Method, device, equipment and medium for extracting altered mineral information

Abstract

The present invention discloses a method, device, equipment and medium for extracting altered mineral information. The method for extracting altered mineral information comprises: determining the potential distribution area of ​​the target altered mineral in the study area according to the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area; gridding the potential distribution area and determining the attribute information of each pixel in the grid unit, wherein the attribute information at least includes the spectral characteristics, topographic characteristics, geochemical characteristics and key band characteristics corresponding to the pixel; performing cluster analysis on the pixels in each grid unit respectively, and determining the sample pixels of the potential distribution area based on the cluster analysis results; training an altered mineral information extraction model using the sample attribute information of the sample pixels to obtain the trained altered mineral information extraction model; and inputting the attribute information of each pixel in the study area into the altered mineral information extraction model to obtain the altered mineral distribution information of the study area.

Description

Technical Field

[0001] The present invention relates to the technical field of remote sensing geological survey, and in particular to a method, device, equipment and medium for extracting altered mineral information. Background Art

[0002] The spatial distribution of altered minerals is closely related to the mineralization process. Therefore, in mineral resource exploration, accurately identifying and extracting altered mineral information is of great significance for ore deposit location and resource assessment. With the development of remote sensing technology, remote sensing imagery provides rich spectral information and can obtain multimodal data such as spatial characteristics, topographic information, and geochemical information. These data can reveal the characteristics of surface minerals from multiple dimensions. However, traditional altered mineral information extraction methods generally rely on spectral characteristics and pixel-by-pixel analysis to identify altered minerals. However, this method ignores the inherent correlation between the geographic spatial characteristics of pixels and fails to fully consider the spatial similarity and distribution patterns of altered minerals within a specific area.

[0003] Therefore, how to effectively integrate regional spatial, spectral, topographic and geochemical multimodal data, give full play to their complementary advantages, and improve the accuracy and efficiency of alteration mineral information extraction is a technical problem that technical personnel in this field urgently need to solve. Summary of the Invention

[0004] The purpose of the present invention is to provide a method, device, equipment and medium for extracting altered mineral information, so as to effectively integrate regional spatial, spectral, topographic and geochemical multimodal data, give full play to their complementary advantages, and improve the accuracy and efficiency of altered mineral information extraction.

[0005] In order to achieve the above object, the present invention provides the following technical solutions:

[0006] A method for extracting alteration mineral information comprises: determining the potential distribution area of ​​the target altered mineral in the study area based on the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area; gridding the potential distribution area and determining the attribute information of each pixel in the grid unit of the potential distribution area; wherein the attribute information at least includes a first spectral characteristic, a first topographic characteristic, a first geochemical characteristic corresponding to the pixel, and a first key band characteristic related to the target altered mineral; performing cluster analysis on the pixels of each of the grid units respectively, and determining sample pixels of the potential distribution area based on the cluster analysis results; determining the sample attribute information of the sample pixels as training samples, training a pre-constructed alteration mineral information extraction model, and obtaining a trained alteration mineral information extraction model; inputting the second spectral characteristic, the second topographic characteristic, the second geochemical characteristic, and the second key band characteristic related to the target altered mineral of each pixel in the study area into the trained alteration mineral information extraction model to obtain the alteration mineral distribution information of the study area.

[0007] In an optional embodiment of the present application, the potential distribution area of ​​the target altered mineral in the study area is determined based on the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area, including: determining the key band related to the target altered mineral, and extracting the band information of the multispectral or hyperspectral satellite image data based on the key band to obtain the key band raster data of the study area, wherein each key band raster data corresponds to each key band information; analyzing the key band raster data of the study area by principal component analysis or band ratio to obtain the indicative raster data of the target altered mineral; using Gaussian filtering , performing filtering analysis on the indicative grid data to obtain the filtered indicative grid data; performing threshold segmentation on the filtered indicative grid data based on a preset indicative grid data threshold to determine the potential distribution area; or, through a spectral angle matching method, analyzing the spectrum of the multispectral or hyperspectral satellite image data according to the standard spectral characteristics of the target altered mineral to obtain the indicative spectral angle index of the target altered mineral; using Gaussian filtering, performing filtering analysis on the indicative spectral angle index to obtain the filtered indicative spectral angle index; performing threshold segmentation on the filtered indicative spectral angle index based on a preset indicative spectral angle index threshold to determine the potential distribution area.

[0008] In an optional embodiment of the present application, the potential distribution area of ​​the target altered mineral in the study area is determined based on the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area, including: determining the key band related to the target altered mineral, and extracting the band information of the multispectral or hyperspectral satellite image data based on the key band to obtain the key band raster data of the study area, wherein each key band raster data corresponds to each key band information; analyzing the key band raster data of the study area by principal component analysis to determine the first indicative raster data of the target altered mineral; filtering and analyzing the first indicative raster data by Gaussian filtering to obtain the filtered first indicative raster data; performing threshold segmentation on the filtered first indicative raster data based on a preset first indicative raster data threshold to obtain the first potential distribution area of ​​the target altered mineral; and analyzing the key bands of the study area by band ratio. The method comprises the following steps: analyzing the key band raster data to determine the second indicative raster data of the target altered mineral; filtering and analyzing the second indicative raster data by Gaussian filtering to obtain the filtered second indicative raster data; performing threshold segmentation on the filtered second indicative raster data based on a preset second indicative raster data threshold to obtain a second potential distribution area of ​​the target altered mineral; analyzing the spectrum of the multispectral or hyperspectral satellite image data by spectral angle matching according to the standard spectral characteristics of the target altered mineral to obtain the indicative spectral angle index of the target altered mineral; filtering and analyzing the indicative spectral angle index by Gaussian filtering to obtain the filtered indicative spectral angle index; performing threshold segmentation on the filtered indicative spectral angle index based on a preset indicative spectral angle index threshold to determine the third potential distribution area of ​​the target altered mineral; and determining the overlapping area of ​​the first potential distribution area, the second potential distribution area, and the third potential distribution area as the potential distribution area.

[0009] In an optional embodiment of the present application, cluster analysis is performed on the pixels of each of the grid units respectively, and sample pixels of the potential distribution area are determined based on the cluster analysis results, including: for any of the grid units, randomly selecting k pixels to be determined as the initial cluster center points of the grid unit; wherein k is a positive integer; based on the distance between each pixel in the grid unit and the initial cluster center point, the initial cluster of the grid unit is determined; for any initial cluster, based on the distance between each pixel in the initial cluster and the initial representative pixel, the initial representative pixel is updated to determine the updated representative pixel in the initial cluster; the representative pixel and other pixels within a preset neighborhood range of the representative pixel are determined as sample pixels.

[0010] In an optional embodiment of the present application, the distance between each pixel in the grid unit and the initial cluster center point, and the distance between each pixel in the initial cluster and the initial representative pixel are determined by the following formula:

[0011] ;

[0012] ;

[0013] ;

[0014] in, Represents the distance between pixel x and the initial representative pixel c or the initial cluster center c; Represents the attribute distance between pixel x and the initial representative pixel c or the initial cluster center c; Represents the spatial distance between pixel x and the initial representative pixel c or the initial cluster center c; represents the weight of the attribute distance; represents the weight of the spatial distance; Represents the i-th attribute information of pixel x; represents the i-th attribute information of the initial representative pixel c or the initial cluster center c; n represents the number of attribute information of pixels; Represents the distance weight of the i-th attribute of the pixel; Represents the projection value of pixel x on the jth axis; represents the projection value of the initial representative pixel c or the initial cluster center c on the j-th axis; m represents the dimension of the spatial coordinate system, m is equal to 2, and the first axis and the second axis of the spatial coordinate system are perpendicular to each other.

