A 3D reconstruction method for multiple magnetic targets based on probability extreme value weighted function

Abstract

The present invention relates to a method for three-dimensional reconstruction of multiple magnetic targets based on a probability extreme value weighted function, and belongs to the field of magnetic detection technology. The method comprises: S1: acquiring magnetic gradient tensor data in an observation plane, dividing the inversion space into cubic grid units, and calculating a kernel matrix for forward modeling of the magnetic gradient tensor; S2: using a correlation imaging method to calculate a normalized mutual correlation coefficient C between each grid and the measured magnetic gradient tensor data ab,q ; S3: Calculate the grid position where the maximum value of the correlation coefficient is located, and obtain the maximum value position of the probability of the magnetic body; S4: Construct a probability extreme value weighting function. Weight each grid according to the distance from the probability maximum value position, and the grid close to the probability maximum value position is given a lower weight; S5: Calculate the distance from each grid to the probability maximum value position, and calculate the re-weighted distance; S6: Add the probability extreme value weighting function to the inversion objective function for solution, and then use the visualization method to obtain the three-dimensional reconstruction result.

Description

Technical Field

[0001] The invention belongs to the technical field of magnetic detection and relates to a three-dimensional reconstruction method of multiple magnetic targets based on a probability extreme value weighted function. Background Art

[0002] The 3D reconstruction technology of magnetic anomaly is to reconstruct the physical parameters and spatial distribution of magnetic targets through the magnetic anomaly values ​​caused by underground or underwater magnetic targets. This technology can effectively detect magnetic targets and is widely used in anti-submarine, unexploded ordnance detection, seabed structure detection, underground ferromagnetic pipeline detection, etc., and has important practical value. Especially for the detection of underwater targets, the reconstructed 3D contour can assist target identification and further enhance the ability to explore the marine environment. At present, the research on 3D reconstruction of multiple magnetic targets is still in its early stages. Similar magnetic inversion techniques are mainly studied in geological exploration, including morphological inversion and physical inversion. Morphological inversion requires prior knowledge of the magnetic susceptibility, which leads to certain limitations in the reconstruction of unknown magnetic targets; and due to the errors in magnetic survey data and the lack of prior information, the traditional physical inversion method is not ideal.

[0003] Therefore, there is an urgent need for a method that can accurately achieve three-dimensional reconstruction in multi-target situations. Summary of the invention

[0004] In view of this, the purpose of the present invention is to provide a three-dimensional reconstruction method for multiple magnetic targets based on a probability extreme value weighted function, which aims to address the shortcomings of the existing three-dimensional reconstruction method for multiple ferromagnetic targets with uneven spatial distribution and solve the problems of low accuracy, poor resolution and unclear contours in three-dimensional reconstruction when there are multiple targets due to errors in magnetic measurement data and lack of prior information.

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

[0006] A method for three-dimensional reconstruction of multiple magnetic targets based on a probability extreme value weighted function specifically comprises the following steps:

[0007] S1: Obtain the magnetic gradient tensor data in the observation plane, divide the inversion space into cubic grid cells, and calculate the kernel matrix of the magnetic gradient tensor forward modeling;

[0008] S2: Calculate the theoretical magnetic gradient tensor generated by the magnetic dipole on each grid cell, and use the correlation imaging method to calculate the normalized cross-correlation coefficient C between it and the measured magnetic gradient tensor data ab,q ;

[0009] S3: The 26-neighborhood method is used to calculate the grid position where the maximum value of the correlation coefficient is located. Since the correlation coefficient represents the probability of the existence of the magnetic body, the maximum value position of the probability of the existence of the magnetic body is obtained;

[0010] S4: Take the probability maximum position as prior information and construct a probability maximum weighting function. Weight each grid according to its distance to the probability maximum position, and the grid close to the probability maximum position is given a lower weight, so that the inversion result is constrained to the space near the probability maximum position;

