental content data during merging, and the total weight is the sum of the spectral weight, texture weight, and elemental content weight. Finally, all the merged and organized data are arranged in coordinate order to form a structured multi-attribute dataset.

[0146] The embodiments of this application realize the structured fusion of multi-source data, and the resulting multi-attribute dataset can accurately associate ore body attribute information of different dimensions such as spectrum, texture, and element content with spatial coordinates.

[0147] This application provides a specific embodiment. Step 105 involves performing multi-source information association processing on the multi-attribute dataset based on the ore body framework to form a multi-attribute feature set, specifically including the following steps:

[0148] Step 501: Based on the ore body framework, integrate the multi-attribute data into the coordinate-bound spectral data, coordinate-bound texture data, and coordinate-bound element content data within the target coordinate range of each target grid cell to form the associated data group of each target grid cell.

[0149] In this embodiment, the ore body framework is first retrieved, and the target coordinate range of each target grid cell is extracted from it. This range is the coordinate boundary of the target grid cell in three-dimensional space. Then, the multi-attribute dataset is traversed to filter out the coordinate data of spectral data, texture data, and element content data that fall within the target coordinate range of each target grid cell. Then, these filtered data are grouped according to the identifier of the target grid cell, and the three types of data corresponding to the same target grid cell are integrated into a whole data set, ultimately forming a unique associated data group for each target grid cell.

[0150] Step 502: Based on the preset spectral range, classify the coordinate-bearing spectral data in each associated data group to determine the target spectral range of each associated data group.

[0151] In this step, the preset spectral range refers to the predefined set of electromagnetic wave wavelength intervals used to classify spectral data, with different intervals corresponding to different spectral feature categories.

[0152] In this embodiment, firstly, a preset spectral range is retrieved. This range includes multiple different wavelength intervals and corresponding spectral category labels. Then, for the coordinate-bearing spectral data in each associated data group, the corresponding wavelength value is extracted, and the matching degree between the wavelength value and each wavelength interval in the preset spectral range is calculated. The matching degree can be obtained by subtracting the absolute difference between the actual wavelength value and the center wavelength value of the interval from the value 1, and then dividing by the wavelength span of the interval. Here, the center wavelength value of the interval refers to the center wavelength value of each wavelength interval in the preset spectral range, and the wavelength span of the interval refers to the difference between the maximum wavelength and the minimum wavelength of each wavelength interval in the preset spectral range.

[0153] Then, based on the category label corresponding to the wavelength range with the highest matching degree, the target spectral range of each associated data group is determined.

[0154] Step 503: Based on the preset texture structure type, classify the texture data with coordinates in each associated data group to determine the target structure type of each associated data group.

[0155] In this step, the preset texture structure type refers to a predefined set of texture morphology categories used to classify texture data, including different texture structure categories such as coarse, continuous, and directional.

[0156] In this embodiment, firstly, a preset texture structure type is retrieved. This type set includes various texture structure categories and corresponding standard values ​​of texture feature parameters. Then, for the texture data with coordinates in each associated data group, its texture feature parameters are extracted, and the similarity between these parameters and the standard values ​​in the preset texture structure type is calculated. The similarity can be obtained by subtracting the absolute difference between the actual texture parameter value and the standard texture parameter value from the numerical value 1, and then dividing by the standard texture parameter value.

[0157] Then, based on the category corresponding to the highest similarity standard value, the target structure type of each associated data group is determined.

[0158] Step 504: Based on the preset element content level, classify the coordinated element content data in each associated data group to determine the target content level of each associated data group.

[0159] In this step, the preset element content level refers to the predefined set of content intervals used to classify element content data, with different intervals corresponding to different content level labels.

[0160] In this embodiment, a preset element content level is first retrieved. This level set includes multiple different element content ranges and corresponding level labels. Then, for the element content data with coordinates in each associated data group, its element content value is extracted, and the preset content range to which the value belongs is determined, so as to determine the target content level of each associated data group.

