Three-dimensional geological model construction method and system based on multi-modal data

Abstract

The application relates to the technical field of three-dimensional geological model construction, in particular to a three-dimensional geological model construction method and system based on multi-modal data, which comprises the following steps: obtaining drilling data, geophysical data, multispectral remote sensing data, geological mapping data and elevation data for three-dimensional geological model construction in a region to be modeled; obtaining deviation coefficients of reflection characteristics of each pixel under different wave bands by the correlation of reflection intensities in the local range of each pixel under different wave bands in the multispectral remote sensing data and the reflection intensity difference of each pixel and the pixels in the local range under the same wavelength, and then obtaining a judgment coefficient of the reflection characteristic similarity between different pixels to cluster the pixels; adjusting a smoothing parameter when filtering the reflection intensity according to the distribution characteristics of the deviation coefficients of the reflection characteristics of the pixels in each cluster; and constructing a three-dimensional geological model by using three-dimensional modeling software. The application can improve the construction precision of the three-dimensional geological model.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of three-dimensional geological model construction, in particular to a three-dimensional geological model construction method and system based on multi-modal data. BACKGROUND

[0002] The three-dimensional geological model can realize the digitization and visualization of the underground geological structure, lithology distribution and structural characteristics, and is widely used in the fields of geological resource exploration, engineering construction and environmental governance. Nowadays, the three-dimensional geological model is constructed combined with multi-modal data, which realizes the all-around and multi-dimensional description of the geological structure, and fully reflects the geological characteristics through different modal data.

[0003] However, in the process of constructing the three-dimensional geological model, due to the influence of various interference factors, the data used for constructing the three-dimensional geological model has poor quality, which further affects the accuracy of constructing the three-dimensional geological model. The remote sensing data reflects the macro geological characteristics of the surface and shallow part. In the actual construction process, interference factors such as atmospheric scattering and sensor noise will cause the quality of remote sensing data to decrease, and there are differences in the interference influence at different times, so that the traditional filtering correction method has a large deviation for the remote sensing data, which affects the accuracy of surface connection and shallow structure description in the process of multi-source data fusion, and causes the construction accuracy of the three-dimensional geological model to decrease. SUMMARY

[0004] In order to solve the above technical problems, the purpose of the present application is to provide a three-dimensional geological model construction method and system based on multi-modal data, and the technical scheme adopted is as follows:

[0005] The embodiment of the present application provides a three-dimensional geological model construction method based on multi-modal data, which comprises the following steps:

[0006] Multi-modal data used for constructing the three-dimensional geological model is obtained in the region to be modeled, including drilling data, geophysical data, multi-spectral remote sensing data, geological mapping data and elevation data;

[0007] The deviation coefficient of the reflection characteristics of each pixel under different wave bands is obtained through the correlation of the reflection intensity in the local range of each pixel under different wave bands of the multi-spectral remote sensing data, and the reflection intensity difference between each pixel and the pixel in the local range under the same wavelength, and the judgment coefficient of the reflection characteristic similarity between different pixels is obtained combined with the position difference and the reflection intensity difference between different pixels, which is used for clustering the pixels, and the smoothing parameter for filtering the reflection intensity corresponding to the pixels in each clustering cluster is adjusted according to the distribution characteristics of the deviation coefficient of the reflection characteristics of the pixels in each clustering cluster.

[0008] According to the processed multi-modal data of the three-dimensional geological model, the three-dimensional geological model is constructed by using a three-dimensional modeling software.

[0009] Preferably, the obtaining of the deviation coefficient of the reflection feature of each pixel is further comprising:

[0010] ;

