Method and system for fusing remote sensing hyperspectral data with geochemical element layers

Abstract

The present invention discloses a method and system for fusing remote sensing hyperspectral data with a geochemical element map. The method comprises: obtaining to-be-processed geochemical data and its corresponding remote sensing hyperspectral image, and performing preprocessing to obtain a geochemical element map and a hyperspectral surface reflectance image; enhancing and extracting feature information from the geochemical element map using a local singularity analysis method to obtain a local singularity index map; extracting diagnostic absorption features associated with preset target chemical elements from the hyperspectral surface reflectance image to obtain a hyperspectral diagnostic absorption feature map; then constructing a regression equation model using a random forest model; and inverting the singularity index based on the regression equation model and the diagnostic absorption features to obtain a high-resolution geochemical element local singularity index map for the target area. The present invention can fuse the geochemical element map with the spectral information of ground objects in the hyperspectral image, enriching the spatial details of the geochemical data and improving its accuracy.

Description

Technical Field

[0001] The present invention relates to the technical field of fusing remote sensing spectral data with geochemical data, and in particular to a method and system for fusing remote sensing hyperspectral data with geochemical element layers. Background Art

[0002] Remote sensing (RS) is a technology that uses sensors to detect the interaction between ground objects and electromagnetic waves in specific spectral bands (such as radiation, reflection, scattering, and polarization) to identify ground objects and their physical and chemical properties. Currently, the commonly used spectral bands for remote sensing are the visible-shortwave infrared, mid-infrared, thermal infrared, and microwave bands. The visible-shortwave infrared wavelength range is 0.38-2.50μm, the mid-infrared wavelength range is 3-5μm, the thermal infrared wavelength range is 8-14μm, and the microwave wavelength range is 0.8-30cm. The development of remote sensing has greatly broadened human vision and visual ability, becoming a powerful tool for studying Earth and planetary sciences and an indispensable technical means for geological research and exploration. Remote sensing technology is used to systematically survey the geological characteristics of a target area, primarily including lithology, alteration zones, fault structures, and alteration or indicator minerals. In recent years, remote sensing technology, thanks to its advantages such as wide coverage, high resolution, high cost-effectiveness and fast data update, complements other geological survey technologies and has become an indispensable means of current earth science research, especially in remote areas with high altitude, high cold and deep cutting.

[0003] Optical remote sensing imagery can be categorized into multispectral and hyperspectral data based on the number of spectra. Multispectral data, exemplified by ASTER (Advanced Spaceborne Thermal Emission and Reflection Radiometer) images, are widely used in lithologic mapping and identification of mineralized alteration anomalies. Key methods include band ratio analysis and principal component analysis, but the degree of quantification is relatively low. The introduction of the Hyperion sensor aboard the EO-1 satellite in 2000 marked the beginning of hyperspectral remote sensing. In recent years, new satellite-borne hyperspectral data sources have emerged, including those from the Ziyuan-1 02D / 02E satellites, the Gaofen-5 satellite, and PRISMA (PRecursore IperSpettrale dellaMissione Applicativa). Space-borne hyperspectral imagery is often the preferred choice for research due to its large coverage area, relatively low acquisition cost, and high spatial resolution (typically 30 meters). Hyperspectral remote sensing imagery (HSI) combines two-dimensional imaging with spectral technology to capture the spatial distribution of surface features while simultaneously recording spectral information using dozens or even hundreds of densely packed, continuous bands. Since its emergence, hyperspectral remote sensing has enabled quantitative and precise identification of rock and mineral types and abundances based on features such as spectral similarity metrics and derived absorption band parameters. It can also, to a certain extent, reveal subtle variations in mineral chemical composition and the temperature and pressure conditions of mineral formation.

[0004] Geochemical data (Geochemistry, Geochemistry) is a crucial foundation for basic geological research and mineral exploration. Through standardized geochemical sampling and high-quality experimental analysis, it determines the content (abundance) of most chemical elements in nature and identifies their spatial distribution and enrichment. In practice, geochemical sampling data are often interpolated to generate regional geochemical element maps. However, due to factors such as manual collection costs, regional variations in sampling quality, and topographical constraints, geochemical data collection often requires relatively few sampling points, resulting in low accuracy. Regionally, geochemical data are typically collected at small scales, such as 1:500,000 or 1:200,000, with sampling points spaced over 1 km apart. This results in low resolution for geochemical element maps generated from interpolated geochemical data collected at these sampling points. Larger scale geochemical data, such as 1:50,000 or 1:10,000, are far from sufficient to cover most regions due to the enormous workload involved. Consequently, in many regional studies, small-scale geochemical data are the only reference.

[0005] Based on the above situation, if hyperspectral data can be fused with geochemical data to obtain higher-resolution element spatial distribution and enrichment, it can not only save the cost of manual sampling, but also closely combine the hyperspectral characteristics with the changes in mineral chemical composition, and the application significance will be clearer. The fusion processing of hyperspectral data and geochemical data is based on the correlation between them. From the perspective of spectral response mechanism, rock minerals have a series of diagnostic spectral characteristic information within the spectral range, such as metal ions (such as Fe 3+ / Fe 2+ ) and electron transfer of Al-OH / Mg-OH and CO3 2- The spectral absorption characteristics of minerals are formed by the vibration of molecular groups. The spectral characteristics of rocks are primarily determined by the abundance of these specific ions (clusters) within the various minerals (including cements) that make up the rocks. This demonstrates the spectral response to geochemical information, and this relevant information is necessarily reflected in the data. Therefore, integrating hyperspectral data with regional geochemical data to accurately determine the spatial distribution and enrichment of elements is particularly urgent and necessary.