[0161] Step 505: Combine the target spectral range, target structure type, and target content level corresponding to all target grid cells to form a multi-attribute feature set.

[0162] In this embodiment, the target spectral range, target structure type, and target content level corresponding to all target grid units are collected. A feature entry is created for each target grid unit, and the target spectral range, target structure type, and target content level of that unit are sequentially filled into the feature entry. Then, according to the three-dimensional spatial arrangement order of the target grid units in the ore body framework, all feature entries are sorted. Finally, the sorted feature entries are integrated into a structured dataset to form a multi-attribute feature set.

[0163] The embodiments of this application realize the deep integration of ore body spatial units and multi-dimensional attribute features. The resulting multi-attribute feature set can accurately reflect the comprehensive attribute features of each ore body spatial unit, providing a structured and targeted data source for subsequent completion processing of discrete attribute data.

[0164] This application provides a specific embodiment. Step 106 involves using a multi-scale attention model based on geological knowledge constraints to perform data completion processing on discrete attribute data in a multi-attribute feature set, in order to construct a spatial block model of non-ferrous metal ore. The specific steps include:

[0165] Step 601: Determine whether there is missing data in each multi-attribute feature group in the multi-attribute feature set. If there is missing data, the multi-attribute feature group is taken as the target feature group. If there is no missing data, the multi-attribute feature group is taken as the complete feature group.

[0166] In this step, a multi-attribute feature set refers to the set of attribute features corresponding to a single target grid cell in the multi-attribute feature set. Missing data refers to attribute data entries in the multi-attribute feature set that were not collected or entered.

[0167] In this embodiment, the multi-attribute feature set is traversed, and each independent multi-attribute feature group is extracted. Then, the attribute data entries of each multi-attribute feature group are checked one by one to determine whether there are any missing data entries or missing values. If the check finds that a multi-attribute feature group has missing data, the multi-attribute feature group is marked as a target feature group. If the check finds that all data entries of a multi-attribute feature group are complete, the multi-attribute feature group is marked as a complete feature group.

[0168] Step 602: Based on the spatial distribution scale of the non-ferrous metal ore body, the target feature group is divided into scales to obtain the first feature subset, the second feature subset, and the third feature subset.

[0169] In this step, the spatial distribution scale of the ore body refers to the scale characteristics of the ore body of non-ferrous metal ore in three-dimensional space, such as its extension range, distribution area and volume.

[0170] In this embodiment, data related to the spatial distribution scale of the non-ferrous metal ore body are first retrieved, specifically including the three-dimensional extension length, distribution area, and volume index of the ore body. Then, these indices are normalized. Specifically, the ratio of the actual extension length of the ore body to the extension length of the largest ore body in the region is determined as the normalized extension length; the ratio of the actual distribution area of ​​the ore body to the distribution area of ​​the largest ore body in the region is determined as the normalized distribution area; and the ratio of the actual volume of the ore body to the volume of the largest ore body in the region is determined as the normalized volume.

[0171] Based on the above normalization results, a total index of the spatial distribution scale of the ore body is calculated. The value of this total index is one-third of the sum of the normalized extension length, normalized distribution area, and normalized volume. On this basis, a scale division threshold is further calculated and set as one-third of the total index of the spatial distribution scale of the ore body.

[0172] Finally, the target feature groups are classified according to the scale division threshold. Specifically, this can be determined by obtaining the distribution range of the target grid cells corresponding to the target feature group in the ore body space. When the distribution range is greater than twice the scale division threshold, the corresponding target feature group is divided into the first feature subset; when the distribution range is between the scale division threshold and twice the scale division threshold, the corresponding target feature group is divided into the second feature subset; when the distribution range is less than the scale division threshold, the corresponding target feature group is divided into the third feature subset.