[0011] wherein, represents the deviation coefficient of the reflection feature of the i-th pixel in the current waveband; represents the difference value of the reflection feature between the i-th pixel and its i-th neighboring pixel in the current waveband; represents the characteristic value of the reflection feature associated change of the i-th neighboring pixel of the i-th pixel compared to the overall neighborhood range; represents the number of the neighboring pixels of the i-th pixel. Preferably, the reflection intensity value of each pixel in the multi-spectral remote sensing data at all wavelengths is mapped to a two-dimensional rectangular coordinate system with wavelength as the horizontal coordinate and reflection intensity value as the vertical coordinate, and a reflection feature curve corresponding to each pixel is obtained by curve fitting. Preferably, a mean curve of the reflection feature curves of each pixel and its neighboring pixels is calculated, and the absolute value of the Pearson correlation coefficient between the reflection feature curve of each pixel and its neighboring pixels and the mean curve at the corresponding reflection intensity value in the current waveband is taken as the characteristic value of the reflection feature associated change of each neighboring pixel compared to the overall neighborhood range. Preferably, the difference value of the reflection intensity value of each pixel and its neighboring pixels at the same wavelength is taken as the absolute value, and the mean value of the absolute values of the difference values of each pixel and its neighboring pixels at all wavelengths in the current waveband is taken as the difference value of the reflection feature of each pixel and its neighboring pixels in the current waveband.

[0012] Preferably, the obtaining of the judgment coefficient of the reflection feature similarity between different pixels is further comprising:

[0013] ;

[0014] wherein,

[0015] represents the judgment coefficient of the reflection feature similarity between the i-th and j-th pixels; represents the Euclidean distance between the i-th and j-th pixels;

[0016] represents the Euclidean distance between the i-th and j-th pixels.

[0017] ​​​​​​​​​​​DTW distance of the reflection intensity value of the first pixel in the first wave band; DTW distance of the reflection intensity value of the first pixel in the first wave band; DTW distance of the reflection intensity value of the first pixel in the first wave band; DTW distance of the reflection intensity value of the first pixel in the first wave band; DTW distance of the reflection intensity value of the first pixel in the first wave band; DTW distance of the reflection intensity value of the first pixel in the first wave band; DTW distance of the reflection intensity value of the first pixel in the first wave band; DTW distance of the reflection intensity value of the first pixel in the first wave band.

[0018] Preferably, in the process of clustering the pixels, the judgment coefficient of the reflection feature similarity between different pixels is taken as the measurement distance of different pixels.

[0019] Preferably, the adjusted Gaussian smoothing parameter of the reflection intensity of the current clustering cluster on the current wave band when Gaussian filtering is performed is The acquisition of the adjusted Gaussian smoothing parameter is further:

[0020] , wherein, the initial Gaussian smoothing parameter of the reflection intensity of the current clustering cluster on the current wave band when Gaussian filtering is performed, the normalized result of the difference value of the reflection intensity deviation feature of the pixel in the current clustering cluster on the current wave band.

[0021] Preferably, for the current wave band, the deviation coefficients of the reflection features of all the pixels in the clustering cluster are statistically analyzed, and the probability statistical result curve is fitted, the KL divergence value between the fitting curve of the current clustering cluster and the fitting curve of each other clustering cluster on the current wave band is calculated, and the mean value of all the KL divergence values is taken as the difference value of the reflection intensity deviation feature of the pixel in the current clustering cluster on the current wave band.

[0022] The embodiment of the application also provides a three-dimensional geological model construction system based on multi-modal data, which comprises a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the steps of the three-dimensional geological model construction method based on multi-modal data according to any one of the above embodiments when executing the computer program.

[0023] As can be seen from the above, the three-dimensional geological model construction method and system based on multi-modal data provided by the application have at least the following beneficial effects:

[0024] In the process of constructing a three-dimensional geological model through multi-modal data, the influence of the quality of multi-source data on the overall construction accuracy leads to the problem of reduced construction accuracy of the three-dimensional geological model. Therefore, the multi-modal data of the region to be modeled is first collected, the influence of regional strip noise and random noise interference differences in the data processing process on the data quality is considered, the reflection characteristics of different positions are analyzed, the interference differences between different positions in different wave bands are compared based on the analysis results, and then the reflection intensity data of all positions is divided. Based on the division results, the interference influence characteristics of different regions and different wave bands are accurately judged and analyzed. The Gaussian smoothing process is optimized through the analysis results, which has the beneficial effect of accurately denoising the reflection intensity data of remote sensing data, improving the correction accuracy of the spatial position of geological mapping data, and improving the accuracy of surface connection and shallow structure description in the three-dimensional model construction process. BRIEF DESCRIPTION OF DRAWINGS

[0025] In order to more clearly illustrate the technical solutions and advantages of the embodiments of the present application or prior art, the drawings needed in the embodiments or prior art description will be briefly introduced. Obviously, the drawings in the following description are only some embodiments of the present application, and those skilled in the art can obtain other drawings according to these drawings without creative labor.

