Mining area soil heavy metal inversion method

Abstract

The invention discloses a mining area soil heavy metal inversion method, which comprises the following steps of performing rough matching on a satellite remote sensing image and an unmanned aerial vehicle remote sensing image to obtain a rough matching satellite subgraph; performing fine matching on the rough matching satellite sub-graph and the unmanned aerial vehicle remote sensing image to obtain a fine matching satellite sub-graph; inputting the fine matching satellite subgraph into a PGGAN model to obtain a super-resolution reconstructed image; resampling the actually measured spectral data of the mining area soil into spectral data consistent with the waveband range of the super-resolution reconstructed image; carrying out characteristic wave band selection on the resampled mining area soil actual measurement spectrum data; respectively training the PLSR model, the PCR model and the MLR model by utilizing the selected characteristic wave bands and the actually measured heavy metal content data of the mining area soil, and taking the model with the highest inversion precision as a final mining area soil heavy metal inversion model; and inverting the heavy metal content of a to-be-detected mining area by using the final mining area soil heavy metal inversion model. The precision of mining area soil heavy metal inversion can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of heavy metal inversion, and more particularly to a method for heavy metal inversion of soil in a mining area. Background Art

[0002] Coal resources are an important economic pillar of my country and belong to non-renewable energy. Therefore, the sustainable mining of coal is related to the development of my country for generations to come. With the rapid development of my country's economy, coal mining will inevitably cause varying degrees of harm to the surrounding environment. In the geological environment monitoring of mining areas, most of the current evaluation or monitoring is based on a single source data structure, and only specific research methods can be formulated according to the needs of the mining area. However, in view of the complex geographical environment of the mining area and the relatively high timeliness requirements, there is a lack of a method for accurately measuring the content of heavy metals in the soil.

[0003] Therefore, how to provide an accurate method for inversion of heavy metals in mining soil is an urgent problem that technicians in this field need to solve. Summary of the invention

[0004] In view of this, the object of the present invention is to provide a method for inversion of heavy metals in mining soil.

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

[0006] A method for inverting heavy metals in mining area soil comprises the following steps:

[0007] S1: Obtain satellite remote sensing images and drone remote sensing images of the mining area;

[0008] S2: performing coarse matching on the satellite remote sensing image and the UAV remote sensing image to obtain a coarse matching satellite sub-image;

[0009] S3: performing precise matching on the coarse matching satellite sub-image and the UAV remote sensing image to obtain a precise matching satellite sub-image;

[0010] S4: inputting the precisely matched satellite sub-image into the PGGAN model to obtain a super-resolution reconstructed image;

[0011] S5: resampling the measured spectral data of the mining area soil into spectral data consistent with the band range of the super-resolution reconstructed image;

[0012] S6: Select characteristic bands from the resampled measured spectral data of the mining area soil;

[0013] S7: using the selected characteristic bands and the measured heavy metal content data of the mining area soil to train the PLSR model, PCR model and MLR model respectively, and taking the model with the highest inversion accuracy as the final mining area soil heavy metal inversion model;

[0014] S8: Using the final mining area soil heavy metal inversion model to invert the heavy metal content in the mining area to be tested.

[0015] Preferably, S2 specifically includes:

[0016] S21: Obtain the latitude and longitude of the center point of the UAV remote sensing image;

[0017] S22: Taking the latitude and longitude position of the center point of the UAV remote sensing image as the center M and the threshold Rt as the interception length, a square satellite subimage is intercepted in the satellite remote sensing image to obtain the coarse matching satellite subimage, wherein the distance from the center M to each side of the coarse matching satellite subimage is Rt.

[0018] Preferably, S3 specifically includes:

[0019] S31: Input the coarse matching satellite sub-image into the M-VGG19 model for feature extraction to obtain the search image feature f S ;

[0020] The UAV remote sensing image is input into the M-VGG19 model for feature extraction to obtain the template image feature f T ;

[0021] S32: Calculate the search image sub-feature f s and the template image sub-feature f t The feature matching quality score function of s ∈f S ,f t ∈f T ;

[0022] S33: searching for a matching quality response maximum value region in the feature matching quality score function, wherein the matching quality response maximum value region is the precise matching satellite sub-image.

[0023] Preferably, the M-VGG19 model only retains the first 8 layers of the VGG19 model, and adds a maximum pooling layer after the third and fifth layers.