[0173] Step 603: Use a multi-scale attention model constrained by geological knowledge to process the first feature subset, the second feature subset, and the third feature subset respectively to obtain the associated feature regions corresponding to the target feature group.

[0174] In this embodiment, a multi-scale attention model constrained by geological knowledge is first invoked. This model incorporates geological knowledge constraints such as the metallogenic regularity and lithological distribution of non-ferrous metal deposits. Then, the first feature subset, the second feature subset, and the third feature subset are respectively input into the model. The model will perform feature extraction and spatial correlation analysis on feature subsets of different scales, identify regions that are strongly correlated with each target feature group in terms of geological features and spatial location, and finally output the associated feature regions corresponding to the target feature groups.

[0175] Step 604: Select the attribute data that is not missing in the associated feature region and meets the geological constraints as the valid attribute data. Based on the valid attribute data, calculate the supplementary data of the discrete attribute data and add the supplementary data to the target feature group to obtain the complete feature group.

[0176] In this embodiment, all attribute data within the associated feature region can be screened one by one to identify attribute data with no missing entries and values ​​that meet geological constraints. These data are then determined as valid attribute data. Next, the completed data of the discrete attribute data is calculated based on the valid attribute data. The completed data is equal to the mean of the valid attribute data multiplied by a geological correction coefficient. This geological correction coefficient is a correction coefficient set according to regional geological patterns and ore body characteristics, and its value is determined by geological experts based on actual mineralization conditions.

[0177] Then, the calculated complete data is added to the corresponding target feature group to fill the gaps in the missing data, and finally the supplemented target feature group is transformed into a complete feature group.

[0178] Step 605: Associate and bind all complete feature groups and corresponding target grid cells in the ore body framework to form a spatial block model of non-ferrous metal ore.

[0179] In this embodiment, a complete set of complete feature groups can be formed by integrating the complete feature groups and the supplemented complete feature groups. Then, the ore body framework is retrieved, and the target grid cells corresponding to each complete feature group are extracted. According to the principle of spatial coordinate consistency, each complete feature group is associated and bound with its corresponding target grid cell. Finally, all the bound target grid cells and complete feature groups are systematically integrated according to the three-dimensional spatial structure of the ore body to construct a spatial block model that can intuitively reflect the spatial distribution characteristics of multiple attributes of non-ferrous metal ore.

[0180] This application embodiment achieves accurate completion of discrete attribute data, making the completed data geologically reasonable. By associating and binding complete feature groups with target grid units to form a spatial block model, the model can completely and accurately reflect the multi-attribute spatial characteristics of non-ferrous metal mines. The final constructed spatial block model provides reliable three-dimensional data support for the exploration planning, resource assessment and development of non-ferrous metal mines.

[0181] Figure 3 This is a schematic diagram of a specific implementation of the multi-attribute spatial feature block model construction system for non-ferrous metal ores provided in this application embodiment, with reference to... Figure 3 The system may include:

[0182] The acquisition module 21 is used to acquire remote sensing images of non-ferrous metal mines, first coordinate information, outline morphology information, element content data and second coordinate information of surface sampling points;

[0183] The analysis module 22 is used to determine the surface cover type of different ore body areas in the remote sensing image. Based on the surface cover type of different ore body areas, texture analysis and spectral analysis are performed on the ore body areas in the remote sensing image to obtain texture information set and spectral information set.

[0184] The calibration module 23 is used to calibrate the first coordinate information and the second coordinate information to obtain the calibrated first coordinate information and the calibrated second coordinate information. Based on the calibrated first coordinate information and the contour morphology information, the ore body framework is constructed.

[0185] The matching module 24 is used to match the spectral information set and texture information set with the calibrated first coordinate information, and to match the element content data with the calibrated second coordinate information to form a multi-attribute dataset.