[0026] Figure 1 The step flow chart of the three-dimensional geological model construction method based on multi-modal data provided by the present application. DETAILED DESCRIPTION

[0027] In order to further illustrate the technical means and effects adopted by the present application to achieve the predetermined invention purpose, the three-dimensional geological model construction method and system based on multi-modal data according to the present application, its specific implementation, structure, features and effects are described in detail as follows. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. In addition, the specific features, structures or characteristics in one or more embodiments can be combined in any suitable form.

[0028] Unless otherwise specified and limited, terms such as “comprising,” “including,” or any other variations thereof are intended to cover a non-exclusive inclusion, such that a circuit structure, article, or device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such an article or device. Without further limitation, an element defined by the phrase “comprising one…” does not exclude the presence of other identical elements in the article or device that includes said element. Furthermore, the term “and / or” as used herein includes any and all combinations of one or more of the associated listed items. All technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0029] The following, in conjunction with the accompanying drawings, details the specific scheme of the three-dimensional geological model construction method and system based on multimodal data provided in this application.

[0030] Please see Figure 1 The document illustrates a flowchart of a method for constructing a three-dimensional geological model based on multimodal data according to an embodiment of this application, including the following steps:

[0031] Step 1: Acquire multimodal data for 3D geological model construction in the area to be modeled, including borehole data, geophysical data, multispectral remote sensing data, geological mapping data, and elevation data.

[0032] First, in this embodiment, data will be acquired during the construction of a 3D geological model based on multimodal data, including borehole data, geophysical data, remote sensing data, geological mapping data, and elevation data. Specifically, in the area to be modeled, lithology and physical parameters, geophysical characteristics, borehole coordinates, and borehole inclination data at different depths will be recorded using core sampling and well logging techniques. Geophysical data will be collected using seismic exploration techniques. Multispectral remote sensing data of the area to be modeled will be acquired using satellite remote sensing. Geological mapping data of the area to be modeled will be acquired through geological surveys, including information on surface outcrops, stratum contact relationships, and fault strikes. Elevation data of the area to be modeled can be collected using a total station, GPS, or lidar.

[0033] Step 2: By analyzing the correlation of reflection intensity within the local area of ​​each pixel in different bands of multispectral remote sensing data, and the difference in reflection intensity between each pixel and the pixels within the local area at the same wavelength, the deviation coefficient of reflection characteristics of each pixel in different bands is obtained. Combining the position and the degree of difference in reflection intensity between different pixels, the similarity judgment coefficient of reflection characteristics between different pixels is obtained, which is used to cluster the pixels. Based on the distribution characteristics of the deviation coefficient of reflection characteristics of pixels within each cluster, the smoothing parameter is adjusted when filtering the reflection intensity of the pixels corresponding to each cluster.

[0034] In the process of three-dimensional model construction, the quality of multi-modal data is a key factor affecting the accuracy of model construction. In the process of collecting drilling data and geophysical data for the single-point lithology characteristics of the area to be modeled, data anomalies and missing values may occur at different depths and in different areas due to interference. Therefore, the adjacent depth difference method and anomaly detection algorithm are used for abnormal value processing and missing value filling. In addition, for the environmental noise interference of geophysical data, Wiener filtering, static correction, and time-depth conversion are used to process the geophysical data. For elevation data, the Kriging interpolation method is used to fill in the blank areas. Geological mapping data need to be further corrected in space based on remote sensing data. However, the quality of remote sensing data affects the accuracy of surface connection and shallow structure description in the process of three-dimensional model construction. When the deviation of remote sensing data is large, it may cause large errors in spatial deviation correction of geological mapping data, affecting the construction accuracy of the final three-dimensional geological model.