[0015] In an optional embodiment of the present application, the sample attribute information of the sample pixel is determined as a training sample, and a pre-constructed altered mineral information extraction model is trained to obtain a trained altered mineral information extraction model, including: inputting the sample attribute information of each of the sample pixels into the pre-constructed altered mineral information extraction model, so that the altered mineral information extraction model combines the multi-head self-attention mechanism and the sample attribute information to determine the predicted distribution probability of the target altered mineral corresponding to the geographical location of each of the sample pixels; determining the loss function of the pre-constructed altered mineral information extraction model based on the predicted distribution probability of the target altered mineral, and training the altered mineral information extraction model according to the loss function to obtain the trained altered mineral information extraction model.

[0016] In an optional embodiment of the present application, the loss function of the altered mineral information extraction model is expressed by the following formula:

[0017] ;

[0018] Wherein, L represents the loss function of the altered mineral information extraction model; the value of y indicates whether the pixel block of the multispectral or hyperspectral satellite image data is the target altered mineral. When y is equal to 1, it indicates that the pixel block is the target altered mineral; when y is equal to 0, it indicates that the pixel block is a non-altered mineral; p represents the predicted probability that the pixel block is the target altered mineral.

[0019] Compared with the existing technology, the method for extracting altered mineral information provided by the present invention determines the sample pixels in the potential distribution area by gridding the potential distribution area of ​​the target altered mineral in the study area, and performs cluster analysis on the pixels in each grid unit to construct sample attribute information as training samples based on the spectral characteristics, topographic characteristics, geochemical characteristics and key band characteristics in each sample pixel to train the altered mineral information extraction model, so that the altered mineral information extraction model can effectively integrate the multi-modal vectors such as space, spectrum, topography and geochemistry of the potential distribution area, give full play to their complementary advantages, and thereby improve the accuracy and efficiency of the altered mineral information extraction model in predicting target altered minerals.

[0020] The present invention also provides a device for extracting altered mineral information, comprising:

[0021] The potential distribution area determination unit is used to determine the potential distribution area of ​​the target altered mineral in the study area based on the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area.

[0022] A grid division unit is used to grid the potential distribution area and determine the attribute information of each pixel in the grid unit of the potential distribution area; wherein the attribute information includes at least a first spectral feature, a first topographic feature, a first geochemical feature corresponding to the pixel, and a first key band feature related to the target altered mineral.

[0023] The cluster analysis unit is used to perform cluster analysis on the pixels of each grid unit respectively, and determine the sample pixels of the potential distribution area based on the cluster analysis results.

[0024] The model training unit is used to determine the sample attribute information of the sample pixel as a training sample, train the pre-built alteration mineral information extraction model, and obtain a trained alteration mineral information extraction model.

[0025] The altered mineral prediction unit is used to input the second spectral characteristics, second topographic characteristics, second geochemical characteristics of each pixel in the study area and the second key band characteristics related to the target altered mineral into the trained altered mineral information extraction model to obtain the altered mineral distribution information of the study area.

[0026] Compared with the prior art, the beneficial effects of the altered mineral information extraction device provided by the present invention are the same as the beneficial effects described in the technical solution of the above-mentioned altered mineral information extraction method, and will not be repeated here.

[0027] The present invention further provides an electronic device, comprising:

[0028] processor;

[0029] a memory for storing instructions executable by the processor;

[0030] The processor is used to execute the above-mentioned method for extracting altered mineral information by running the instructions in the memory.

[0031] Compared with the prior art, the beneficial effects of the electronic device provided by the present invention are the same as the beneficial effects of the method for extracting altered mineral information described in the above technical solution, and will not be described in detail here.

[0032] The present invention also provides a computer storage medium, in which instructions are stored. When the instructions are executed, the above-mentioned method for extracting altered mineral information is implemented.

[0033] Compared with the prior art, the beneficial effects of the computer storage medium provided by the present invention are the same as the beneficial effects of the method for extracting altered mineral information described in the above technical solution, and will not be described in detail here. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The drawings described herein are used to provide a further understanding of the present invention and constitute a part of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute an improper limitation of the present invention. In the drawings:

[0035] Figure 1 Flowchart of the method for extracting altered mineral information provided in the embodiment of this application.

[0036] Figure 2 Structural diagram of the device for extracting altered mineral information provided in an embodiment of the present application.

[0037] Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0038] To facilitate a clear description of the technical solutions of the embodiments of the present invention, the words "first" and "second" are used in the embodiments of the present invention to distinguish between identical or similar items with substantially the same functions and effects. For example, the first threshold and the second threshold are merely used to distinguish between different thresholds and do not limit their order. Those skilled in the art will understand that the words "first" and "second" do not limit the quantity or execution order, and the words "first" and "second" do not necessarily mean different.

[0039] It should be noted that, in the present invention, words such as "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in the present invention should not be construed as being preferred or advantageous over other embodiments or designs. Rather, the use of words such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0040] In the present invention, "at least one" means one or more, and "more" means two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships may exist. For example, A and / or B can mean: the existence of A alone, the existence of A and B at the same time, and the existence of B alone, where A and B can be singular or plural. The character " / " generally indicates that the previous and next associated objects are in an "or" relationship. "At least one of the following items" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b or c can mean: a, b, c, the combination of a and b, the combination of a and c, the combination of b and c, or the combination of a, b and c, where a, b, c can be single or plural.

[0041] The spatial distribution of altered minerals is closely related to the mineralization process. Therefore, in mineral resource exploration, accurately identifying and extracting altered mineral information is of great significance for ore deposit location and resource assessment. With the development of remote sensing technology, remote sensing imagery provides rich spectral information and can obtain multimodal data such as spatial characteristics, topographic information, and geochemical information. These data can reveal the characteristics of surface minerals from multiple dimensions. However, traditional altered mineral information extraction methods generally rely on spectral characteristics and pixel-by-pixel analysis to identify altered minerals. However, this method ignores the inherent correlation between the geographic spatial characteristics of pixels and fails to fully consider the spatial similarity and distribution patterns of altered minerals within a specific area.

[0042] Therefore, how to effectively integrate regional spatial, spectral, topographic and geochemical multimodal data, give full play to their complementary advantages, and improve the accuracy and efficiency of alteration mineral information extraction is a technical problem that technical personnel in this field urgently need to solve.

[0043] In order to solve the above technical problems, the present application provides a method, device, equipment and medium for extracting altered mineral information, which are described in detail one by one in the following embodiments.

[0044] Please refer to Figure 1 , Figure 1 Flowchart of the method for extracting altered mineral information provided in the embodiment of this application.

[0045] like Figure 1 As shown, the method for extracting altered mineral information includes the following steps S101 to S105:

[0046] S101, determining the potential distribution area of ​​the target altered mineral in the study area based on the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area.