[0006] There are usually two processing strategies for the fusion of remote sensing and geochemical data: the first is fitting, which mainly expresses the spatial relationship between the two data information without delving into the causal relationship; the second is coupling, which considers both spatial and causal correlations.

[0007] The fusion of remote sensing and geochemical data based on overlaying requires the use of appropriate numerical conversion methods before data processing to convert the geochemical data into image data. This involves normalizing the data to ensure spatial and dimensional consistency with the remote sensing image, facilitating fusion. Typical numerical conversion methods include color image synthesis and arithmetic operations. However, overlaying-based fusion techniques fail to consider the causal relationships between the data and simply overlay the data based on spatial location. This can lead to false anomalies and lacks interpretability.

[0008] The fusion of remote sensing and chemical data based on coupling can be divided into the data layer, decision layer, and symbol layer. Typical methods for the data layer include pixel-based image weighted fusion, image linear or nonlinear model fusion, and fusion based on wavelet technology. The data layer is quantitative data, while the decision layer and symbol layer are qualitative data. There are significant differences in the processing of quantitative and qualitative data. Remote sensing image classification categories, remote sensing alteration information anomaly classification, and geochemical element anomaly contour grading maps are all qualitative data, and are typically fused using features such as correlation coefficient transformation and information entropy. However, coupled data layer fusion typically uses data dimensionality reduction techniques to fuse remote sensing bands with geochemical element layers, lacking feature screening techniques with clear physical meaning. Fusion at the decision layer and symbol layer is less quantitative and results in larger errors.

[0009] The fusion methods of the two aforementioned strategies are mostly used to fuse multispectral data with geochemical element layers, while there is less research on the fusion of hyperspectral data with geochemical data. Therefore, a method for fusing remote sensing hyperspectral data with geochemical element layers is urgently needed to solve the above problems. Summary of the Invention

[0010] The present invention aims to address the shortcomings of existing technologies by providing a method and system for fusing remote sensing hyperspectral data with geochemical element maps. This method can fuse low-resolution geochemical element maps with the fine spectral information of ground objects in hyperspectral imagery, enriching the spatial detail of geochemical data, reducing the uncertainty of spatial interpolation, enhancing weak geochemical anomaly information, and improving the accuracy of small-scale geochemical data.

[0011] The object of the present invention is achieved through the following technical solutions: In a first aspect, an embodiment of the present invention provides a method for fusing remote sensing hyperspectral data with geochemical element layers, comprising the following steps:

[0012] (1) Obtain the geochemical data to be processed and its corresponding remote sensing hyperspectral image, and preprocess them to obtain geochemical element layers and hyperspectral surface reflectance images;

[0013] (2) Enhance and extract feature information of geochemical element layers through local singularity analysis methods to obtain local singularity index maps; extract diagnostic absorption features related to preset target chemical elements in hyperspectral surface reflectance images to obtain hyperspectral diagnostic absorption feature layers;

[0014] (3) Based on the local singularity index at the actual geochemical sampling point and the corresponding hyperspectral diagnostic absorption characteristics, a regression equation model is constructed using the random forest model;

[0015] (4) Based on the regression equation model and diagnostic absorption characteristics, the singularity index at other locations in the target area except the actual sampling point is inverted to obtain a high-resolution local singularity index map of geochemical elements in the target area.

[0016] Furthermore, in step (1), the geochemical data is preprocessed, specifically including:

[0017] First, the zero values ​​and outliers in the geochemical data are processed; then the geochemical data of the point data type are converted into a raster geochemical element layer using the spatial interpolation method; among them, the raster pixel size in the geochemical element layer is the same as the sampling interval of the geochemical data of the point data type.

[0018] Furthermore, in step (1), the remote sensing hyperspectral image is preprocessed, specifically including:

[0019] According to the satellite preset parameters and terrain characteristics, the remote sensing hyperspectral image is subjected to radiometric calibration, atmospheric correction and orthorectification processing, and then denoised using the hyperspectral image mixed noise removal method to obtain the denoised hyperspectral surface reflectance image.

[0020] Furthermore, the feature information enhancement and extraction of the geochemical element layer by the local singularity analysis method to obtain the local singularity index map specifically includes:

[0021] For the characteristic information in the geochemical element layer, the local singularity analysis method is used to perform local anomaly enhancement and singularity measurement of the field density in the fractal space to determine the fractal density and local singularity index; that is, for each grid cell in the geochemical element layer, the field value in the neighborhood centered on the grid cell in the fractal space is calculated; assuming that the field value obeys a multifractal distribution, the field density on the neighborhood is calculated based on the field value and neighborhood size in the neighborhood and the Euclidean space dimension, and the field density is singularly measured in the fractal space to obtain the local singularity index at each grid cell in the geochemical element layer; based on the local singularity index at each grid cell, a local singularity index map is generated and resampled to the same spatial resolution as the hyperspectral surface reflectance image.

[0022] Furthermore, the diagnostic absorption characteristics include the deepest absorption wavelength position, the deepest absorption depth, the maximum absorption area and the absorption spacing.