[0035] Based on the above analysis process, in this embodiment, for the acquired remote sensing data, first, consider the error caused by atmospheric scattering in the actual collection process. Through radiation calibration, geometric correction, and atmospheric correction, the surface reflection characteristic data is obtained to avoid the influence of atmospheric scattering interference. However, in the actual collection process, cloud cover may cause strip noise in local areas, and electronic noise and pulse interference may occur during data transmission, causing random noise interference, which increases the error in spectral feature judgment and blurs the stratigraphic boundary in remote sensing data, thereby affecting the accuracy of multi-source data fusion. Therefore, to improve the accuracy of three-dimensional geological model construction based on multi-modal data, consider the influence of remote sensing data deviation on surface connection and shallow structure description in the actual processing process, and perform optimization processing on the interference characteristics of the collected remote sensing data after correction, thereby improving the construction accuracy of the three-dimensional geological model. The specific analysis and processing process is as follows:

[0036] First, due to the large difference in geological characteristics at different positions in the area to be modeled, the interference characteristics of the reflection characteristics under different interference differences are different. Therefore, for each pixel in the multi-spectral remote sensing data, the reflection intensity value of each pixel at all wavelengths is mapped to a two-dimensional rectangular coordinate system with wavelength as the horizontal coordinate and reflection intensity value as the vertical coordinate. At the same time, in this embodiment, the least squares method is used to fit the mapped results, and the fitted results are taken as the reflection characteristic curve of each pixel. Considering the consistency of local reflection characteristics of geological structures, the mean curve of the reflection characteristic curves of each pixel and its neighboring pixels is calculated. The mean curve is calculated by calculating the mean value of the reflection intensity values of each pixel and its neighboring pixels at the same wavelength. The neighboring pixels are the pixels in the 8-neighborhood of each pixel.

[0037] Further, for each pixel wavelength range is divided, and then based on the difference in reflection characteristics of different wave bands in the local range is compared, in the process of data acquisition and transmission interference, may lead to local reflection light intensity appears random high frequency interference and strip noise, resulting in geological area in the local range under the reflection characteristics of different directions and different frequency band difference, therefore, for each pixel corresponding wavelength range is divided evenly, specifically, in the embodiment, for wavelength uniform division of the number is 20, that is, 20 wave bands.

[0038] Further, taking the current wave band as an example, in the embodiment, the absolute value of the Pearson correlation coefficient between the reflection characteristic curve corresponding to each pixel of the adjacent pixel and the mean curve in the current wave band is calculated, and is taken as the feature value of the correlation change of the reflection characteristic of each adjacent pixel compared with the overall neighborhood range. The larger the feature value is, the more significant the reflection characteristic of the position of the pixel in the corresponding wave band compared with the overall reflection characteristic of the neighborhood range, and the greater the influence of the interference on the center pixel on the accuracy of the deviation judgment.

[0039] Further, based on the above analysis, the reflection spectral characteristics in the 8-neighborhood range of each pixel are analyzed, and the correlation characteristics of each pixel in different wave bands compared with the overall neighborhood range are compared. The more significant the correlation characteristics are, the more significant the reflection intensity value in different wave bands is due to strip noise and random noise interference. Specifically, for the current wave band, the difference between the reflection intensity value of each pixel and its adjacent pixel at the same wavelength is taken as the absolute value, and the mean value of the difference between each pixel and its adjacent pixel at all wavelengths in the current wave band is taken as the difference value of the reflection characteristics of each pixel and its adjacent pixel in the current wave band. Based on the difference value of the reflection characteristics of the pixel in the current wave band and the difference of the reflection characteristics of each pixel in different wave bands compared with the overall neighborhood range, the degree of deviation of the reflection characteristics of the pixel in different wave bands due to interference is analyzed, and the deviation coefficient of the reflection characteristics of each pixel in each wave band is constructed. Preferably, the specific calculation relationship is:

[0040] ;

[0041] wherein, represents the deviation coefficient of the reflection characteristics of the i-th pixel in the current wave band; represents the difference value of the reflection characteristics of the i-th pixel and its i-th adjacent pixel in the current wave band; represents the feature value of the correlation change of the i-th adjacent pixel of the i-th pixel compared with the overall neighborhood range reflection characteristics; ​​​​​Indicates the first The number of neighboring pixels of a given pixel.