[0047] Altered minerals are secondary minerals formed after primary minerals undergo chemical changes due to geological processes (such as hydrothermal activity, weathering, etc.). Their spectral characteristics can be regarded as the "fingerprint" of the altered minerals and used for their identification.

[0048] The standard spectral characteristics refer to the consistent and repeatable spectral response pattern exhibited by a certain mineral under its typical conditions. In practical applications, the standard spectral characteristics are usually based on laboratory measurements or extracted from an existing spectral database.

[0049] The purpose of the above S101 is to identify the target altered minerals in the study area and determine the potential distribution area where the target altered minerals may exist.

[0050] In practical applications, the above S101 can be implemented through methods such as principal component analysis (PCA), band ratio method (BR), and spectral angle mapping method (SAM).

[0051] Specifically, in an optional implementation manner of the present application, the above S101 includes the following S11 to S13:

[0052] S11, determining the key bands related to the target altered mineral, and extracting band information from the multispectral or hyperspectral satellite image data based on the key bands to obtain key band raster data of the study area, wherein each key band raster data corresponds to each key band information.

[0053] Key band raster data refers to satellite image data within different wavelength ranges obtained through the multispectral or hyperspectral satellite image data. In the embodiment of the present application, the key band raster data can be obtained by analyzing the standard spectrum curve of the target altered mineral to extract the key spectral features therein, so as to select sensitive bands that can reflect the characteristics of the altered mineral. The analysis of the standard spectrum curve includes but is not limited to the following aspects:

[0054] The position of the characteristic absorption valley in the standard spectrum curve is identified, and the central wavelength of the absorption valley is measured (wherein the central wavelength of the absorption valley is used to indicate the characteristic spectral response of the mineral), thereby clarifying the spectral range associated with the target altered mineral.

[0055] Quantify the depth of the spectral absorption valley (the depth of the spectral absorption valley is used to reflect the response intensity of the mineral in the spectrum to evaluate the significance of the mineral presence and the sensitivity of the band to altered minerals), and calculate the change amplitude of its relative reflectivity or absorbance.

[0056] The width of the spectral absorption valley at half height is calculated to characterize the width of the absorption valley. The half-height width parameter can assist in determining the characteristics of the absorption valley, thereby providing support for the precise definition of the sensitive band.

[0057] S12, analyzing the key grid data by principal component analysis or band ratio method to determine the indicative grid data of the target altered mineral.

[0058] Principal component analysis is used to reduce redundant information in band raster data to extract indicative raster data related to the target altered minerals.

[0059] Band ratio analysis is used to enhance spectral features associated with the target alteration minerals while suppressing background noise and other interfering factors.

[0060] S13, using Gaussian filtering to perform filtering analysis on the indicative grid data to obtain the filtered indicative grid data; based on a preset indicative grid data threshold, performing threshold segmentation on the filtered indicative grid data to determine the potential distribution area.

[0061] The above S13 refers to the use of adaptive Gaussian filtering to dynamically adjust the size of the filtering window according to the local distribution characteristics of the indicative grid data, and then perform filtering analysis on the indicative grid data of the study area, and then combine the preset indicative grid data threshold to segment the filtered indicative grid data to determine the potential distribution area.

[0062] In another optional implementation manner of the present application, the above S101 may also be implemented by the following S21 to S22:

[0063] S21, analyzing the spectrum of the multispectral or hyperspectral satellite image data according to the standard spectral characteristics of the target altered mineral by a spectral angle matching method to obtain an indicative spectral angle index of the target altered mineral.

[0064] S22, using Gaussian filtering to perform filtering analysis on the indicative spectral angle index to obtain the filtered indicative spectral angle index; based on a preset indicative spectral angle index threshold, performing threshold segmentation on the filtered indicative spectral angle index to determine the potential distribution area.

[0065] Spectral angle matching is a spectral matching algorithm that determines similarity by calculating the angle between the spectrum of an unknown sample and a known standard spectrum. The smaller the angle, the closer the two are, indicating the presence of the same or similar minerals.

[0066] The above S21 and S22 refer to obtaining the indicative spectral angle index that best reflects the target altered mineral through the spectral angle matching method, and then using the adaptive Gaussian filtering method to dynamically adjust the size of the filter window according to the local distribution characteristics of the indicative spectral angle index data, and then perform filtering analysis on the study area, and then combine the preset indicative spectral angle index threshold to perform threshold segmentation on the filtered indicative spectral angle index to determine the potential distribution area.

[0067] It can be understood that in the actual application process, different methods are used to determine the potential distribution area of ​​alteration minerals from different dimensions through the above-mentioned principal component analysis, band ratio and spectral angle matching. In an optional embodiment of the present application, the potential distribution area can also be determined in combination with the above three methods to effectively reduce the uncertainty that may be introduced by determining the potential distribution area through a single method.

[0068] That is, the first potential distribution area, the second potential distribution area, and the third potential distribution area are determined using the principal component analysis, the band ratio method, and the spectral angle matching method. The intersection of the first potential distribution area, the second potential distribution area, and the third potential distribution area is then determined as the final potential distribution area.

[0069] In another optional embodiment of the present application, in order to further improve the accuracy of the potential distribution area and eliminate areas in these areas such as cities, roads, and buildings that are significantly affected by human activities, high-resolution land use data can also be combined to identify artificial structures and non-geological anomaly areas in the potential distribution area; and artificial structures and non-geological anomaly areas in the potential distribution area are removed by mask elimination.

[0070] S102, gridding the potential distribution area and determining the attribute information of each pixel in the grid unit of the potential distribution area; wherein the attribute information includes at least a first spectral feature, a first topographic feature, a first geochemical feature corresponding to the pixel, and a first key band feature related to the target altered mineral.

[0071] The above S102 means that, in the potential distribution area, a fixed grid division strategy is adopted to divide the potential distribution area into multiple grid units (for example, the potential distribution area is divided into grid units of 256×256 pixels), and attribute information of each pixel in the grid unit is determined.

[0072] Among them, each pixel in the grid unit corresponds to the geographical location of the pixel, and each pixel includes a series of numerical values, which are the attribute information mentioned in S102 above, that is, the attribute information includes the spectral characteristics, topographic characteristics, geochemical characteristics corresponding to the geographical location of the pixel, and key band characteristics related to the target altered minerals.

[0073] The spectral features can be determined based on the reflectivity information of each band in the multispectral or hyperspectral satellite image data; the terrain features can be derived terrain features such as slope and aspect obtained based on a digital elevation model (DEM); the geochemical features can be obtained through field collection, remote sensing inversion, and other means, including: major and trace element concentration change data, geochemical background information on mineral genesis, etc.; the key spectral features related to the target altered mineral can be obtained through analysis of the standard spectral curve of the target altered mineral mentioned in S101 above. In addition, in order to improve the relationship between attribute information and the environment, the key spectral features can also include band combination indicators such as vegetation index in addition to parameters such as absorption valley characteristics and reflection peak characteristics.

[0074] In addition, in the actual application process, in order to facilitate the training of the alteration mineral information extraction model used to predict alteration minerals through the attribute information of the pixel, after extracting the spectral characteristics, topographic characteristics, geochemical characteristics and key band characteristics, it is also necessary to ensure the consistency of these feature data in spatial resolution.