[0186] The association module 25 is used to perform multi-source information association processing on multi-attribute datasets based on the ore body framework to form a multi-attribute feature set;

[0187] Module 26 is used to construct a spatial block model of non-ferrous metal mines by using a multi-scale attention model based on geological knowledge constraints to complete the discrete attribute data in the multi-attribute feature set.

[0188] The multi-attribute spatial feature block model construction system for non-ferrous metal ores in this application embodiment is used to implement the aforementioned multi-attribute spatial feature block model construction method for non-ferrous metal ores. Therefore, the specific implementation of the multi-attribute spatial feature block model construction system for non-ferrous metal ores can be found in the embodiment section of the multi-attribute spatial feature block model construction method for non-ferrous metal ores mentioned above. The specific implementation can be referred to the description of the corresponding embodiments, and will not be repeated here.

[0189] This application also provides an electronic device, comprising: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of the above-described method for constructing a multi-attribute spatial feature block model of a non-ferrous metal ore.

[0190] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described method for constructing a multi-attribute spatial feature block model of non-ferrous metal ores.

[0191] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as USB flash drives, read-only memory, random access memory, portable hard drives, magnetic disks, or optical disks.

[0192] The embodiments of the present invention also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in the embodiments of the method for constructing a multi-attribute spatial feature block model of any non-ferrous metal ore.

[0193] Those skilled in the art will further 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, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. 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.

[0194] The above provides a detailed description of the method and system for constructing a multi-attribute spatial feature block model of non-ferrous metal ores. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are merely for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make various improvements and modifications to this application without departing from its principles, and these improvements and modifications also fall within the protection scope of this application.

Claims

1. A method for constructing a multi-attribute spatial feature block model of non-ferrous metal ores, characterized in that, include: The method involves acquiring remote sensing images of non-ferrous metal mines, first coordinate information, outline morphology information, element content data of surface sampling points, and second coordinate information. The first coordinate information refers to the coordinate data of the surface location corresponding to the remote sensing image of the non-ferrous metal mine, and the second coordinate information refers to the coordinate data of the location corresponding to the surface sampling point in the non-ferrous metal mine. The surface cover type of different ore body areas in the remote sensing image is determined. Based on the surface cover type of different ore body areas, texture analysis and spectral analysis are performed on the ore body areas in the remote sensing image to obtain texture information set and spectral information set. The first coordinate information and the second coordinate information are calibrated to obtain calibrated first coordinate information and calibrated second coordinate information. Based on the calibrated first coordinate information and the contour morphology information, the ore body framework is constructed. The spectral information set and the texture information set are matched with the calibrated first coordinate information, and the element content data is matched with the calibrated second coordinate information to form a multi-attribute dataset. Based on the ore body framework, the multi-attribute dataset is subjected to multi-source information association processing to form a multi-attribute feature set; A multi-scale attention model based on geological knowledge constraints is used to complete the discrete attribute data in the multi-attribute feature set in order to construct a spatial block model of non-ferrous metal mines. A multi-scale attention model based on geological knowledge constraints is used to perform data completion processing on the discrete attribute data in the multi-attribute feature set, in order to construct a spatial block model of non-ferrous metal deposits, including: Determine whether there is missing data in each multi-attribute feature group in the multi-attribute feature set. If there is missing data, the multi-attribute feature group is taken as the target feature group. If there is no missing data, the multi-attribute feature group is taken as the complete feature group. Based on the spatial distribution scale of the non-ferrous metal ore body, the target feature group is divided by scale to obtain a first feature subset, a second feature subset, and a third feature subset; A multi-scale attention model constrained by geological knowledge is used to process the first feature subset, the second feature subset, and the third feature subset respectively to obtain the associated feature regions corresponding to the target feature group; The attribute data that are not missing and meet the geological constraints in the associated feature region are taken as valid attribute data. Based on the valid attribute data, the complete data of the discrete attribute data is calculated and the complete data is added to the target feature group to obtain the complete feature group. All complete feature groups and corresponding target grid cells in the ore body framework are associated and bound to form a spatial block model of the non-ferrous metal ore.