[0042] Understandably, the larger the calculated deviation coefficient, the more severe the impact of noise interference on the reflection characteristics of the pixel's location, considering the differences between the reflection characteristics of each pixel's location and those of neighboring pixels in different directions, as well as the correlation analysis of the reflection characteristics of different pixels with the overall domain range.

[0043] Step 2: Based on the above analysis, during the acquisition of multispectral remote sensing data of the area to be modeled, the deviation characteristics caused by interference at each location in different bands are comprehensively analyzed, and then the differences in multispectral reflectance intensity between different locations are analyzed; based on the analysis results, the multispectral remote sensing data are divided, and the strip noise and random noise interference in different regions are analyzed.

[0044] Therefore, for any two pixels in multispectral remote sensing data, data with similar reflection characteristics are divided by comparing the differences in reflection intensity between different pixels in different bands. Specifically, the DTW distance of the reflection intensity values ​​corresponding to the two pixels is calculated within each band, and the Euclidean distance between the two pixels is also calculated. The larger the Euclidean distance, the farther the distance between the corresponding positions of the pixels. Conversely, the larger the DTW distance, the greater the difference in reflection intensity between the corresponding pixels in the current band. Furthermore, the mean value of the deviation coefficient of the reflection characteristics of the two pixels within each band is calculated. The larger the mean value, the greater the difference in reflection characteristics between the two pixels due to interference within the corresponding band. Based on the above analysis process, the judgment coefficient of the similarity of reflection characteristics between pixels is calculated, and its calculation relationship is as follows:

[0045] ;

[0046] in, Indicates the first The and the first The coefficient for determining the similarity of reflection features between individual pixels; Indicates the first The and the first Euclidean distance between pixels; Indicates the first The and the first The pixel in the first DTW distance for reflection intensity values ​​within each band; Indicates the first Within the first band The and the first The mean of the deviation coefficients of the reflectance characteristics of each pixel. Indicates the first the sum of the mean of the deviation coefficients of the reflection characteristics of all arbitrary two pixels in a wave band, m is the preset wave band number, and in this embodiment, m is 20, wherein, The greater the value is, the greater the influence of the reflection intensity difference between the pixels in the corresponding wave band on the accuracy of the reflection characteristic similarity judgment is; the greater the calculated judgment coefficient is, the greater the reflection characteristic difference between the pixels in different wave band ranges is.

[0047] Further, all the pixels are taken as inputs, and a K-means clustering algorithm is used to cluster and divide the pixels, wherein the measurement distance between different pixels in the clustering and dividing process is the judgment coefficient of the reflection characteristic similarity between different pixels, the cluster number and the maximum iteration number are 8 and 200 respectively; the implementation process of the specific K-means clustering algorithm is known to those skilled in the art, and will not be described in detail.

[0048] Step 3: Based on the above clustering and dividing results, the pixels with similar reflection characteristics under the influence of different wave band interference in the actual acquisition process are divided, based on the dividing results, the deviation of the reflection intensity data under the influence of the difference of different regions is analyzed, and then the filtering processing effect of the reflection intensity data of different regions is optimized.

[0049] Specifically, for the current wave band, for each cluster in the division, first, the probability statistics of the deviation coefficients of the reflection characteristics of all the pixels in the cluster are performed, and the probability statistics results are curve fitted by the least square method; further, the KL divergence values between the fitting curves of the current cluster and each other cluster in the current wave band are calculated, and the mean of all the KL divergence values is taken as the difference value of the reflection intensity deviation characteristics of the pixels in the current cluster in the current wave band.

[0050] Further, the difference values of the reflection intensity deviation characteristics obtained for all the clusters are normalized, and in this embodiment, the Softmax function is used for normalization, and based on the normalization result, the reflection intensity data of the pixels in each cluster under different wave bands is subjected to Gaussian filtering processing.