[0075] Specifically, the resolution of the spectral features, topographic features, and geochemical features can be adjusted to be consistent with the multispectral or hyperspectral satellite image data by resampling, and then these features can be fused according to the channel to generate a multimodal feature tensor: X, where , C represents the number of feature channels, that is, the total number of spectral features, topographic features, geochemical features, and key band features; H and W represent the spatial dimensions of the features, that is, the number of rows and columns.

[0076] S103 , performing cluster analysis on the pixels of each grid unit respectively, and determining the sample pixels of the potential distribution area based on the cluster analysis results.

[0077] In order to further determine the intrinsic correlation between the spatial characteristics of the target altered minerals in the study area and the topography, geochemistry, and spectrum, this application performs cluster analysis on the pixels in the grid cells through the above S103.

[0078] Specifically, performing cluster analysis on the pixels of each grid unit and determining the sample pixels of the potential distribution area based on the cluster analysis results includes the following steps S31 to S34:

[0079] S31 : For any of the grid cells, randomly select k pixels to be determined as initial cluster center points of the grid cell; wherein k is a positive integer.

[0080] S32: Determine an initial cluster of the grid unit according to the distance between each pixel in the grid unit and the initial cluster center point.

[0081] S33 , for any initial cluster, updating the initial representative pixel according to the distance between each pixel in the initial cluster and the initial representative pixel, and determining an updated representative pixel in the initial cluster.

[0082] Specifically, the distance between each pixel in the grid unit and the initial cluster center point, and the distance between each pixel in the initial cluster and the initial representative pixel are determined by the following formulas (1) to (3):

[0083] (1);

[0084] (2);

[0085] (3);

[0086] in, Represents the distance between pixel x and the initial representative pixel c or the initial cluster center c; Represents the attribute distance between pixel x and the initial representative pixel c or the initial cluster center c; Represents the spatial distance between pixel x and the initial representative pixel c or the initial cluster center c; represents the weight of the attribute distance; represents the weight of the spatial distance; Represents the i-th attribute information of pixel x; represents the i-th attribute information of the initial representative pixel c or the initial cluster center c; n represents the number of attribute information of pixels; Represents the distance weight of the i-th attribute of the pixel; Represents the projection value of pixel x on the jth axis; represents the projection value of the initial representative pixel c or the initial cluster center c on the j-th axis; m represents the dimension of the spatial coordinate system, m is equal to 2, and the first axis and the second axis of the spatial coordinate system are perpendicular to each other.

[0087] Furthermore, the weight of spatial distance Weight of attribute distance , which is used to adjust the contribution of attribute information and spatial information to the differences between different pixels.

[0088] In the actual application process, and The search range can be tested gradually by grid search for all possible weight combinations. and , calculate its overall silhouette coefficient As the evaluation index of clustering quality, we choose The largest group and As the final result.

[0089] in, ,and The closer the value is to 1, the better the clustering effect is and the higher the density between different pixels is. The closer the value is to -1, the worse the clustering effect is, and there are pixels in the cluster that are not suitable for the cluster.

[0090] Furthermore, the overall silhouette coefficient It can be determined by the following formula (4):

[0091] (4);

[0092] in, represents the overall silhouette coefficient; Represents the silhouette coefficient of the i-th pixel; Represents the average distance between the i-th pixel and other pixels in the cluster; Represents the average distance between the i-th pixel and other clusters; The smaller it is, the closer the i-th pixel is to the cluster center, and the tighter the cluster is. The larger the value, the more obvious the separation of the i-th pixel from other clusters.

[0093] In actual application, in order to achieve clustering of pixels of grid units, the above-mentioned process from S32 to S34 can be repeatedly performed until the cluster center (ie, the representative pixel) no longer changes or the number of iterations reaches the iteration threshold.

[0094] S34: Determine the representative pixel and other pixels within a preset neighborhood of the representative pixel as sample pixels.

[0095] It can be understood that in the potential distribution area of ​​altered minerals, for a pixel with altered minerals, other pixels adjacent to the pixel also have corresponding altered minerals. The closer the pixels are, the more similar the corresponding attribute information of these pixels is.

[0096] On this basis, in order to construct the spatial correlation between the representative pixel and other pixels in the cluster, other pixels within the preset neighborhood range of the representative pixel (such as 32×32 pixel range) can be selected as sample pixels, thereby ensuring the spatial uniformity of the sample pixel distribution, while fully reflecting the representative characteristics of the representative pixel and avoiding deviations caused by the randomness of a single pixel.

[0097] In addition, for cluster centers close to the grid unit boundary, other pixels can be determined by mirror filling or expanding the preset neighborhood range, thereby avoiding the situation where the preset neighborhood range is incomplete due to boundary restrictions.

[0098] S104: Determine the sample attribute information of the sample pixel as a training sample, train a pre-built alteration mineral information extraction model, and obtain a trained alteration mineral information extraction model.

[0099] The altered mineral information extraction model can be understood as a deep learning model. In the embodiment of the present application, the altered mineral information extraction model is built based on the Transformer architecture of the multi-head self-attention mechanism. The model realizes effective modeling of sequence data by introducing the self-attention mechanism, the multi-head attention mechanism and a series of auxiliary technologies (such as position encoding, layer normalization, etc.), which is conducive to improving the ability to process long sequence data. It also has the characteristics of high parallelization, which is conducive to improving the training efficiency of the model.

[0100] In the embodiment of the present application, the geographical location corresponding to the sample pixel can be understood as the geographical location where altered minerals exist. The training of the pre-built altered mineral information extraction model includes:

[0101] The sample attribute information of each of the sample pixels is input into the pre-constructed altered mineral information extraction model, so that the altered mineral information extraction model combines the multi-head self-attention mechanism and the sample attribute information to determine the predicted distribution probability of the target altered mineral in the geographical location corresponding to each of the sample pixels; the loss function of the pre-constructed altered mineral information extraction model is determined based on the predicted distribution probability of the target altered mineral, and the altered mineral information extraction model is trained according to the loss function to obtain the trained altered mineral information extraction model.

[0102] In actual application, the process of training the pre-built alteration mineral information extraction model based on the training samples is divided into two processes: encoding and decoding.

[0103] First, before encoding and decoding, each sample pixel input into the altered mineral information extraction model must be divided into blocks; for the sample pixels composed of the representative pixels and other pixels within the preset neighborhood of the representative pixels, first, they are divided into image blocks of size p×p, where p is a positive integer, and the size of p can be set based on actual conditions, such as an image block of size 8×8.

[0104] The total number of blocks can be expressed by the following formula (5):

[0105] (5);

[0106] Where N represents the number of image blocks; H and W represent the spatial dimensions of the sample attribute information of the sample pixel. Furthermore, each image block is a vector of fixed length after being flattened, and its length is equal to , where C represents the number of feature channels in the sample attribute information.

[0107] Assume that the number of attention heads of the altered mineral information extraction model is , then the length of the sample attribute information input by each attention head is .

[0108] During the encoding process, first, the query vector (Q-Query), key vector (K-Key), and value vector (V-Value) of the input features are calculated for each attention head.

[0109] Among them, the query vector , key vector , value vector ,in, , , is the projection matrix of each head, with dimensions ; X represents the input data.