[0024] Preferably, the expression of the feature matching quality score function is:

[0025] Q(s,t)=P(t|s)·P(s|t);

[0026] Wherein, Q(s, t) represents the feature matching quality score function; P(t|s) represents the matching similarity between sub-image t in template image T and sub-image s in search image S relative to all other sub-images in template image T; P(s|t) represents the matching similarity between sub-image s in search image S and sub-image t in template image T relative to all other sub-images in search image S; wherein template image T is the UAV remote sensing image, and search image S is the coarse matching satellite sub-image.

[0027] Preferably, for a sub-image t in the template image T, there is a unique sub-image s in the search image S that matches the sub-image t, and the sub-image t is also the only sub-image in the template image T that matches the sub-image s, then P(t|s)=P(s|t)=1, and the matching quality perception function Q(s,t)=1;

[0028] For sub-image t in template image T, there are N sub-images in search image S that match sub-image t, then P(s|t) = 1 / N, P(t|s) = 1, and matching quality perception function Q(s, t) = 1 / N;

[0029] For a sub-image s in the search image S, there are M sub-images in the template image T that match the sub-image s, then P(s|t) = 1, P(t|s) = 1 / M, and the matching quality perception function Q(s, t) = 1 / M;

[0030] There are N sub-images in the search image S that match M sub-images in the template image T, then P(s|t) = 1 / N, P(t|s) = 1 / M, and the matching quality perception function Q(s, t) = 1 / NM;

[0031] There is no matching sub-image between the template image T and the search image S, and the matching quality perception function Q(s,t)=0.

[0032] Preferably, S33 obtains the precise matching satellite sub-image based on the following formula:

[0033] φ=arg max s {∑ s∈S max{Q(s,t)|t∈T}};

[0034] Among them, φ represents the maximum value of the matching quality response.

[0035] Preferably, S6 specifically includes:

[0036] S61: Calculate various spectral indicators of the resampled mining area soil measured spectral data;

[0037] S62: Calculate the correlation between each spectral index and the measured heavy metal content data of the mining area soil;

[0038] S63: Select the band with the greatest correlation as the characteristic band.

[0039] Preferably, the spectral indicators include spectrum, reciprocal of spectrum, logarithm of spectrum, logarithm of reciprocal of spectrum, reciprocal of logarithm of spectrum, first-order differential of spectrum, second-order differential of spectrum, first-order differential of reciprocal logarithm of spectrum, second-order differential of reciprocal logarithm of spectrum, first-order differential of reciprocal logarithm of spectrum and second-order differential of reciprocal logarithm of spectrum; among which, the spectrum among the 12 spectral indicators refers to the measured spectral data of mining area soil after resampled.

[0040] Preferably, S63 specifically includes: selecting the first ten bands of the first three spectral indicators with the largest correlation as the characteristic bands.

[0041] It can be seen from the above technical solutions that, compared with the prior art, the present invention discloses a method for heavy metal inversion in mining soil, which can improve the accuracy of heavy metal inversion. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying creative work.

[0043] Figure 1 A flowchart of a method for inverting heavy metals in mining soil provided by the present invention;

[0044] Figure 2 A schematic diagram of rough matching provided by the present invention;

[0045] Figure 3 This is a schematic diagram of the structure of the M-VGG19 model provided by the present invention. DETAILED DESCRIPTION

[0046] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0047] like Figure 1 As shown, the embodiment of the present invention discloses a method for inverting heavy metals in mining soil, comprising the following steps:

[0048] S1: Obtain satellite remote sensing images and drone remote sensing images of the mining area;

[0049] Specifically: remote sensing satellite images passing through the mining area are obtained and pre-processed to obtain the satellite remote sensing images; according to the satellite transit period, the UAV flight time is determined and the UAV flight route is designed, and remote sensing UAV images are obtained by the UAV and pre-processed to obtain the UAV remote sensing images;

[0050] Preprocessing mainly includes: radiation calibration, atmospheric correction, image fusion, etc.