[0051] In this embodiment, specifically, for each cluster, the reflection intensity of all the pixels in the cluster under each wave band is taken as input, and a Gaussian filter is used to filter and denoise the reflection intensity data under each wave band; the smoothing parameters of the Gaussian filtering processing of different wave bands are adjusted for the influence characteristics of the strip noise and random noise interference of different regional pixels, so as to improve the filtering processing effect of the reflection intensity data under different wave bands in different regions, avoid the problem of large deviation of remote sensing data caused by strip noise and random noise interference, and further affect the accuracy of the three-dimensional geological model construction; wherein the relationship formula of the Gaussian smoothing parameter adjustment of different clusters in the processing process is:

[0052] , represents the adjusted Gaussian smoothing parameter when the current cluster in the current band is subjected to Gaussian filtering of the reflectance intensity, represents the initial Gaussian smoothing parameter when the current cluster in the current band is subjected to Gaussian filtering of the reflectance intensity, and the size is 2, represents the normalized result of the difference value of the reflectance intensity deviation feature of the pixels in the current cluster in the current band.

[0053] It can be understood that the difference of the reflectance intensity of the region where all the pixels in the cluster are located in the corresponding band under the influence of noise interference is more significant, that is, the difference value of the reflectance intensity deviation feature is larger, and the corresponding noise interference is more significant. Therefore, the smoothing parameter of the Gaussian filtering processing is larger, so as to reduce the influence of noise interference. The specific Gaussian filtering processing process is known to those skilled in the art and will not be described here.

[0054] Step 3: According to the processed three-dimensional geological model multi-modal data, a three-dimensional modeling software is used to construct the three-dimensional geological model.

[0055] Further, the three-dimensional geological model is constructed according to the processed three-dimensional geological model multi-modal data. Preferably, in the present embodiment, the multi-modal information data of the processed three-dimensional geological model is taken as input, and the GOCAD three-dimensional modeling software is used to construct the three-dimensional geological model.

[0056] Specifically, first, a terrain surface model of the region to be modeled is generated based on the elevation data, which is used as the upper boundary of the three-dimensional geological model, so that the three-dimensional geological model is accurately connected with the surface form; further, the fused stratum interface, fault and rock mass structure information is imported into the three-dimensional modeling software, and the three-dimensional form of each layer is constructed through surface modeling technology, and the contact relationship between the layers is determined, for example, the integrated contact and the unconformable contact, and the intersection relationship between the fault and the stratum is processed through Boolean operation to ensure the accuracy of the structure logic.

[0057] Further, under the model framework constructed above, the embodiment adopts the collaborative Kriging interpolation algorithm to distribute the discrete attribute data such as lithology and porosity to the three-dimensional grid to obtain an attribute model; further, the overall morphology of the constructed model is visualized by the three-dimensional visualization technology, and the model accuracy is evaluated by using the verification drilling data, for example, by comparing the deviation between the predicted stratum depth and the actual drilling depth to evaluate, if the deviation meets the standard, the final three-dimensional geological model is output; if it does not meet the standard, it returns to the data fusion stage for re-optimization until the model meets the accuracy requirement. It should be noted that the implementer can use other existing technologies for three-dimensional model construction and accuracy evaluation to realize the specific embodiment, and the specific embodiment is not limited in the embodiment.

[0058] Based on the same inventive concept as the above method, the embodiment of the present application also provides a three-dimensional geological model construction system based on multi-modal data, which comprises a memory, a processor and a computer program stored in the memory and running on the processor, and the processor implements the steps of the three-dimensional geological model construction method based on multi-modal data according to any one of the above embodiments when executing the computer program.

[0059] It can be understood that the above-mentioned sequence of the embodiments of the present application is only for description, and does not represent the advantages and disadvantages of the embodiments. And the above describes the specific embodiments of the present application. In addition, the processes depicted in the drawings do not necessarily require the specific order or continuous order shown to achieve the desired results. In some embodiments, multi-task processing and parallel processing are possible or can be advantageous.

[0060] Each embodiment in the present application is described in a progressive manner, and the same and similar parts between each embodiment can be referred to each other, and each embodiment mainly describes the difference from other embodiments.

[0061] The above is only the embodiment of the present application, and is not used to limit the scope of the present application. Any equivalent structure or equivalent process conversion using the content of the present application specification and drawings, or direct or indirect application in other related technical fields, are also included in the protection scope of the present application.