[0011] S5: If the magnetic body is unevenly distributed in the three directions of x, y, and z, when calculating the distance from each grid to the probability maximum position, different weights should be given to the three directions based on prior information, that is, the reweighted distance should be calculated;

[0012] S6: adding the probability extreme value weighting function to the inversion objective function for solution, and then using the visualization method to obtain the three-dimensional reconstruction result;

[0013] Furthermore, in step S1, assuming that the sides of the cube are a, b, and c, and the center coordinates are (x0, y0, z0), the three-component expression of the magnetic induction intensity of the magnetic cube at any observation point is:

[0014]

[0015]

[0016]

[0017] Where (x, y, z) is the coordinate of the observation point in the measurement plane, μ0 is the magnetic permeability in vacuum, J is the total magnetization intensity of the cube, L0, M0, N0 are the direction cosines of the total magnetization intensity, (ξ, η, ζ) is the coordinate of the source point of the cube field, r = [(ξ-x) 2 +(η-y) 2 +(ζ-z) 2 ] 1 / 2 , represents the distance between the observation point and any point in the cube;

[0018] By differentiating the three-component expressions of magnetic induction intensity in different directions, the corresponding expressions of magnetic gradient tensor components can be obtained.

[0019] Furthermore, in step S1, due to the superposition of potential fields, the magnetic gradient tensor component of the entire model for any observation point is equal to the linear superposition of the magnetic gradient tensor components generated by the grid units constituting the model at that point.

[0020] Further, in step S2, the normalized correlation coefficient C ab,q The calculation formula is:

[0021]

[0022] Among them, B ab(x i ,y i ,z i ) is a component of the measured magnetic gradient tensor at any point in the measurement area, B ab,q (x i ,y i ,z i ) is a component of the magnetic gradient tensor generated by the underground magnetic dipole q at any point in the survey area, a, b = x, y, z, N is the total number of observation points in the survey area.

[0023] Further, in step S3, the mutual correlation coefficient C ab,q Indicates the possibility of the existence of a dipole at the q point. The mutual correlation coefficient C ab,q Indicates the possibility that a magnetic substance exists at this q point. ab,q It is between -1 and 1. The closer it is to 1, the greater the probability of the existence of a magnetic body. Therefore, the maximum value position of the correlation coefficient reflects the location of the maximum value of the probability of the existence of a magnetic body. If there are n magnetic targets, n such maximum probability positions can be obtained accordingly.

[0024] Further, in step S4, the probability extreme value weighting function is constructed using the position coordinates of the probability maximum value of the magnetic body, and its calculation expression is:

[0025]

[0026] Where z0 is the height of the observation plane relative to the horizontal reference plane; z j is the central depth of the j-th cubic unit; n represents the number of maximum values ​​of the probability of the existence of magnetic bodies; r 1j To r nj Represents the distance from the center of the jth grid unit to the 1st to nth probability maximum positions; r0 is a small stable constant. In order to avoid the model being too concentrated at the probability maximum position, r0 is taken as the side length of a grid unit;

[0027] Further, in step S5, the reweighted distance r from the center of the jth grid unit to the probability maximum position ij The calculation expression for (i=1, 2, ..., n) is:

[0028]

[0029] Where (x i ,y i , z i ) is the coordinate of the position of the maximum probability of the ith estimated magnetic body, (x j ,y j , z j ) is the coordinate of the center of the jth grid cell, a x , ay , a z They are the prior proportion weights in the three directions respectively. If there is no prior information, the default value is 1.

[0030] Further, in step S6, the expression of the inversion objective function is:

[0031]

[0032] The first term is the data fitting function, which represents the degree of fit between the predicted data and the measured data. d represents the observed data, m is the model physical property parameter to be solved, and G is the mapping matrix from the physical property space to the data space, also known as the kernel matrix;

[0033] The second term is the model objective function, which is used to reduce the degrees of freedom in the inversion to obtain a unique solution. Among them, m0 is the reference model, which can be set to a zero-value vector if there is no corresponding prior information; W m is the model weighting matrix, which is composed of different types of constraint matrices, such as the probability extreme value weighting matrix and the roughness matrix; β is the regularization parameter, which is used to adjust the weight between the data fitting objective function and the model objective function.