[0023] Furthermore, the extraction of diagnostic absorption features related to preset target chemical elements in the hyperspectral surface reflectance image to obtain a hyperspectral diagnostic absorption feature layer specifically includes:

[0024] First, the spectrum in the local spectral interval of the hyperspectral surface reflectance image is continuum-removed; then, a plurality of continuous band spectra are fitted in sequence using a bivariate polynomial; the fitting function is differentiated, and the wavelength at the position where the derivative is 0 is obtained as the absorption wavelength position, and the absorption depth is obtained by subtracting the fitting function value at the absorption wavelength position from 1; secondly, the deepest absorption depth is obtained by comparing multiple absorption depths in the local spectral interval, and the deepest absorption wavelength position is obtained according to the absorption wavelength position corresponding to the deepest absorption depth; then, the maximum absorption area is obtained according to the area of ​​the triangle formed by the three spectral measured points closest to the deepest absorption wavelength position, and the absorption spacing is obtained according to the distance between the deepest absorption wavelength position and the second deepest absorption wavelength position; finally, a hyperspectral diagnostic absorption feature layer is obtained according to the deepest absorption depth, deepest absorption wavelength position, maximum absorption area and absorption spacing in multiple local spectral intervals in the hyperspectral surface reflectance image.

[0025] Furthermore, the local singularity index at the actual geochemical sampling point and the corresponding hyperspectral diagnostic absorption characteristics are used to construct a regression equation model using a random forest model, specifically including:

[0026] Based on the local singularity index map and the hyperspectral diagnostic absorption feature layer, at the position corresponding to the actual geochemical sampling point, the local singularity index of the geochemical elements at the corresponding position is used as the output of the regression equation model, and the diagnostic absorption feature at the corresponding position is used as the input of the regression equation model. The random forest model is used to construct the regression equation model.

[0027] Furthermore, the inversion of the singularity index at other locations in the target area except the actual sampling point based on the regression equation model and the diagnostic absorption characteristics to obtain a high-resolution local singularity index map of geochemical elements in the target area as a whole specifically includes:

[0028] According to the self-similarity of the Earth system, a regression equation model between the local singularity index and the diagnostic absorption characteristics established at the actual geochemical sampling point is used to invert the singularity index at other positions in the target area except the actual sampling points. That is, the diagnostic absorption characteristics at other positions in the target area except the actual sampling points are used as the input of the regression equation model to obtain the singularity index at the corresponding position, so as to obtain a high-resolution local singularity index map of geochemical elements in the target area as a whole.

[0029] A second aspect of an embodiment of the present invention provides a system for implementing the above-mentioned method of fusing remote sensing hyperspectral data with geochemical element layers, comprising:

[0030] The data reading and preprocessing module is used to obtain the geochemical data to be processed and the remote sensing hyperspectral image consistent with the coverage area of ​​the geochemical data to be processed, and preprocess them to obtain the geochemical element layer and the hyperspectral surface reflectance image;

[0031] The feature information enhancement and extraction module is used to enhance and extract feature information of the geochemical element layer through the local singularity analysis method to obtain a local singularity index map; extract diagnostic absorption features related to preset target chemical elements in the hyperspectral surface reflectance image to obtain a hyperspectral diagnostic absorption feature layer;

[0032] A regression function building module is used to construct a regression equation model using a random forest model based on the local singularity index and the corresponding hyperspectral diagnostic absorption characteristics at the actual geochemical sampling points;

[0033] The geochemical element local singularity layer inversion module is used to invert the singularity index at locations other than the actual sampling points in the target area based on the regression equation model and diagnostic absorption characteristics, so as to obtain a high-resolution geochemical element local singularity index map for the entire target area.

[0034] A third aspect of an embodiment of the present invention provides an electronic device, comprising a memory and a processor, wherein the memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the above-mentioned method of fusing remote sensing hyperspectral data with geochemical element layers.

[0035] The beneficial effects of the present invention are as follows: the present invention constructs a method and system for fusing low-resolution geochemical element layers with diagnostic features of high-resolution hyperspectral data. This method can enrich the spatial details of geochemical data, reduce the uncertainty of spatial interpolation, enhance weak geochemical anomaly information, and improve the accuracy of small-scale geochemical data without increasing the workload and difficulty of geochemical sampling. After the geochemical data and hyperspectral diagnostic feature information are fused, comprehensive prospecting information can be quantitatively enhanced, providing a more reliable basis for geological prospecting. The present invention uses diagnostic absorption features in hyperspectral data that are closely related to mineralization information to achieve fusion with geochemical data. On the one hand, the parameters of the diagnostic absorption features, such as the deepest absorption wavelength position, deepest absorption depth, absorption area, and absorption spacing, are dense planar layers that can not only reflect the distribution of surface minerals, but also quantitatively indicate the continuous spatial variation of the abundance of specific ions / groups to a certain extent. On the other hand, the entire data fusion process has clear geological significance and strong interpretability. There is a significant scale effect between geochemical data and hyperspectral data. The present invention uses a local singularity analysis method to enhance the geochemical element layer before fusion, which helps to alleviate false anomaly information caused by the scale effect. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a flow chart of the method for fusing remote sensing hyperspectral data with geochemical element layers of the present invention;

[0037] Figure 2 is a schematic diagram of the hyperspectral diagnostic absorption characteristics of the present invention;

[0038] Figure 3 This is a structural diagram of a system for fusing remote sensing hyperspectral data with geochemical element layers of the present invention;

[0039] Figure 4 is a hyperspectral surface reflectance image in one embodiment of the present invention;

[0040] Figure 5 It is a 1:200,000 copper element layer in one embodiment of the present invention;

[0041] Figure 6 This is a 1:200,000 copper element local singularity index diagram in one embodiment of the present invention;

[0042] Figure 7 is a color composite image of the hyperspectral diagnostic absorption characteristics in one embodiment of the present invention;

[0043] Figure 8 This is a high-resolution copper element local singularity index inversion map in one embodiment of the present invention. DETAILED DESCRIPTION

[0044] The exemplary embodiments will be described in detail herein, with examples shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. Obviously, the drawings used in the following description are only some embodiments of the present invention, and it is possible for a person of ordinary skill in the art to derive other drawings based on these drawings without inventive effort. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0045] The terms used in this invention are for the purpose of describing specific embodiments only and are not intended to limit the invention. The singular forms "a," "the," and "the" used in this invention and the appended claims are also intended to include plural forms unless the context clearly indicates otherwise. It should also be understood that the term "and / or" as used herein refers to and includes any or all possible combinations of one or more of the associated listed items.