[0051] S2: performing coarse matching on the satellite remote sensing image and the UAV remote sensing image to obtain a coarse matching satellite sub-image;

[0052] In one embodiment, S2 specifically includes:

[0053] S21: Obtain the longitude and latitude of the center point of the UAV remote sensing image; wherein the longitude and latitude of the center point of the UAV remote sensing image is the longitude and latitude position of the camera recorded at the time of imaging the UAV remote sensing image;

[0054] S22: Taking the latitude and longitude position of the center point of the UAV remote sensing image as the center M, and taking the threshold Rt as the interception length, intercept a square satellite sub-image in the satellite remote sensing image to obtain the coarse matching satellite sub-image, wherein the distance from the center M to each side of the coarse matching satellite sub-image is Rt (such as Figure 2 shown).

[0055] In this embodiment, the rough matching specifically includes the following steps:

[0056] a) Select a typical test area to be flown, and then calculate the size of the ground area covered by the satellite remote sensing image and the UAV remote sensing image based on the upper left and lower right coordinate points of the latitude and longitude of the satellite remote sensing image and the UAV remote sensing image. The purpose is to ensure that the satellite remote sensing image contains the area covered by the UAV remote sensing image.

[0057] b) Calculate the distance Lx from the center point of the UAV remote sensing image to the lateral boundary and the distance Ly from the longitudinal boundary of the satellite remote sensing image. The purpose is to select a suitable threshold Rt according to the size of the distance Lx and the distance Ly. In the present invention, the value range of the threshold Rt is 100m-200m.

[0058] c) Obtain the ground distance covered by a unit pixel of the satellite remote sensing image based on the resolution of the satellite remote sensing image.

[0059] d) Calculating the pixel coordinates (IonT, IlaT) of the center of the coarse matching satellite sub-image on the satellite remote sensing image according to the longitude and latitude of the UAV remote sensing image and the ground coverage distance of the unit pixel of the satellite remote sensing image.

[0060] e) Taking the pixel coordinates (IonT, IlaT) as the center and the threshold Rt as the interception length, a square satellite subimage is intercepted in the satellite remote sensing image to obtain the coarse matching satellite subimage, wherein the distance from the center M to each side of the coarse matching satellite subimage is Rt.

[0061] The above method can be used to obtain the coarse matching satellite sub-images of each UAV remote sensing image. The coarse matching can reduce the amount of calculation for subsequent fine matching and improve the matching efficiency.

[0062] S3: performing precise matching on the coarse matching satellite sub-image and the UAV remote sensing image to obtain a precise matching satellite sub-image;

[0063] In one embodiment, S3 specifically includes:

[0064] S31: Input the coarse matching satellite sub-image into the M-VGG19 model for feature extraction to obtain the search image feature f S ;

[0065] In this embodiment, the M-VGG19 model only retains the first 8 layers of the VGG19 model, and adds a maximum pooling layer after the third and fifth layers.

[0066] like Figure 3 As shown, the M-VGG19 model includes a first convolutional layer, a first ReLU activation function, a second convolutional layer, a second ReLU activation function, a third convolutional layer, a third ReLU activation function, a first maximum pooling layer, a fourth convolutional layer, a fourth ReLU activation function, a fifth convolutional layer, a fifth ReLU activation function, a second maximum pooling layer, a sixth convolutional layer, a sixth ReLU activation function, a seventh convolutional layer, a seventh ReLU activation function, an eighth convolutional layer and an eighth ReLU activation function, which are connected in sequence.

[0067] The expression of the ReLU activation function is f(x)=max(0,x), where x represents the feature input to the ReLU activation function.

[0068] The UAV remote sensing image is input into the M-VGG19 model for feature extraction to obtain the template image feature f T ;

[0069] S32: Calculate the search image sub-feature f s and the template image sub-feature f t The feature matching quality score function of s ∈f S ,f t ∈f T ;

[0070] It can be understood that: searching for image sub-features f sRepresents the feature corresponding to the sub-image s in the search image S; the template image sub-feature f t Represents the features corresponding to the sub-image t in the template image T;

[0071] S33: searching for a matching quality response maximum value region in the feature matching quality score function, wherein the matching quality response maximum value region is the precise matching satellite sub-image.

[0072] Specifically: S33 obtains the precise matching satellite sub-image based on the following formula;

[0073] φ=arg max s {∑ s∈S max{Q(s,t)|t∈T}};

[0074] Among them, φ represents the maximum value of the matching quality response.