[0110] Afterwards, the similarity matrix A is calculated, where the similarity matrix A can be expressed by the following formula (6):

[0111] (6);

[0112] in, is a scaling factor used to balance the gradient; the Softmax function is used to normalize the dot product results to obtain the final similarity matrix, which represents the relative importance or similarity between each query and each key.

[0113] Furthermore, an attention weighted value is calculated based on the similarity matrix, and the attention weighted value is expressed by the following formula (7):

[0114] (7);

[0115] in, represents the attention weight value; A represents the similarity matrix; V represents the value vector.

[0116] After obtaining the attention weights, the results of each attention head are concatenated and a linear transformation is applied to form the input of the next layer:

[0117] (8);

[0118] in, represents the input of the next layer; h represents the h-th attention head; Represents a linear transformation matrix, which is used to adjust the dimension of the input result to be consistent with the input data.

[0119] After multi-head attention calculation, the features of each pixel in the image block are independently processed by the feedforward neural network (FFN). Specifically, the feedforward neural network processing the features of each pixel can be implemented by the following formula (9):

[0120] (9);

[0121] in, Represents the features of the feedforward neural network pixel x; and is the weight matrix; and is a bias term; in practical applications, the ReLU activation function can be used to ensure nonlinear feature extraction.

[0122] For each layer of multi-head attention and feedforward neural network, the altered mineral information extraction model adds the input to the output through residual connection, and then obtains the result and performs layer normalization (LayerNorm) to stabilize the training and avoid gradient disappearance or explosion. Specifically, the normalization process can be implemented by the following formulas (10) and (11):

[0123] (10);

[0124] (11);

[0125] in, Represents the input of a multi-head attention mechanism or feedforward neural network; and Indicates output; Representation layer normalization operation; and Represent the outputs after residual connection and layer normalization operations respectively.

[0126] Furthermore, during the encoding process, after processing through multiple encoding layers, the altered mineral information extraction model will obtain the feature representation of each pixel block, and finally perform classification through an output layer using a Sigmoid activation function to obtain the predicted distribution probability of the target altered mineral.

[0127] Specifically, the predicted distribution probability can be expressed by the following formula (12):

[0128] (12);

[0129] in, A feature vector representing a pixel block; represents the weight matrix; represents the bias term.

[0130] Further, The activation function formula can be expressed by the following formula (13):

[0131] (13);

[0132] in, The activation function is used to map the output value to (0, 1) to represent the probability that the content in each pixel block is an altered mineral.

[0133] Furthermore, after obtaining the predicted distribution probability of the target altered mineral output by the pre-constructed altered mineral information extraction model after identifying the sample attribute information, the corresponding loss function can be constructed in combination with the predicted distribution probability, and the altered mineral information extraction model can be trained with the goal of minimizing the loss value of the loss function.

[0134] Specifically, the loss function of the altered mineral information extraction model can be expressed by the following formula (14):

[0135] (14);

[0136] Wherein, L represents the loss function of the altered mineral information extraction model; the value of y indicates whether the pixel block of the multispectral or hyperspectral satellite image data is the target altered mineral. When y is equal to 1, it indicates that the pixel block is the target altered mineral; when y is equal to 0, it indicates that the pixel block is a non-altered mineral; p represents the predicted probability that the pixel block is the target altered mineral.

[0137] That is to say, in actual application, the input data of the altered mineral information extraction model is the attribute information of the study area, and the output data of the altered mineral information extraction model is the probability that a pixel block composed of multiple adjacent pixels is the target altered mineral.

[0138] In an optional embodiment of the present application, the performance of the altered mineral information extraction model may be evaluated in combination with the predicted probability that each pixel block output by the altered mineral information extraction model is an altered mineral.

[0139] Specifically, for each pixel block, assuming that the predicted probability value output by the altered mineral information extraction model is P, when P is greater than or equal to the preset probability threshold T, the pixel block is determined to correspond to the target altered mineral. When P is less than the preset probability threshold T, the non-target altered mineral corresponding to the pixel block is determined, and the matching situation between the result of the threshold division and whether the pixel block is actually the target altered mineral is statistically analyzed, and then a confusion matrix is ​​constructed.

[0140] Please refer to Table 1, which is a schematic diagram of the confusion matrix provided in an embodiment of the present application.

[0141] Table 1:

[0142] P is greater than T P is less than T Actually the target alteration mineral (y=1) True Positive (TP) False Negative (FN) Actually non-target alteration mineral (y=0) False Positive (FP) True Negative (TN)

[0143] Among them, TP represents the number of pixel blocks predicted by the altered mineral information extraction model as target altered minerals and predicted correctly; FN represents the number of pixel blocks predicted by the altered mineral information extraction model as non-target altered minerals and predicted incorrectly; FP represents the number of pixel blocks predicted by the altered mineral information extraction model as target altered minerals and predicted incorrectly; TN represents the number of pixel blocks predicted by the altered mineral information extraction model as non-target altered minerals and predicted correctly.

[0144] Furthermore, based on the results obtained in Table 1 above, the trained alteration mineral information extraction model was evaluated, and the evaluation items included accuracy, precision, recall rate, and F1 score.

[0145] The accuracy rate is used to reflect the overall prediction accuracy of the alteration mineral information extraction model, and the accuracy rate can be calculated using the following formula (15):

[0146] (15);

[0147] in, Indicates the accuracy rate.

[0148] Accuracy is used to indicate the correct proportion of the predicted target alteration minerals. The accuracy can be calculated using the following formula (16):

[0149] (16);

[0150] in, It represents the accuracy, and the higher the accuracy, the more reliable the prediction result of the alteration mineral information extraction model.

[0151] The recall rate represents the proportion of target altered minerals that are correctly identified. The recall rate can be calculated using the following formula (17):

[0152] (17);

[0153] in, It represents the recall rate. A higher recall rate indicates a higher probability that the alteration mineral information extraction model captures most of the target alteration minerals.

[0154] The F1 score represents the harmonic mean of the precision and the recall, and is used to comprehensively evaluate the performance of the alteration mineral information extraction model. The F1 score can be calculated using the following formula (18):

[0155] (18);

[0156] Here, F1 represents the F1 score.

[0157] S105, inputting the second spectral characteristics, second topographic characteristics, second geochemical characteristics and second key band characteristics related to the target altered mineral of each pixel in the study area into the trained altered mineral information extraction model to obtain the altered mineral distribution information of the study area.

[0158] That is, after completing the training of the altered mineral information extraction model, the altered mineral information extraction model can be applied to the entire study area, that is, the spectral characteristics, topographic characteristics, geochemical characteristics of each pixel in the study area and the key band characteristics related to the target altered mineral are input into the altered mineral information extraction model to obtain the altered mineral distribution information of the study area.

[0159] In actual application, after obtaining the altered mineral distribution information, random sampling can be carried out in the target altered mineral area and non-target altered mineral area calibrated by the altered mineral distribution information to collect corresponding soil or rock samples, and geological survey methods (such as X-ray diffraction analysis, spectral analysis) can be used to verify and analyze the output results of the altered mineral information extraction model, identify the source of errors and factors that may affect the model output errors, and iteratively optimize the altered mineral information extraction model and altered mineral distribution information to generate more accurate altered mineral distribution information.