[0046] It should be understood that although the terms "first," "second," "third," etc. may be used in the present invention to describe various information, such information should not be limited to these terms. These terms are merely used to distinguish information of the same type from one another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present invention. Depending on the context, the term "if" as used herein may be interpreted as "when," "when," or "in response to determining."

[0047] The present invention will be described in detail below with reference to the accompanying drawings. Unless there is any conflict, the features of the following embodiments and implementations may be combined with each other.

[0048] See also Figure 1 The method for fusing remote sensing hyperspectral data with geochemical element layers of the present invention specifically comprises the following steps:

[0049] (1) Obtain the geochemical data to be processed and its corresponding remote sensing hyperspectral image, and preprocess them to obtain geochemical element layers and hyperspectral surface reflectance images.

[0050] It should be noted that the geochemical data to be processed and the remote sensing hyperspectral imagery corresponding to the geographic location of the geochemical data to be processed correspond to each other. Geochemical data can be obtained through a variety of channels, depending on the scale of the regional geochemical data and the target area. Applications can be made from public databases and online platforms, such as Geological Cloud, the National Regional Geochemical Exploration Database, the National Geological Data Center Portal, the EarthChem Database, and the United States Geological Survey (USGS) Data Portal. Remote sensing hyperspectral imagery of a specific area can be obtained through online platforms, such as the National Remote Sensing Data and Application Service Platform and the USGS Data Portal.

[0051] In this embodiment, the original geochemical data is of point data type, and the remote sensing hyperspectral image is of planar raster data type. Therefore, it is necessary to convert the point data type geochemical data into a raster data type through preprocessing so that the data formats of the geochemical data and the remote sensing hyperspectral image match. Preprocessing the geochemical data specifically includes: first processing zero values ​​and outliers in the geochemical data, specifically by replacing them with values ​​that are 30%-60% smaller than the corresponding detection line or the median of the corresponding neighborhood; then using a spatial interpolation method to convert the point data type geochemical data into a raster geochemical element layer, wherein the raster pixel size in the geochemical element layer is the same as the sampling spacing of the point data type geochemical data.

[0052] Specifically, for zero values ​​and outliers in geochemical data, the median of the corresponding neighborhood can be used to replace the corresponding zero value or outlier. In addition, different element testing equipment will have their own corresponding detection lines. Therefore, zero values ​​and outliers can also be replaced with values ​​far below the detection line, such as using values ​​30%-60% less than the corresponding detection line. Afterwards, spatial interpolation methods are used to convert the point-type geochemical data into a raster-shaped geochemical element layer. The interpolation result output raster pixel size is consistent with the sampling interval of the point-type geochemical data. That is, in the geochemical element layer, the raster pixel size is the same as the sampling interval of the point-type geochemical data. The coordinate system of the geochemical element layer is consistent with that of the hyperspectral image, so that the two match in geographic space.

[0053] Furthermore, spatial interpolation methods include Kriging and IDW. Kriging and IDW are both commonly used spatial interpolation methods, widely used in fields such as geographic information systems (GIS), geological exploration, environmental science, and meteorology. Kriging interpolation achieves spatial interpolation by analyzing the spatial autocorrelation of data to provide an optimal unbiased estimate of the value of an unknown point and providing an estimate error. IDW is suitable for predicting the value of an unknown point based on the values ​​of known points. Its core concept is that the closer the known point is to the predicted point, the greater its influence on the predicted value.

[0054] In this embodiment, the remote sensing hyperspectral image is preprocessed, specifically including: performing conventional processes such as radiometric calibration, atmospheric correction, and orthorectification on the remote sensing hyperspectral image according to preset satellite parameters and topographic features, and then using a fast and parameter-free hyperspectral image mixed noise removal method (FastHyMix) to perform denoising to obtain a denoised hyperspectral surface reflectance image.

[0055] It should be understood that radiometric calibration, atmospheric correction, and orthorectification are some common processes for processing remote sensing hyperspectral images. Among them, radiometric calibration is the process of converting the original digital quantization values ​​recorded by the remote sensing sensor into physically meaningful radiation quantities (such as radiation brightness, reflectivity, etc.); atmospheric correction is a key step in remote sensing image processing, which aims to eliminate the influence of atmospheric scattering, absorption and reflection, and convert the radiation signal received by the sensor into the true reflectivity or radiance of the surface to improve the quantitative accuracy of remote sensing data; orthorectification is a key step in remote sensing image processing, which aims to eliminate image deformation caused by terrain undulation and sensor geometric distortion, and generate orthorectified images with uniform scale and no distortion, so that they have map-level geometric accuracy and can be directly used for measurement and analysis.