[0075] In one embodiment, the expression of the feature matching quality score function is:

[0076] Q(s,t)=P(t|s)·P(s|t);

[0077]

[0078] Wherein, Q(s, t) represents the feature matching quality score function; P(t|s) represents the matching similarity between sub-image t in template image T and sub-image s in search image S relative to all other sub-images in template image T; P(s|t) represents the matching similarity between sub-image s in search image S and sub-image t in template image T relative to all other sub-images in search image S; wherein template image T is the UAV remote sensing image, and search image S is the coarse matching satellite sub-image; β represents a positive weight, ρ() represents a cosine operation, s′ represents all other sub-images in search image S except sub-image s; t′ represents all other sub-images in template image T except sub-image t.

[0079] In one embodiment, for a sub-image t in a template image T, there is a unique sub-image s in the search image S that matches the sub-image t, and the sub-image t is also the only sub-image in the template image T that matches the sub-image s, then P(t|s)=P(s|t)=1, and the matching quality perception function Q(s,t)=1;

[0080] For sub-image t in template image T, there are N sub-images in search image S that match sub-image t, then P(s|t) = 1 / N, P(t|s) = 1, and matching quality perception function Q(s, t) = 1 / N;

[0081] For a sub-image s in the search image S, there are M sub-images in the template image T that match the sub-image s, then P(s|t) = 1, P(t|s) = 1 / M, and the matching quality perception function Q(s, t) = 1 / M;

[0082] There are N sub-images in the search image S that match M sub-images in the template image T, then P(s|t) = 1 / N, P(t|s) = 1 / M, and the matching quality perception function Q(s, t) = 1 / NM;

[0083] There is no matching sub-image between the template image T and the search image S, and the matching quality perception function Q(s,t)=0.

[0084] It can be understood that: assuming that the size of the UAV remote sensing image before rough matching is The size of satellite remote sensing images is The size of the coarse matching satellite sub-image obtained by the coarse matching method of the present invention is The extracted feature dimension size is d. According to the matching process, the image feature vector cosine operation is first calculated, and the calculation result is stored in an L′×l-dimensional matrix D. ij Represents the sub-feature s of the search image i and the sub-feature t of the template image j The result of the cosine operation is O(L′ld) in computation time and O(L′l) in memory usage. Then the feature matching quality score function Q(s,t) is calculated. According to the definition of Q(s,t), an L′×l-dimensional matrix Q can also be used to store Q(s,t). The computation time complexity is O(L′l) and the memory usage complexity is O(L′l). Finally, the matching area with the highest matching quality score is found in Q(s,t). The search process is to first find the maximum value in each row of Q to obtain an L′×1 vector, and then adjust the vector dimension to S. w ′×S h ′-dimensional score matrix, and finally T w ×T h The sliding window searches for the area with the highest score in the score matrix as the best matching area, and the best matching area is the precise matching satellite sub-image.

[0085] S4: inputting the precisely matched satellite sub-image into the PGGAN model to obtain a super-resolution reconstructed image;

[0086] S5: resampling the measured spectral data of the mining area soil into spectral data consistent with the band range of the super-resolution reconstruction image (the spectral shape of the spectral data of the mining area soil measured after resampling remains unchanged, but the original band is resampled into a spectral curve consistent with the band range of the super-resolution reconstruction image, and the spectral range after resampling is consistent with the band range of the super-resolution reconstruction image);

[0087] It can be understood that the measured spectral data of the mining area soil is obtained based on the following steps:

[0088] First, we use drone remote sensing images to get a general understanding of the topography of the mining area, formulate a sampling plan, and lay out strip sampling belts according to the actual location and geographical characteristics of the site to sample the topsoil. When sampling, we use GPS to locate the sampling points and record the sampling point number, coordinates, surrounding environment and other information.

[0089] There are 31 tailings sampling points, and the sampling depth is about 5 cm. The same and uniform samples are collected at each sampling point, and the weeds and obvious large pieces of gravel in the samples are removed, put into paper sample bags with labels, and brought back to the laboratory for heavy metal content and spectral measurement. The present invention collected 100 soil samples, and after removing animal and plant substances and air-drying, crushing, and passing through a nylon sieve, they were used for chemical testing and analysis of heavy metal content, and then used for laboratory spectral data collection and acquisition.

[0090] When measuring the soil spectrum, the spectrometer sensor is placed 0.3m above the soil sample surface, the field of view is set to 8°, and a 1000W halogen lamp is placed 0.5m above the soil surface to provide nearly parallel light for the soil sample. Each soil sample is measured three times from the beginning to the end, and the average spectrum is taken as the original spectrum of the sample to reduce the observation error caused by human factors.