2. The method according to claim 1, characterized in that, The surface cover type of different ore body regions in the remote sensing image is determined. Based on the surface cover type of different ore body regions, texture analysis and spectral analysis are performed on the ore body regions in the remote sensing image to obtain texture information sets and spectral information sets, including: Based on the geological boundary data of the non-ferrous metal mine and the pixel grayscale values ​​of the remote sensing image, multiple ore body regions are determined. Based on the remote sensing reflectance, physical structure parameters and color characteristic parameters of each ore body area, the land cover type of each ore body area is determined. Select feature dimensions and adaptation parameter values ​​that match the land cover type from a preset feature dimension library to determine the texture extraction rules for the land cover type; Based on the texture extraction rules for different surface cover types, a texture information set is extracted from the remote sensing image. Based on the mineral spectral properties of each land cover type, the target wavelength range is extracted from the remote sensing image as the target spectral band, and the target spectral bands of all land cover types are integrated to form a spectral information set.

3. The method according to claim 2, characterized in that, Select feature dimensions and adaptation parameter values ​​that match the land cover type from a preset feature dimension library to determine the texture extraction rules for the land cover type, including: Select the feature dimension and matching parameter value that are suitable for each land cover type from the preset feature dimension library. When the land cover type is oxidized zone, the matching feature dimensions are roughness dimension and continuity dimension. When the land cover type is primary zone, the matching feature dimension is roughness dimension. When the land cover type is weathered rock layer, the matching feature dimensions are continuity dimension and directionality dimension. Based on the characteristic dimensions of each land cover type, a texture analysis method for each land cover type is determined. The texture analysis method for the roughness dimension is to statistically analyze the gray value difference within a pixel block of a certain size; the texture analysis method for the continuity dimension is to statistically analyze the percentage of pixels that meet the continuity length threshold; and the texture analysis method for the directionality dimension is to statistically analyze the percentage of texture directions that meet the angular range. The feature dimensions, adaptation parameter values, and texture analysis methods of each land cover type are integrated to form texture extraction rules for each land cover type.

4. The method according to claim 1, characterized in that, The first coordinate information and the second coordinate information are calibrated to obtain calibrated first coordinate information and calibrated second coordinate information. Based on the calibrated first coordinate information and the contour morphology information, a ore body framework is constructed, including: Multiple evenly distributed and immovable natural outcrops or artificial markers are selected as measurement marks within the exploration area of ​​the non-ferrous metal mine, and the standard coordinate data of each measurement mark is obtained. Calculate the first deviation value between the first coordinate at each measurement mark in the first coordinate information and the corresponding standard coordinate data, and calibrate the first coordinate information according to the first deviation value to obtain the calibrated first coordinate information; Calculate the second coordinate and the second deviation value of the corresponding standard coordinate data at each measurement mark in the second coordinate information, and calibrate the second coordinate information according to the second deviation value to obtain the calibrated second coordinate information; Based on the calibrated first coordinate information, the exploration area is divided into multiple grid cells, and the grid coordinates of each grid cell are obtained; The terrain undulation data and ore body boundary data in the contour morphology information are associated and bound with the corresponding grid cells to obtain multiple target grid cells; Based on the grid coordinates, all target grid cells are arranged in three-dimensional space to obtain the ore body framework.

5. The method according to claim 1, characterized in that, The spectral information set and the texture information set are matched with the calibrated first coordinate information, and the element content data is matched with the calibrated second coordinate information to form a multi-attribute dataset, including: The first image range of each target spectral band in the spectral information set is obtained in the remote sensing image, and all coordinate data in the first image range of the calibrated first coordinate information are matched with the corresponding target spectral band to obtain multiple spectral data with coordinates. The second image range of each texture data group in the texture information set is obtained in the remote sensing image, and all coordinate data in the calibrated first coordinate information within the second image range are matched with the corresponding texture data group to obtain multiple texture data with coordinates. The sampling point positions of the element content data are obtained, and the coordinate data corresponding to the sampling point positions in the calibrated second coordinate information are matched with the element content data to obtain element content data with coordinates. The coordinate-bearing spectral data, the coordinate-bearing texture data, and the coordinate-bearing element content data are integrated to form a multi-attribute dataset.