[0056] It should be noted that, considering that the spectral resolution of hyperspectral sensors is high but the signal-to-noise ratio is low, and that images interfered with by noise are often obtained due to factors such as inconsistent component response, water vapor absorption and atmospheric effects, it is therefore very necessary to denoise the hyperspectral images. In this embodiment, a fast parameter-free hyperspectral image mixed noise removal method is used for denoising to obtain a denoised hyperspectral surface reflectance image. Among them, FastHyMix is ​​a commonly used hyperspectral image mixed noise removal method. This method uses a Gaussian mixture model to describe the complex distribution of mixed noise and utilizes two main characteristics of hyperspectral data, namely low rank in the spectral domain and high correlation in the spatial domain; the Gaussian mixture model can effectively estimate the intensity of Gaussian noise and the location of sparse noise. This method combines the low rank advantage of subspace-based representation and utilizes the spatial correlation of hyperspectral images by introducing a powerful deep image prior (extracted from a neural denoising network).

[0057] (2) The feature information of the geochemical element layer is enhanced and extracted by the local singularity analysis method to obtain a local singularity index map; the diagnostic absorption features related to the preset target chemical elements in the hyperspectral surface reflectance image are extracted to obtain a hyperspectral diagnostic absorption feature layer.

[0058] In this embodiment, the feature information of the geochemical element layer is enhanced and extracted by the local singularity analysis method to obtain the local singularity index map, which specifically includes: for the feature information in the geochemical element layer, the local singularity analysis method is used to perform local anomaly enhancement and singularity measurement of the field density in the fractal space to determine the fractal density. and the local singularity index Specifically, for each grid cell in the geochemical element layer, the fractal space with the grid cell as the is the center and the neighborhood size is Neighborhood Field value within ; Assume that the field value obeys a multifractal distribution, then ; Based on the neighborhood Field value within and the neighborhood The Euclidean space dimension and neighborhood size Calculate the neighborhood Field density on , and its calculation formula is:

[0059]

[0060] Where, Indicates the position of the grid cell in the geochemical element layer. represents the neighborhood size, Represented in grid cells A neighborhood with a center of size , Represents neighborhood Field values ​​(such as element content) within, E represents the Euclidean space dimension, Represents the grid cells in the neighborhood The fractal density at Represents the grid cells in the neighborhood The local singularity index at , Represents neighborhood The field density (such as element content) on the surface is measured; the field density is singularly measured in fractal space according to the above formula to obtain the local singularity index of each grid cell in the geochemical element layer; based on the local singularity index at each grid cell, a local singularity index map is generated and resampled to the same spatial resolution as the hyperspectral surface reflectance image.

[0061] It should be noted that for geochemical fields, the local singularity index often reflects the enrichment law of mineralized elements. The local singularity index can quantitatively characterize the enrichment and depletion of mineralized elements. E refers to the Euclidean space dimension, which is usually 1, 2, and 3, corresponding to one-dimensional space, two-dimensional space, and three-dimensional space, respectively. When , that is, the location The element abundance at is independent of the neighborhood size and has non-singularity; when When , that is, the local positive singularity increases sharply as the neighborhood scale decreases, representing the position The element enrichment at When , that is, the local negative singularity decreases sharply with the decrease of the neighborhood scale, representing the position Element depletion at different locations. Element distribution patterns vary across geographic locations, resulting in distinct singularities at different locations. The most significant difference between local singularity analysis and traditional statistical methods is that local singularity analysis focuses on a small number of anomalies that conform to fractal distributions, whereas traditional statistical methods measure the majority of non-singular data in a region. Therefore, after processing a geochemical element layer using local singularity analysis, a local singularity index map is generated.

[0062] In this embodiment, the hyperspectral surface reflectance image not only records spatial detail information but also provides dense and continuous spectral information. The diagnostic absorption characteristics of a specific spectral band are closely related to the mineral structure and composition. Therefore, the diagnostic absorption characteristics related to the preset target chemical elements in the hyperspectral surface reflectance image are quantitatively extracted. Among them, the diagnostic absorption characteristics include the deepest absorption wavelength position, deepest absorption depth, maximum absorption area and absorption spacing, etc. Figure 2 As shown. Extracting diagnostic absorption features related to preset target chemical elements in the hyperspectral surface reflectance image to obtain a hyperspectral diagnostic absorption feature layer, specifically including: first, removing the continuum of the spectrum in the local spectral interval in the hyperspectral surface reflectance image; then using a bivariate polynomial to fit multiple continuous band spectra in sequence; secondly, since it is known that the derivative of the absorption wavelength position is 0, the wavelength at the position where the derivative is 0 is obtained as the absorption wavelength position by deriving the fitting function, and the absorption depth is obtained by subtracting the fitting function value at the absorption wavelength position from 1; then, there may be multiple absorption valleys in a local spectral interval, that is, there may be multiple absorption wavelength positions and absorption depths in a local spectral interval, therefore, the deepest absorption depth is obtained by comparing multiple absorption depths in the local spectral interval, that is, multiple absorption depths. The maximum value in the degree is the deepest absorption depth, and the deepest absorption wavelength position is obtained according to the absorption wavelength position corresponding to the deepest absorption depth, that is, the absorption wavelength position corresponding to the deepest absorption depth is the deepest absorption wavelength position; secondly, the maximum absorption area is obtained according to the triangular area formed by the three spectral measured points closest to the deepest absorption wavelength position, that is, the triangular area formed by the three spectral measured points closest to the deepest absorption wavelength position is the maximum absorption area, and the absorption spacing is obtained according to the distance between the deepest absorption wavelength position and the second deepest absorption wavelength position, that is, the distance between the deepest absorption wavelength position and the second deepest absorption wavelength position is the absorption spacing; finally, the hyperspectral diagnostic absorption feature layer is obtained according to the deepest absorption depth, deepest absorption wavelength position, maximum absorption area and absorption spacing in multiple local spectral intervals in the hyperspectral surface reflectance image.