[0091] Subsequently, the abnormal bands are removed from the original spectrum, and then the spectral data after the abnormal bands are removed is smoothed using the weighted moving average method to obtain the measured spectral data of the mining area soil described in S5.

[0092] The Box-plot tool in SPSS software was used to detect outliers in heavy metal content, and the intercept lines Q3+1.5×IQR and Q1+1.5×IQR (where Q3 is the upper quartile, Q1 is the lower quartile, and IQR is the quartile interval) of the box plot were used as threshold ranges to eliminate outliers in the soil heavy metal content data, thereby obtaining the actual measured heavy metal content data in the soil of the mining area.

[0093] S6: Select characteristic bands from the resampled measured spectral data of the mining area soil;

[0094] In one embodiment, S6 specifically includes:

[0095] S61: Calculate various spectral indicators of the resampled mining area soil measured spectral data;

[0096] S62: Calculate the correlation between each spectral index and the measured heavy metal content data of the mining area soil;

[0097] S63: Select the band with the greatest correlation as the characteristic band.

[0098] In this embodiment, the spectral indicators include spectrum, reciprocal of spectrum, logarithm of spectrum, logarithm of reciprocal of spectrum, reciprocal of logarithm of spectrum, first-order differential of spectrum, second-order differential of spectrum, first-order differential of reciprocal logarithm of spectrum, second-order differential of reciprocal logarithm of spectrum, first-order differential of reciprocal logarithm of spectrum and second-order differential of reciprocal logarithm of spectrum; among them, the spectrum among the 12 spectral indicators refers to the measured spectral data of soil in the mining area after resampling.

[0099] In this embodiment, S63 specifically includes: selecting the first ten bands of the first three spectral indicators with the greatest correlation as the characteristic bands.

[0100] S7: The PLSR model, PCR model and MLR model were trained respectively using the selected characteristic bands and the measured heavy metal content data of the mining area soil, and the model with the highest inversion accuracy was used as the final mining area soil heavy metal inversion model.

[0101] Specifically: the selected characteristic bands are used as input, and the measured heavy metal content data of the mining area soil are used as output for polynomial fitting to obtain the trained PLSR model, PCR model and MLR model;

[0102] The present invention collected a total of 100 sets of data, each set of data including a corresponding characteristic band and a corresponding measured heavy metal content data, 70 sets of data were used as training sets, and 30 sets of data were used as validation sets to train the PLSR model, PCR model, and MLR model respectively;

[0103] The present invention adopts the determination coefficient R 2 And root mean square error RMSE are used to test the inversion accuracy of PLSR model, PCR model and MLR model.

[0104] S8: Using the final mining area soil heavy metal inversion model to invert the heavy metal content in the mining area to be tested.

[0105] Specifically: real-time satellite remote sensing images of the mining area to be monitored are obtained, and characteristic bands corresponding to the mining area to be monitored in the real-time satellite remote sensing images are input into the final mining area soil heavy metal inversion model to obtain real-time heavy metal content data of the mining area to be monitored.

[0106] In this specification, each embodiment is described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the embodiments can be referred to each other. For the device disclosed in the embodiment, since it corresponds to the method disclosed in the embodiment, the description is relatively simple, and the relevant parts can be referred to the method part.

[0107] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one 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 invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A method for inversion of heavy metals in mining soil, characterized in that: The following steps are involved: S1: Obtain satellite remote sensing images and drone remote sensing images of the mining area; S2: performing coarse matching on the satellite remote sensing image and the UAV remote sensing image to obtain a coarse matching satellite sub-image; S3: performing precise matching on the coarse matching satellite sub-image and the UAV remote sensing image to obtain a precise matching satellite sub-image; S4: inputting the precisely matched satellite sub-image into the PGGAN model to obtain a super-resolution reconstructed image; S5: resampling the measured spectral data of the mining area soil into spectral data consistent with the band range of the super-resolution reconstructed image; S6: Select characteristic bands from the resampled measured spectral data of the mining area soil; S7: using the selected characteristic bands and the measured heavy metal content data of the mining area soil to train the PLSR model, PCR model and MLR model respectively, and taking the model with the highest inversion accuracy as the final mining area soil heavy metal inversion model; S8: Using the final mining area soil heavy metal inversion model to invert the heavy metal content in the mining area to be tested.