[0160] To sum up, the method for extracting altered mineral information provided in the embodiment of the present application determines the sample pixels in the potential distribution area of ​​the altered mineral by gridding the potential distribution area of ​​the target altered mineral in the study area, and performs cluster analysis on the pixels in each grid unit to determine the sample pixels in the potential distribution area of ​​the altered mineral. Based on the spectral characteristics, topographic characteristics, geochemical characteristics and key band characteristics in each sample pixel, the sample attribute information is constructed as a training sample to train the altered mineral information extraction model, so that the altered mineral information extraction model can effectively integrate the multi-modal vectors such as space, spectrum, topography and geochemistry of the potential distribution area, give full play to their complementary advantages, and thereby improve the accuracy and efficiency of the altered mineral information extraction model in predicting the target altered mineral.

[0161] This application also provides a device for extracting altered mineral information. Please refer to Figure 2 , Figure 2Structural diagram of the device for extracting altered mineral information provided in an embodiment of the present application.

[0162] like Figure 2 As shown, the device for extracting altered mineral information includes:

[0163] The potential distribution area determination unit 201 is used to determine the potential distribution area of ​​the target altered mineral in the study area based on the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area.

[0164] The grid division unit 202 is used to grid the potential distribution area and determine the attribute information of each pixel in the grid unit of the potential distribution area; wherein the attribute information includes at least the first spectral feature, the first topographic feature, the first geochemical feature and the first key band feature related to the target altered mineral corresponding to the pixel.

[0165] The cluster analysis unit 203 is configured to perform cluster analysis on the pixels of each grid unit, and determine the sample pixels of the potential distribution area based on the cluster analysis results.

[0166] The model training unit 204 is configured to determine the sample attribute information of the sample pixel as a training sample, and train a pre-built alteration mineral information extraction model to obtain a trained alteration mineral information extraction model.

[0167] The altered mineral prediction unit 205 is used to input the second spectral characteristics, second topographic characteristics, second geochemical characteristics and second key band characteristics related to the target altered mineral of each pixel in the study area into the trained altered mineral information extraction model to obtain the altered mineral distribution information of the study area.

[0168] In an optional embodiment of the present application, the potential distribution area of ​​the target altered mineral in the study area is determined based on the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area, including: determining the key band related to the target altered mineral, and extracting the band information of the multispectral or hyperspectral satellite image data based on the key band to obtain the key band raster data of the study area, wherein each key band raster data corresponds to each key band information; analyzing the key band raster data of the study area by principal component analysis or band ratio to obtain the indicative raster data of the target altered mineral; using Gaussian filtering , performing filtering analysis on the indicative grid data to obtain the filtered indicative grid data; performing threshold segmentation on the filtered indicative grid data based on a preset indicative grid data threshold to determine the potential distribution area; or, through a spectral angle matching method, analyzing the spectrum of the multispectral or hyperspectral satellite image data according to the standard spectral characteristics of the target altered mineral to obtain the indicative spectral angle index of the target altered mineral; using Gaussian filtering, performing filtering analysis on the indicative spectral angle index to obtain the filtered indicative spectral angle index; performing threshold segmentation on the filtered indicative spectral angle index based on a preset indicative spectral angle index threshold to determine the potential distribution area.

[0169] In an optional embodiment of the present application, the potential distribution area of ​​the target altered mineral in the study area is determined based on the standard spectral characteristics of the target altered mineral and at least one of the multispectral or hyperspectral satellite image data of the study area, including: determining the key band related to the target altered mineral, and extracting the band information of the multispectral or hyperspectral satellite image data based on the key band to obtain the key band raster data of the study area, wherein each key band raster data corresponds to each key band information; analyzing the key band raster data of the study area by principal component analysis to determine the first indicative raster data of the target altered mineral; filtering and analyzing the first indicative raster data by Gaussian filtering to obtain the filtered first indicative raster data; performing threshold segmentation on the filtered first indicative raster data based on a preset first indicative raster data threshold to obtain the first potential distribution area of ​​the target altered mineral; and analyzing the key bands of the study area by band ratio. The method comprises the following steps: analyzing the key band raster data to determine the second indicative raster data of the target altered mineral; filtering and analyzing the second indicative raster data by Gaussian filtering to obtain the filtered second indicative raster data; performing threshold segmentation on the filtered second indicative raster data based on a preset second indicative raster data threshold to obtain a second potential distribution area of ​​the target altered mineral; analyzing the spectrum of the multispectral or hyperspectral satellite image data by spectral angle matching according to the standard spectral characteristics of the target altered mineral to obtain the indicative spectral angle index of the target altered mineral; filtering and analyzing the indicative spectral angle index by Gaussian filtering to obtain the filtered indicative spectral angle index; performing threshold segmentation on the filtered indicative spectral angle index based on a preset indicative spectral angle index threshold to determine the third potential distribution area of ​​the target altered mineral; and determining the overlapping area of ​​the first potential distribution area, the second potential distribution area, and the third potential distribution area as the potential distribution area.

[0170] In an optional embodiment of the present application, cluster analysis is performed on the pixels of each of the grid units respectively, and sample pixels of the potential distribution area are determined based on the cluster analysis results, including: for any of the grid units, randomly selecting k pixels to be determined as the initial cluster center points of the grid unit; wherein k is a positive integer; based on the distance between each pixel in the grid unit and the initial cluster center point, the initial cluster of the grid unit is determined; for any initial cluster, based on the distance between each pixel in the initial cluster and the initial representative pixel, the initial representative pixel is updated to determine the updated representative pixel in the initial cluster; the representative pixel and other pixels within a preset neighborhood range of the representative pixel are determined as sample pixels.

[0171] In an optional embodiment of the present application, the distance between each pixel in the grid unit and the initial cluster center point, and the distance between each pixel in the initial cluster and the initial representative pixel are determined by the following formula:

[0172] ;

[0173] ;

[0174] ;

[0175] in, Represents the distance between pixel x and the initial representative pixel c or the initial cluster center c; Represents the attribute distance between pixel x and the initial representative pixel c or the initial cluster center c; Represents the spatial distance between pixel x and the initial representative pixel c or the initial cluster center c; represents the weight of the attribute distance; represents the weight of the spatial distance; Represents the i-th attribute information of pixel x; represents the i-th attribute information of the initial representative pixel c or the initial cluster center c; n represents the number of attribute information of pixels; Represents the distance weight of the i-th attribute of the pixel; Represents the projection value of pixel x on the jth axis; represents the projection value of the initial representative pixel c or the initial cluster center c on the j-th axis; m represents the dimension of the spatial coordinate system, m is equal to 2, and the first axis and the second axis of the spatial coordinate system are perpendicular to each other.

[0176] In an optional embodiment of the present application, the sample attribute information of the sample pixel is determined as a training sample, and a pre-constructed altered mineral information extraction model is trained to obtain a trained altered mineral information extraction model, including: inputting the sample attribute information of each of the sample pixels into the pre-constructed altered mineral information extraction model, so that the altered mineral information extraction model combines the multi-head self-attention mechanism and the sample attribute information to determine the predicted distribution probability of the target altered mineral corresponding to the geographical location of each of the sample pixels; determining the loss function of the pre-constructed altered mineral information extraction model based on the predicted distribution probability of the target altered mineral, and training the altered mineral information extraction model according to the loss function to obtain the trained altered mineral information extraction model.