[0063] It should be understood that geochemical data typically contains data on multiple elements. Different minerals contain different combinations of elements, which in turn correspond to different spectral characteristics. For example, if the target study area is a copper mine, the spectral characteristics corresponding to the copper element should be selected. Therefore, before fusion, it is necessary to determine what the preset target chemical element is. The preset target chemical element is related to the study area and research objectives. In addition, diagnostic absorption characteristics include but are not limited to the four characteristics mentioned above. In actual use, it is necessary to select characteristics related to the preset target chemical element. Therefore, if necessary, other characteristics related to the preset target chemical element can also be selected.

[0064] (3) Based on the local singularity index at the actual geochemical sampling points and the corresponding hyperspectral diagnostic absorption characteristics, a regression equation model was constructed using the random forest model.

[0065] Specifically, based on the local singularity index map and the hyperspectral diagnostic absorption feature layer, at the position corresponding to the actual geochemical sampling point, the local singularity index of the geochemical element at the corresponding position is used as the output of the regression equation model, and the diagnostic absorption feature at the corresponding position is used as the input of the regression equation model, and the random forest model is used to construct the regression equation model.

[0066] Furthermore, the regression equation model is expressed as:

[0067]

[0068] Where, Represents the local singularity index of the geochemical elements at the corresponding position of the actual geochemical sampling point, represents the i-th diagnostic absorption feature at the corresponding position, n is the total number of diagnostic absorption features, represents the nonlinear regression function, is the correction factor.

[0069] (4) Based on the regression equation model and diagnostic absorption characteristics, the singularity index at other locations in the target area except the actual sampling point is inverted to obtain a high-resolution local singularity index map of geochemical elements in the target area.

[0070] Specifically, based on the self-similarity of the Earth system, a regression equation model between the local singularity index and the diagnostic absorption characteristics established at the actual geochemical sampling point is also applicable to other locations within the target area. This regression equation model is applied to other locations within the target area except the actual sampling point. Therefore, the regression equation model is used to invert the singularity index at other locations within the target area except the actual sampling point. That is, the diagnostic absorption characteristics at other locations within the target area except the actual sampling point are used as input to the regression equation model to obtain the singularity index at the corresponding location, thereby obtaining a high-resolution local singularity index map of geochemical elements for the entire target area. In this way, the fusion of remote sensing hyperspectral data and geochemical element layers can be achieved. The resulting high-resolution local singularity index map of geochemical elements can directly indicate the degree of enrichment and depletion of chemical elements.

[0071] It should be understood that the high-resolution geochemical element local singularity index map is obtained by fusing the original geochemical data (low resolution) with the high-resolution hyperspectral data. The entire regression process achieves the purpose of fusing the two types of data.

[0072] It should be noted that the present invention uses diagnostic absorption features in hyperspectral data that are more relevant to geochemical data to achieve data fusion. Other features that can indicate geological information to a certain extent can also be fused using similar technical processes.

[0073] For example, we first obtained the GF-5 hyperspectral data and the regional 1:200,000 scale copper (Cu) geochemical data of a porphyry copper mine. According to the method steps provided by the present invention, the pre-processed hyperspectral data and geochemical data, i.e., the hyperspectral surface reflectance image and the 1:200,000 copper element layer, are obtained. Figure 4 and Figure 5 The local singularity analysis of the copper element layer is performed to obtain the local singularity index diagram, as shown in Figure 6 As shown in the figure. According to the alteration zoning characteristics of porphyry copper deposits, the diagnostic absorption characteristics of characteristic minerals are mainly distributed in the short-wave infrared spectrum of 2100-2400nm. Therefore, the deepest absorption wavelength position, deepest absorption depth, maximum absorption area, absorption spacing and other characteristics are extracted based on the short-wave infrared spectrum of the hyperspectral surface reflectance image, as shown in the figure. Figure 7As shown, red (R) represents the deepest absorption wavelength position, green (G) represents the deepest absorption depth, and blue (B) represents the absorption spacing. At the corresponding positions of the actual sampling points of the geochemical data (a total of 55 sampling points), based on the local singularity index map and the hyperspectral diagnostic absorption feature layer, a random forest model is used to construct a regression equation model; and based on the hyperspectral diagnostic absorption feature layer, the regression equation model is used to invert the singularity index at other positions in the region to obtain a regional high-resolution copper element local singularity index inversion map, as shown below. Figure 8 The results show that the fusion of the hyperspectral data obtained by the method steps provided by the present invention and the geochemical element layer is rich in spatial detail information and is more consistent with the distribution of known mineral points.

[0074] It is worth mentioning that an embodiment of the present invention further provides a system for fusing remote sensing hyperspectral data with geochemical element layers, which is used to implement the method for fusing remote sensing hyperspectral data with geochemical element layers in the above embodiment.

[0075] like Figure 3 As shown, the system includes a data reading and preprocessing module, a feature information enhancement and extraction module, a regression function establishment module and a geochemical element local singularity layer inversion module.

[0076] In this embodiment, the data reading and preprocessing module includes a geochemical data reading module, a remote sensing hyperspectral data reading module, a geochemical data preprocessing module, and a remote sensing hyperspectral data preprocessing module. The geochemical data reading module is used to correctly read the geochemical data to be processed; the remote sensing hyperspectral data reading module is used to correctly read the remote sensing hyperspectral imagery that corresponds to the area covered by the geochemical data to be processed; the geochemical data preprocessing module is used to preprocess the geochemical data to obtain a geochemical element layer; and the remote sensing hyperspectral data preprocessing module is used to preprocess the remote sensing hyperspectral imagery to obtain a hyperspectral surface reflectance image.