Claims

1. A method for constructing a three-dimensional geological model based on multi-modal data, characterized in that, The method comprises the following steps: acquiring multi-modal data for three-dimensional geological model construction in a region to be modeled, including drilling data, geophysical data, multispectral remote sensing data, and geological mapping data and elevation data; obtaining a deviation coefficient of the reflection characteristics of each pixel at different wave bands by correlating the reflection intensities in the local range of each pixel at different wave bands in the multispectral remote sensing data, and the difference in the reflection intensities of each pixel and the pixels in the local range at the same wavelength, and obtaining a judgment coefficient of the similarity of the reflection characteristics between different pixels for clustering the pixels, and adjusting the smoothing parameter for filtering the reflection intensities of the pixels in each cluster according to the distribution characteristics of the deviation coefficient of the reflection characteristics of the pixels in each cluster; constructing a three-dimensional geological model according to the processed multi-modal data of the three-dimensional geological model and using a three-dimensional modeling software; the judgment coefficient of the similarity of the reflection characteristics between different pixels is further obtained by: ； wherein, represents the judgment coefficient of the similarity of the reflection characteristics between the first and the first pixels; represents the Euclidean distance between the first and the first pixels; represents the DTW distance of the reflection intensity values of the first and the first pixels in the first band; represents the mean of the deviation coefficients of the reflection characteristics of the first and the first pixels in the first band, represents the cumulative sum of the means of the deviation coefficients of the reflection characteristics of all arbitrary two pixels in the first band, and m is the preset number of the divided bands. 2.The method of claim 1, wherein, the deviation coefficient of the reflection characteristics of each pixel is further obtained by: ； in, Indicates the current band's lower number Deviation coefficient of individual pixel reflectance characteristics; Indicates the first The pixel and its first The difference in reflection characteristics of each nearest neighbor pixel in the current band; No. The first pixel The feature value of the reflection feature of each nearest neighbor pixel relative to the overall neighborhood range changes; Indicates the first The number of neighboring pixels of a given pixel. 3.The method of claim 2, wherein, mapping the reflection intensity values of each pixel at all wavelengths in the multispectral remote sensing data to a two-dimensional rectangular coordinate system with the wavelength as the horizontal coordinate and the reflection intensity value as the vertical coordinate, and obtaining the reflection characteristic curve of each pixel by curve fitting. 4.The method of claim 3, wherein, calculating the mean curve of the reflection characteristic curves of each pixel and its neighboring pixels, and calculating the absolute value of the Pearson correlation coefficient between the reflection characteristic curve of each pixel and its neighboring pixels and the mean curve at the corresponding reflection intensity value in the current wave band as the characteristic value of the reflection characteristic correlation change of each neighboring pixel relative to the overall neighborhood range. 5.The method of claim 2, wherein, calculating the difference between the reflection intensity values of each pixel and its neighboring pixels at the same wavelength, taking the absolute value, and calculating the mean value of the difference values obtained at all wavelengths in the current wave band as the difference value of the reflection characteristics of each pixel and its neighboring pixels at the current wave band. 6.The method of claim 1, wherein, In the process of clustering the pixels, the judgment coefficient of the similarity of the reflection characteristics between different pixels is used as the distance between different pixels. 7.The method of claim 1, wherein, The adjusted Gaussian smoothing parameter when the reflection intensity of the current cluster on the current waveband is subjected to Gaussian filtering The obtaining further comprises: wherein, represents the initial Gaussian smoothing parameter when the Gaussian filter is applied to the reflectance intensity of the current cluster in the current waveband, represents the normalized difference value of the deviation feature of the reflectance intensity of the pixels in the current cluster in the current waveband. 8.The method of claim 7, wherein, For the current wave band, the deviation coefficients of the reflection characteristics of all the pixels in the cluster are statistically analyzed, and the probability statistical result is curve fitted to calculate the KL divergence value between the fitted curves of the current cluster and each cluster at the current wave band, and the mean value of all the KL divergence values is used as the difference value of the reflection intensity deviation characteristics of the pixels in the current cluster at the current wave band. 9.A three-dimensional geological model construction system based on multi-modal data, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, The processor implements the steps of the three-dimensional geological model construction method based on multi-modal data according to any one of claims 1-8 when executing the computer program.

Images