[0177] In an optional embodiment of the present application, the loss function of the altered mineral information extraction model is expressed by the following formula:

[0178] ;

[0179] Wherein, L represents the loss function of the altered mineral information extraction model; the value of y indicates whether the pixel block of the multispectral or hyperspectral satellite image data is the target altered mineral. When y is equal to 1, it indicates that the pixel block is the target altered mineral; when y is equal to 0, it indicates that the pixel block is a non-altered mineral; p represents the predicted probability that the pixel block is the target altered mineral.

[0180] The above-mentioned device embodiments provided in this embodiment and the method embodiments of this application are based on the same application concept and can execute the altered mineral information extraction method provided in any of the above-mentioned embodiments of this application, and have the corresponding functional modules and beneficial effects of executing the altered mineral information extraction method. For technical details not fully described in this embodiment, please refer to the specific processing content of the altered mineral information extraction method provided in the above-mentioned embodiments of this application, and will not be repeated here.

[0181] The present application also provides an electronic device. Figure 3 , Figure 3 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0182] like Figure 3 As shown, the electronic device includes:

[0183] Processor 210;

[0184] a memory 200 for storing instructions executable by the processor 210;

[0185] The processor 210 is configured to execute the method for extracting alteration mineral information disclosed in any of the above embodiments by running instructions in the memory 200 .

[0186] The processor 210, the memory 200, the communication interface 220, the input device 230 and the output device 240 are interconnected via a bus.

[0187] A bus may include a pathway that transfers information between components of a computer system.

[0188] Processor 210 can be a general-purpose processor, such as a general-purpose central processing unit (CPU), a microprocessor, or the like. It can also be an application-specific integrated circuit (ASIC), or one or more integrated circuits used to control the execution of the program of the present invention. It can also be a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic device, discrete gate or transistor logic device, or discrete hardware components.

[0189] The processor 210 may include a main processor, and may also include a baseband chip, a modem, and the like.

[0190] Memory 200 stores programs that implement the technical solutions of the present invention and may also store an operating system and other key services. Specifically, the programs may include program code, which includes computer operating instructions. More specifically, memory 200 may include read-only memory (ROM), other types of static storage devices capable of storing static information and instructions, random access memory (RAM), other types of dynamic storage devices capable of storing information and instructions, disk storage, flash memory, and the like.

[0191] The input device 230 may include a device for receiving data and information input by a user, such as a keyboard, a mouse, a camera, a scanner, a touch screen, etc.

[0192] Output device 240 may include devices that allow information to be output to a user, such as a display screen, printer, speakers, etc.

[0193] The communication interface 220 may include any device such as a transceiver to communicate with other devices or communication networks, such as Ethernet, a radio access network (RAN), a wireless local area network (WLAN), etc.

[0194] The processor 210 executes the program stored in the memory 200 and calls other devices, and can be used to implement each step of any method for extracting alteration mineral information provided in the above embodiments of the present application.

[0195] In addition to the above-mentioned methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, which, when executed by a processor, enable the processor to execute the steps in the method for extracting altered mineral information of various embodiments of the present application.

[0196] The computer program product may be written in any combination of one or more programming languages ​​to implement the program code for performing the operations of the embodiments of the present application, including object-oriented programming languages ​​such as Java, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's computing device, as a standalone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0197] In addition, the embodiments of the present application may also be a storage medium on which a computer program is stored, and the computer program is used by a processor to execute the steps in the method for extracting altered mineral information in various embodiments of the present application.

[0198] For the sake of simplicity, the aforementioned method embodiments are described as a series of action combinations. However, those skilled in the art should be aware that this application is not limited by the order of the actions described, because according to this application, certain steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also be aware that the embodiments described in this specification are all preferred embodiments, and the actions and modules involved are not necessarily required by this application.

[0199] It should be noted that the various embodiments in this specification are described in a progressive manner, with each embodiment focusing on the differences from other embodiments. Similarities between the various embodiments can be referred to in conjunction with each other. For device embodiments, since they are generally similar to method embodiments, their description is relatively simple, and for relevant details, reference can be made to the description of the method embodiments.

[0200] The steps in the methods of each embodiment of the present application can be adjusted in sequence, merged, and deleted according to actual needs, and the technical features recorded in each embodiment can be replaced or combined.

[0201] The modules and sub-modules in the devices and terminals in the various embodiments of the present application can be merged, divided, and deleted according to actual needs.

[0202] In the several embodiments provided in this application, it should be understood that the disclosed terminals, devices, and methods can be implemented in other ways. For example, the terminal embodiments described above are merely illustrative. For example, the division of modules or submodules is merely a logical function division. In actual implementation, there may be other division methods, such as multiple submodules or modules can be combined or integrated into another module, or some features can be ignored or not executed. In addition, the mutual coupling or direct coupling or communication connection shown or discussed can be an indirect coupling or communication connection through some interface, device or module, which can be electrical, mechanical or other forms.

[0203] The modules or submodules described as separate components may or may not be physically separate, and the components of the modules or submodules may or may not be physical modules or submodules, that is, they may be located in one place or distributed across multiple network modules or submodules. Some or all of the modules or submodules may be selected to achieve the purpose of this embodiment according to actual needs.

[0204] In addition, each functional module or submodule in each embodiment of the present application may be integrated into a processing module, or each module or submodule may exist physically separately, or two or more modules or submodules may be integrated into a single module. The above-mentioned integrated modules or submodules may be implemented in the form of hardware or software functional modules or submodules.

[0205] Professionals may further appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of the two. In order to clearly illustrate the interchangeability of hardware and software, the above description has generally described the components and steps of each example according to their functions. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians may use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0206] The steps of the methods or algorithms described in conjunction with the embodiments disclosed herein may be implemented directly using hardware, software executed by a processor, or a combination of the two. The software may be stored in random access memory (RAM), internal memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, a hard disk, a removable disk, a CD-ROM, or any other form of storage medium known in the art.

[0207] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "comprising," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or device comprising a series of elements includes not only those elements, but also other elements not explicitly listed, or elements inherent to such process, method, article, or device. In the absence of further limitations, an element defined by the phrase "comprising a ..." does not exclude the presence of additional identical elements in the process, method, article, or device comprising the element.

[0208] The above description of the disclosed embodiments will enable those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is to be construed in the widest manner consistent with the principles and novel features disclosed herein.