2. A method for inversion of heavy metals in mining soil according to claim 1, characterized in that: S2 specifically includes: S21: Obtain the latitude and longitude of the center point of the UAV remote sensing image; S22: Taking the latitude and longitude position of the center point of the UAV remote sensing image as the center M and the threshold Rt as the interception length, a square satellite subimage is intercepted in the satellite remote sensing image to obtain the coarse matching satellite subimage, wherein the distance from the center M to each side of the coarse matching satellite subimage is Rt.

3. The method for inversion of heavy metals in mining soil according to claim 1, characterized in that: S3 specifically includes: S31: Input the coarse matching satellite sub-image into the M-VGG19 model for feature extraction to obtain the search image feature f S ; The UAV remote sensing image is input into the M-VGG19 model for feature extraction to obtain the template image feature f T ; S32: Calculate the search image sub-feature f s and the template image sub-feature f t The feature matching quality score function of s ∈f S ,f t ∈f T ; S33: searching for a matching quality response maximum value region in the feature matching quality score function, wherein the matching quality response maximum value region is the precise matching satellite sub-image.

4. A method for inversion of heavy metals in mining soil according to claim 3, characterized in that: The M-VGG19 model only retains the first 8 layers of the VGG19 model and adds maximum pooling layers after the third and fifth layers.

5. The method for inversion of heavy metals in mining soil according to claim 3, characterized in that: The expression of the feature matching quality score function is: Q(s,t)=P(t|s)·P(s|t); Wherein, Q(s, t) represents the feature matching quality score function; P(t|s) represents the matching similarity between sub-image t in template image T and sub-image s in search image S relative to all other sub-images in template image T; P(s|t) represents the matching similarity between sub-image s in search image S and sub-image t in template image T relative to all other sub-images in search image S; wherein template image T is the UAV remote sensing image, and search image S is the coarse matching satellite sub-image.

6. A method for inversion of heavy metals in mining soil according to claim 5, characterized in that: For a sub-image t in the template image T, there is a unique sub-image s in the search image S that matches the sub-image t, and the sub-image t is also the only sub-image in the template image T that matches the sub-image s, then P(t|s)=P(s|t)=1, and the matching quality perception function Q(s,t)=1; For sub-image t in template image T, there are N sub-images in search image S that match sub-image t, then P(s|t) = 1 / N, P(t|s) = 1, and matching quality perception function Q(s, t) = 1 / N; For a sub-image s in the search image S, there are M sub-images in the template image T that match the sub-image s, then P(s|t) = 1, P(t|s) = 1 / M, and the matching quality perception function Q(s, t) = 1 / M; There are N sub-images in the search image S that match M sub-images in the template image T, then P(s|t) = 1 / N, P(t|s) = 1 / M, and the matching quality perception function Q(s, t) = 1 / NM; There is no matching sub-image between the template image T and the search image S, and the matching quality perception function Q(s,t)=0.

7. The method for inversion of heavy metals in mining soil according to claim 3, characterized in that: S33 obtains the precise matching satellite sub-image based on the following formula; φ=argmax s {∑ s∈S max{Q(s,t)|t∈T}}; Among them, φ represents the maximum value of the matching quality response.

8. The method for inversion of heavy metals in mining soil according to claim 6, characterized in that: S6 specifically includes: S61: Calculate various spectral indicators of the resampled mining area soil measured spectral data; S62: Calculate the correlation between each spectral index and the measured heavy metal content data of the mining area soil; S63: Select the band with the greatest correlation as the characteristic band.

9. A method for inversion of heavy metals in mining soil according to claim 8, characterized in that: The spectral indicators include spectrum, reciprocal of spectrum, logarithm of spectrum, logarithm of reciprocal of spectrum, reciprocal of logarithm of spectrum, first-order differential of spectrum, second-order differential of spectrum, first-order differential of reciprocal logarithm of spectrum, second-order differential of reciprocal logarithm of spectrum, first-order differential of reciprocal logarithm of spectrum and second-order differential of reciprocal logarithm of spectrum; among them, the spectrum among the 12 spectral indicators refers to the measured spectral data of soil in the mining area after resampling.

10. The method for inversion of heavy metals in mining soil according to claim 8, characterized in that: S63 specifically includes: selecting the first ten bands of the first three spectral indicators with the largest correlation as the characteristic bands.

Images