6. The method according to claim 1, characterized in that, Based on the ore body framework, the multi-attribute dataset undergoes multi-source information association processing to form a multi-attribute feature set, including: Based on the ore body framework, the multi-attribute data are integrated into the coordinate-bound spectral data, coordinate-bound texture data, and coordinate-bound element content data within the target coordinate range of each target grid cell, forming an associated data group for each target grid cell. Based on a preset spectral range, the coordinate-bearing spectral data in each associated data group are categorized to determine the target spectral range for each associated data group. Based on the preset texture structure type, the texture data with coordinates in each associated data group are classified to determine the target structure type of each associated data group; Based on the preset element content level, the coordinate element content data in each associated data group are classified to determine the target content level of each associated data group. The target spectral range, target structure type, and target content level corresponding to all target grid units are combined to form a multi-attribute feature set.

7. A system for constructing multi-attribute spatial feature block models of non-ferrous metal ores, characterized in that, include: The acquisition module is used to acquire remote sensing images of non-ferrous metal mines, first coordinate information, outline morphology information, element content data and second coordinate information of surface sampling points. The first coordinate information refers to the coordinate data of the surface location corresponding to the remote sensing image of the non-ferrous metal mine, and the second coordinate information refers to the coordinate data of the location corresponding to the surface sampling point in the non-ferrous metal mine. The analysis module is used to determine the surface cover type of different ore body areas in the remote sensing image, and to perform texture analysis and spectral analysis on the ore body areas in the remote sensing image based on the surface cover type of different ore body areas to obtain texture information set and spectral information set; The calibration module is used to calibrate the first coordinate information and the second coordinate information to obtain calibrated first coordinate information and calibrated second coordinate information, and to construct the ore body framework based on the calibrated first coordinate information and the contour morphology information. The matching module is used to match the spectral information set and the texture information set with the calibrated first coordinate information, and to match the element content data with the calibrated second coordinate information to form a multi-attribute dataset. The association module is used to perform multi-source information association processing on the multi-attribute dataset based on the ore body framework to form a multi-attribute feature set; The construction module is used to perform data completion processing on the discrete attribute data in the multi-attribute feature set using a multi-scale attention model based on geological knowledge constraints, so as to construct a spatial block model of non-ferrous metal mines. Specifically, the construction module is used to determine whether there is missing data in each multi-attribute feature group in the multi-attribute feature set. If there is missing data, the multi-attribute feature group is taken as the target feature group; if there is no missing data, the multi-attribute feature group is taken as the complete feature group. Based on the spatial distribution scale of the non-ferrous metal ore body, the target feature group is divided by scale to obtain a first feature subset, a second feature subset, and a third feature subset; A multi-scale attention model constrained by geological knowledge is used to process the first feature subset, the second feature subset, and the third feature subset respectively to obtain the associated feature regions corresponding to the target feature group; The attribute data that are not missing and meet the geological constraints in the associated feature region are taken as valid attribute data. Based on the valid attribute data, the complete data of the discrete attribute data is calculated and the complete data is added to the target feature group to obtain the complete feature group. All complete feature groups and corresponding target grid cells in the ore body framework are associated and bound to form a spatial block model of the non-ferrous metal ore.

8. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the method for constructing a multi-attribute spatial feature block model of a non-ferrous metal ore as described in any one of claims 1 to 6 when executing the computer program.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, enables the construction method of a multi-attribute spatial feature block model of non-ferrous metal ores as described in any one of claims 1 to 6.

Images