[0077] In this embodiment, the feature information enhancement and extraction module is used to enhance and extract feature information of the geochemical element layer through the local singularity analysis method to obtain a local singularity index map; extract the diagnostic absorption features related to the preset target chemical elements in the hyperspectral surface reflectance image to obtain a hyperspectral diagnostic absorption feature layer.

[0078] In this embodiment, the regression function building module is used to construct a regression equation model using a random forest model based on the local singularity index at the actual geochemical sampling point and the corresponding hyperspectral diagnostic absorption characteristics.

[0079] In this embodiment, the geochemical element local singularity layer inversion module is used to invert the singularity index at other locations in the target area except the actual sampling points based on the regression equation model and the diagnostic absorption characteristics to obtain a high-resolution geochemical element local singularity index map of the entire target area.

[0080] In summary, the present invention provides a method and system for fusing remote sensing hyperspectral data with geochemical element layers. Specifically, the method comprises the following steps: first, obtaining the geochemical data to be processed and the hyperspectral data corresponding to its geographical location, preprocessing the geochemical data to form a geochemical element layer, and simultaneously preprocessing the hyperspectral image. Then, the geochemical element layer is subjected to local singularity analysis to enhance weak anomaly information, obtain a local singularity index map, and extract a hyperspectral diagnostic absorption feature layer closely related to the preset elements; a nonlinear regression relationship is established between the local singularity index at the corresponding position of the actual sampling point of the geochemical data and the hyperspectral diagnostic absorption feature, and a regression equation model is obtained. Then, the regression equation model is used to invert the hyperspectral diagnostic absorption feature to obtain a high-resolution geochemical element local singularity index layer, which can directly indicate the regional element enrichment degree.

[0081] Corresponding to the aforementioned embodiment of the method for fusing remote sensing hyperspectral data with geochemical element layers, the present invention also provides an embodiment of an electronic device.

[0082] An embodiment of the present invention provides an electronic device comprising one or more processors and a memory coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the method for fusing remote sensing hyperspectral data with geochemical element layers in the above embodiment.

[0083] The embodiment of the electronic device of the present invention can be applied to any device with data processing capabilities, and the any device with data processing capabilities can be a device such as a computer or an electronic device. The electronic device embodiment can be implemented by software, or by hardware or a combination of software and hardware. Taking software implementation as an example, as an electronic device in a logical sense, it is formed by the processor of any device with data processing capabilities in which it is located to read the corresponding computer program instructions in the non-volatile memory into the memory and run them. From the hardware level, the hardware structure of any device with data processing capabilities in which the electronic device of the present invention is located, in addition to the processor, memory, network interface, and non-volatile memory, the any device with data processing capabilities in which the electronic device in the embodiment is located can also include other hardware according to the actual function of the any device with data processing capabilities, which will not be described in detail.

[0084] The implementation process of the functions and effects of each unit in the above electronic device is specifically described in the implementation process of the corresponding steps in the above method, and will not be repeated here.

[0085] For the electronic device embodiment, since it basically corresponds to the method embodiment, the relevant parts can be referred to the partial description of the method embodiment. The electronic device embodiment described above is only illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules can be selected according to actual needs to achieve the purpose of the solution of the present invention. Those of ordinary skill in the art can understand and implement it without paying any creative work.

[0086] An embodiment of the present invention further provides a computer-readable storage medium having a program stored thereon. When the program is executed by a processor, the method for fusing remote sensing hyperspectral data with geochemical element layers in the above embodiment is implemented.

[0087] The computer-readable storage medium may be an internal storage unit of any device with data processing capabilities described in any of the aforementioned embodiments, such as a hard disk or memory. The computer-readable storage medium may also be any device with data processing capabilities, such as a plug-in hard disk, a smart media card (SMC), an SD card, a flash card, etc. equipped on the device. Furthermore, the computer-readable storage medium may also include both an internal storage unit of any device with data processing capabilities and an external storage device. The computer-readable storage medium is used to store the computer program and other programs and data required by any device with data processing capabilities, and may also be used to temporarily store data that has been output or is to be output.

[0088] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the various embodiments of the present invention.

Claims

1. A method for fusing remote sensing hyperspectral data with geochemical element layers, characterized in that: The following steps are involved: (1) Obtain the geochemical data to be processed and its corresponding remote sensing hyperspectral image, and preprocess them to obtain geochemical element layers and hyperspectral surface reflectance images; (2) Enhance and extract feature information of the geochemical element layer by using the local singularity analysis method to obtain a local singularity index map; extract diagnostic absorption features related to the preset target chemical elements in the hyperspectral surface reflectance image to obtain a hyperspectral diagnostic absorption feature layer; the diagnostic absorption features include the deepest absorption wavelength position, deepest absorption depth, maximum absorption area and absorption spacing; The step of extracting diagnostic absorption features related to preset target chemical elements from the hyperspectral surface reflectance image to obtain a hyperspectral diagnostic absorption feature layer specifically includes: First, the spectrum in the local spectral interval of the hyperspectral surface reflectance image is continuum-removed; then, a plurality of continuous band spectra are fitted in sequence using a bivariate polynomial; then, the fitting function is differentiated, and the wavelength at the position where the derivative is 0 is obtained as the absorption wavelength position, and the absorption depth is obtained by subtracting the fitting function value at the absorption wavelength position from 1; secondly, the deepest absorption depth is obtained by comparing multiple absorption depths in the local spectral interval, and the deepest absorption wavelength position is obtained according to the absorption wavelength position corresponding to the deepest absorption depth; then, the maximum absorption area is obtained according to the area of ​​the triangle formed by the three spectral measured points closest to the deepest absorption wavelength position, and the absorption spacing is obtained according to the distance between the deepest absorption wavelength position and the second deepest absorption wavelength position; finally, a hyperspectral diagnostic absorption feature layer is obtained according to the deepest absorption depth, deepest absorption wavelength position, maximum absorption area and absorption spacing in multiple local spectral intervals in the hyperspectral surface reflectance image; (3) Based on the local singularity index at the actual geochemical sampling point and the corresponding hyperspectral diagnostic absorption characteristics, a regression equation model is constructed using the random forest model; (4) Based on the regression equation model and diagnostic absorption characteristics, the singularity index at other locations in the target area except the actual sampling point is inverted to obtain a high-resolution local singularity index map of geochemical elements in the target area.