Claims

1. A method for extracting altered mineral information, characterized in that: include: Determine the key bands related to the target altered minerals, and extract band information from the multispectral or hyperspectral satellite image data of the study area based on the key bands to obtain key band raster data of the study area, wherein each key band raster data corresponds to each key band information; Analyzing the key band raster data of the study area by principal component analysis to determine the first indicative raster data of the target altered mineral; Using Gaussian filtering, the first indicative raster data is filtered and analyzed to obtain filtered first indicative raster data; based on a preset first indicative raster data threshold, the first indicative raster data is threshold segmented to obtain a first potential distribution area of ​​the target altered mineral; Analyzing the key band raster data of the study area by means of band ratios to determine the second indicative raster data of the target altered mineral; Using Gaussian filtering, the second indicative raster data is filtered and analyzed to obtain filtered second indicative raster data; based on a preset second indicative raster data threshold, the filtered second indicative raster data is threshold segmented to obtain a second potential distribution area of ​​the target altered mineral; By using a spectral angle matching method, the spectrum of the multispectral or hyperspectral satellite image data is analyzed according to the standard spectral characteristics of the target altered mineral to obtain an indicative spectral angle index of the target altered mineral; The indicative spectral angle index is filtered and analyzed by Gaussian filtering to obtain the filtered indicative spectral angle index; based on a preset indicative spectral angle index threshold, the filtered indicative spectral angle index is threshold segmented to determine the third potential distribution area of ​​the target altered mineral; determining an overlapping area among the first potential distribution area, the second potential distribution area, and the third potential distribution area as a potential distribution area; Gridding the potential distribution area and determining attribute information of each pixel in the grid unit of the potential distribution area; wherein the attribute information includes at least a first spectral feature, a first topographic feature, a first geochemical feature corresponding to the pixel, and a first key band feature related to the target altered mineral; Performing cluster analysis on the pixels of each grid unit respectively, and determining the sample pixels of the potential distribution area based on the cluster analysis results; Determining the sample attribute information of the sample pixel as a training sample, training a pre-built alteration mineral information extraction model to obtain a trained alteration mineral information extraction model; The second spectral characteristics, second topographic characteristics, second geochemical characteristics of each pixel in the study area and the second key band characteristics related to the target altered mineral are input into the trained altered mineral information extraction model to obtain the altered mineral distribution information of the study area.

2. The method for extracting altered mineral information according to claim 1, characterized in that: The performing cluster analysis on the pixels of each grid unit respectively and determining the sample pixels of the potential distribution area based on the cluster analysis results includes: For any of the grid cells, k pixels are randomly selected to be determined as the initial cluster center points of the grid cell; wherein k is a positive integer; Determining an initial cluster of the grid unit according to the distance between each pixel in the grid unit and the initial cluster center point; For any initial cluster, the initial representative pixel is updated according to the distance between each pixel in the initial cluster and the initial representative pixel, and an updated representative pixel in the initial cluster is determined; The representative pixel and other pixels within a preset neighborhood of the representative pixel are determined as sample pixels.

3. The method for extracting altered mineral information according to claim 2, characterized in that: The distance between each pixel in the grid unit and the initial cluster center point, as well as the distance between each pixel in the initial cluster and the initial representative pixel, are determined by the following formula: d total (x,c)=w a ·d a (x,c)+w s ·d s (x,c); Among them, d total (x,c) represents the distance between pixel x and the initial representative pixel c or the initial cluster center c; d a (x,c) represents the attribute distance between pixel x and the initial representative pixel c or the initial cluster center c; d s (x,c) represents the spatial distance between pixel x and the initial representative pixel c or the initial cluster center c; w a represents the weight of the attribute distance; w s represents the weight of the spatial distance; x i Represents the i-th attribute information of pixel x; c i represents the i-th attribute information of the initial representative pixel c or the initial cluster center c; n represents the number of attribute information of pixels; Represents the distance weight of the i-th attribute of the pixel; Represents the projection value of pixel x on the jth axis; represents the projection value of the initial representative pixel c or the initial cluster center c on the j-th axis; m represents the dimension of the spatial coordinate system, m is equal to 2, and the first axis and the second axis of the spatial coordinate system are perpendicular to each other.

4. The method for extracting altered mineral information according to claim 1, characterized in that: The method of determining the sample attribute information of the sample pixel as a training sample and training a pre-built alteration mineral information extraction model to obtain a trained alteration mineral information extraction model includes: Inputting the sample attribute information of each sample pixel into the pre-built altered mineral information extraction model, so that the altered mineral information extraction model combines the multi-head self-attention mechanism and the sample attribute information to determine the predicted distribution probability of the target altered mineral at the geographical location corresponding to each sample pixel; The loss function of the pre-constructed altered mineral information extraction model is determined based on the predicted distribution probability of the target altered mineral, and the altered mineral information extraction model is trained according to the loss function to obtain the trained altered mineral information extraction model.

5. The method for extracting altered mineral information according to claim 4, characterized in that: The loss function of the altered mineral information extraction model is expressed by the following formula: L=-[y·log(p)+(1-y)·log(1-p)]; Wherein, L represents the loss function of the altered mineral information extraction model; the value of y indicates whether the pixel block of the multispectral or hyperspectral satellite image data is the target altered mineral. When y is equal to 1, it indicates that the pixel block is the target altered mineral; when y is equal to 0, it indicates that the pixel block is a non-altered mineral; p represents the predicted probability that the pixel block is the target altered mineral.

6. A device for extracting altered mineral information, characterized in that: include: A potential distribution area determination unit is used to determine the key bands related to the target altered mineral, and extract band information from the multispectral or hyperspectral satellite image data of the study area based on the key bands to obtain the key band raster data of the study area, wherein each key band raster data corresponds to each key band information; the key band raster data of the study area is analyzed by principal component analysis to determine the first indicative raster data of the target altered mineral; the first indicative raster data is filtered and analyzed by Gaussian filtering to obtain the filtered first indicative raster data; based on a preset first indicative raster data threshold, the filtered first indicative raster data is threshold segmented to obtain the first potential distribution area of ​​the target altered mineral; the key band raster data of the study area is analyzed by band ratio to determine the second indicative raster data of the target altered mineral; The second indicative raster data is filtered and analyzed by Gaussian filtering to obtain the filtered second indicative raster data; based on a preset second indicative raster data threshold, the filtered second indicative raster data is threshold segmented to obtain a second potential distribution area of ​​the target altered mineral; the spectrum of the multispectral or hyperspectral satellite image data is analyzed according to the standard spectral characteristics of the target altered mineral by a spectral angle matching method to obtain an indicative spectral angle index of the target altered mineral; the indicative spectral angle index is filtered and analyzed by Gaussian filtering to obtain the filtered indicative spectral angle index; based on a preset indicative spectral angle index threshold, the filtered indicative spectral angle index is threshold segmented to determine a third potential distribution area of ​​the target altered mineral; and the overlapping area of ​​the first potential distribution area, the second potential distribution area, and the third potential distribution area is determined as a potential distribution area; a grid division unit, configured to divide the potential distribution area into grids and determine attribute information of each pixel in the grid cells of the potential distribution area; wherein the attribute information includes at least a first spectral feature, a first topographic feature, a first geochemical feature corresponding to the pixel, and a first key band feature associated with the target altered mineral; A cluster analysis unit, configured to perform cluster analysis on the pixels of each grid unit, and determine the sample pixels of the potential distribution area based on the cluster analysis results; A model training unit is used to determine the sample attribute information of the sample pixel as a training sample, train a pre-built alteration mineral information extraction model, and obtain a trained alteration mineral information extraction model; The altered mineral prediction unit is used to input the second spectral characteristics, second topographic characteristics, second geochemical characteristics of each pixel in the study area and the second key band characteristics related to the target altered mineral into the trained altered mineral information extraction model to obtain the altered mineral distribution information of the study area.

7. An electronic device, characterized in that: include: processor; a memory for storing instructions executable by the processor; The processor is used to execute the method for extracting alteration mineral information according to any one of claims 1 to 5 by running instructions in the memory.

8. A computer storage medium, characterized in that The computer storage medium stores instructions, and when the instructions are executed, the method for extracting altered mineral information according to any one of claims 1 to 5 is implemented.

Images