[0034] Theoretically, the minimization of φ(m) is The solution m under these conditions is the optimal solution of the objective function.

[0035]

[0036] Simplifying the above formula, we can get the inversion target equation:

[0037] (G T G+βW m T W m )m=(G T d+βW m T W m m0)

[0038] Further, in step S6, the inversion target equation is solved using the conjugate gradient method according to the measured magnetic gradient tensor components to estimate the magnetic susceptibility of each grid cell.

[0039] Further, in step S6, the specific steps of achieving high-precision three-dimensional reconstruction for multiple magnetic targets include:

[0040] The optimal solution m of the physical property parameters is rearranged into a three-dimensional matrix according to the spatial position of the grid cells;

[0041] 40% of the local extreme value of the optimal solution m is used as the segmentation threshold, and the grid cell values ​​with magnetic susceptibility less than this threshold are set to 0;

[0042] The three-dimensional matrix is ​​processed using visualization technology to achieve high-precision three-dimensional reconstruction of multiple magnetic targets.

[0043] The beneficial effect of the present invention is that compared with the prior art, it is the first time to use the correlation imaging method to extract prior information from the magnetic gradient tensor components, and construct a probability extreme value weighting function to constrain the inversion target equation, which can not only eliminate the influence of the background field, but also reasonably allocate the weights of the grid units, so that the imaging results shrink to the probability maximum position and the contour is clearer, while improving the longitudinal and lateral resolution. Compared with the current three-dimensional inversion method, the imaging results are more accurate, the magnetic parameters and three-dimensional contours of the reconstructed abnormal body model are closer to the real target, and the inversion effect has been significantly improved, thereby achieving high-precision three-dimensional reconstruction of multiple magnetic targets.

[0044] Other advantages, objectives and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained through the following description. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be described in detail below in conjunction with the accompanying drawings, wherein:

[0046] Figure 1 It is a flow chart of the method for three-dimensional reconstruction of multiple magnetic targets based on probability extreme value weighting function of the present invention;

[0047] Figure 2 is the magnetic body distribution of the theoretical model and the magnetic gradient tensor component Bzz calculated by forward modeling;

[0048] Figure 3 It is the cross-sectional diagram of the related imaging results of the theoretical model in the y=0 plane;

[0049] Figure 4 This is the horizontal 3D reconstruction result of the theoretical model; Figure 4 (a) weighting for using extreme probability values; Figure 4 (b) weighting the extreme value of the probability of non-use;

[0050] Figure 5 This is the result of the vertical 3D reconstruction of the theoretical model; Figure 5 (a) weighting for using extreme probability values; Figure 5 (b) Weighting for the extreme value of unused probability. DETAILED DESCRIPTION

[0051] The following describes the embodiments of the present invention by specific examples, and those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be noted that the illustrations provided in the following embodiments only illustrate the basic concept of the present invention in a schematic manner, and the following embodiments and features in the embodiments can be combined with each other without conflict.

[0052] Among them, the drawings are only used for illustrative explanations, and they only represent schematic diagrams rather than actual pictures, and should not be understood as limitations on the present invention. In order to better illustrate the embodiments of the present invention, some parts of the drawings may be omitted, enlarged or reduced, and do not represent the size of actual products. For those skilled in the art, it is understandable that some well-known structures and their descriptions in the drawings may be omitted.

[0053] The same or similar numbers in the drawings of the embodiments of the present invention correspond to the same or similar parts; in the description of the present invention, it should be understood that if the terms "upper", "lower", "left", "right", "front", "rear", etc. indicate the orientation or position relationship, they are based on the orientation or position relationship shown in the drawings, which is only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operate in a specific orientation. Therefore, the terms describing the position relationship in the drawings are only used for illustrative purposes and cannot be understood as limiting the present invention. For ordinary technicians in this field, the specific meanings of the above terms can be understood according to specific circumstances.