2. The method for fusing remote sensing hyperspectral data with geochemical element layers according to claim 1, characterized in that: In step (1), the geochemical data is preprocessed, specifically including: First, the zero values ​​and outliers in the geochemical data are processed; then the geochemical data of the point data type are converted into a raster geochemical element layer using the spatial interpolation method; among them, the raster pixel size in the geochemical element layer is the same as the sampling interval of the geochemical data of the point data type.

3. The method for fusing remote sensing hyperspectral data with geochemical element layers according to claim 1, characterized in that: In step (1), the remote sensing hyperspectral image is preprocessed, specifically including: According to the satellite preset parameters and terrain characteristics, the remote sensing hyperspectral image is subjected to radiometric calibration, atmospheric correction and orthorectification processing, and then denoised using the hyperspectral image mixed noise removal method to obtain the denoised hyperspectral surface reflectance image.

4. The method for fusing remote sensing hyperspectral data with geochemical element layers according to claim 1, characterized in that: The method of enhancing and extracting characteristic information of the geochemical element layer by the local singularity analysis method to obtain a local singularity index map specifically includes: For the characteristic information in the geochemical element layer, the local singularity analysis method is used to perform local anomaly enhancement and singularity measurement of the field density in the fractal space to determine the fractal density and local singularity index; that is, for each grid cell in the geochemical element layer, the field value in the neighborhood centered on the grid cell in the fractal space is calculated; assuming that the field value obeys a multifractal distribution, the field density on the neighborhood is calculated based on the field value and neighborhood size in the neighborhood and the Euclidean space dimension, and the field density is singularly measured in the fractal space to obtain the local singularity index at each grid cell in the geochemical element layer; based on the local singularity index at each grid cell, a local singularity index map is generated and resampled to the same spatial resolution as the hyperspectral surface reflectance image.

5. The method for fusing remote sensing hyperspectral data with geochemical element layers according to claim 1, characterized in that: The method uses a random forest model to construct a regression equation model based on the local singularity index and the corresponding hyperspectral diagnostic absorption characteristics at the actual geochemical sampling point, specifically including: Based on the local singularity index map and the hyperspectral diagnostic absorption feature layer, at the position corresponding to the actual geochemical sampling point, the local singularity index of the geochemical elements at the corresponding position is used as the output of the regression equation model, and the diagnostic absorption feature at the corresponding position is used as the input of the regression equation model. The random forest model is used to construct the regression equation model.

6. The method for fusing remote sensing hyperspectral data with geochemical element layers according to claim 1, characterized in that: The inversion of the singularity index at other locations in the target area except the actual sampling point based on the regression equation model and the diagnostic absorption characteristics to obtain a high-resolution local singularity index map of geochemical elements in the target area specifically includes: According to the self-similarity of the Earth system, a regression equation model between the local singularity index and the diagnostic absorption characteristics established at the actual geochemical sampling point is used to invert the singularity index at other positions in the target area except the actual sampling points. That is, the diagnostic absorption characteristics at other positions in the target area except the actual sampling points are used as the input of the regression equation model to obtain the singularity index at the corresponding position, so as to obtain a high-resolution local singularity index map of geochemical elements in the target area as a whole.

7. A system for implementing the method for fusing remote sensing hyperspectral data with geochemical element layers according to any one of claims 1 to 6, characterized in that: include: The data reading and preprocessing module is used to obtain the geochemical data to be processed and the remote sensing hyperspectral image consistent with the coverage area of ​​the geochemical data to be processed, and preprocess them to obtain the geochemical element layer and the hyperspectral surface reflectance image; The feature information enhancement and extraction module is used to enhance and extract feature information of the geochemical element layer through the local singularity analysis method to obtain a local singularity index map; extract diagnostic absorption features related to preset target chemical elements in the hyperspectral surface reflectance image to obtain a hyperspectral diagnostic absorption feature layer; A regression function building module is used to construct a regression equation model using a random forest model based on the local singularity index and the corresponding hyperspectral diagnostic absorption characteristics at the actual geochemical sampling points; The geochemical element local singularity layer inversion module is used to invert the singularity index at locations other than the actual sampling points in the target area based on the regression equation model and diagnostic absorption characteristics, so as to obtain a high-resolution geochemical element local singularity index map for the entire target area.

8. An electronic device comprising a memory and a processor, characterized in that: The memory is coupled to the processor; wherein the memory is used to store program data, and the processor is used to execute the program data to implement the method for fusing remote sensing hyperspectral data with geochemical element layers according to any one of claims 1 to 6.

Images