[0054] See also Figure 1 The present invention provides a method for three-dimensional reconstruction of multiple magnetic targets based on a probability extreme value weighted function, which specifically includes the following steps:

[0055] S1: Obtain the magnetic gradient tensor data in the observation plane, divide the inversion space into cubic grid units, and calculate the kernel matrix of the magnetic gradient tensor forward modeling; specifically including:

[0056] The inversion space is divided into cubic grid units, and magnetic targets of various complex shapes are simulated by combining grid units. Assuming that the sides of the cube are a, b, and c, and the center coordinates are (x0, y0, z0), the three-component expression of the magnetic induction intensity of the magnetic cube at any observation point is:

[0057]

[0058]

[0059]

[0060] Where (x, y, z) is the coordinate of the observation point in the measurement plane, μ0 is the magnetic permeability in vacuum, J is the total magnetization intensity of the cube, L0, M0, N0 are the direction cosines of the total magnetization intensity, (ξ, η, ζ) is the coordinate of the source point of the cube field, r = [(ξ-x) 2 +(η-y) 2 +(ζ-z) 2 ] 1 / 2 , represents the distance between the observation point and any point in the cube;

[0061] By differentiating the three-component expressions of magnetic induction intensity in different directions, the corresponding expressions of magnetic gradient tensor components can be obtained.

[0062] Due to the superposition of potential fields, the magnetic gradient tensor component of the entire model for any observation point is equal to the linear superposition of the magnetic gradient tensor components generated at that point by the grid units that make up the model.

[0063] S2: Use the correlation imaging method to calculate the normalized cross-correlation coefficient C between the theoretical magnetic gradient tensor and the observed magnetic gradient tensor generated by the magnetic dipole at each grid cell ab,q ; Specifically include:

[0064] The theoretical magnetic gradient tensor generated by the magnetic dipole on each grid cell is calculated using the forward operator, and the normalized correlation coefficient C between it and the observed magnetic gradient tensor is calculated using the correlation imaging method. ab,q , the calculation formula is:

[0065]

[0066] Among them, B ab (x i ,y i ,z i ) is a component of the measured magnetic gradient tensor at any point in the measurement area, B ab,q (x i ,y i ,z i ) is a component of the magnetic gradient tensor generated by the underground magnetic dipole q at any point in the survey area, a, b = x, y, z, N is the total number of observation points in the survey area.

[0067] S3: The position of the calculated maximum value of the correlation coefficient, that is, the position of the maximum value of the probability of the existence of the magnetic body; specifically including:

[0068] The correlation coefficient C ab,qArrange them into a three-dimensional matrix for visualization. Then compare the mutual correlation coefficient of each grid with the mutual correlation coefficient of its 26 adjacent grids. If it is the maximum value, the coordinates of the grid are the location of a magnetic body maximum probability. If there are n magnetic targets, n such maximum probability locations can be obtained accordingly.

[0069] S4: Using the obtained probability maximum position to construct a probability extreme value weighting function as a constraint for the inversion objective equation; specifically including:

[0070] The probability extreme value weighting function is constructed using the position coordinates of the maximum value of the probability of the existence of the magnetic body. Its calculation expression is:

[0071]

[0072] Where z0 is the height of the observation plane relative to the horizontal reference plane; z j is the central depth of the j-th cubic unit; n represents the number of maximum values ​​of the probability of the existence of magnetic bodies; r 1j To r nj Represents the distance from the center of the jth grid unit to the 1st to nth probability maximum positions; r0 is a small stable constant. In order to avoid the model being too concentrated at the probability maximum position, r0 is taken as the side length of a grid unit;

[0073] S5: Give different weights to the three directions x, y, and z according to the prior information, that is, calculate the reweighted distance; specifically, it includes:

[0074] The reweighted distance r from the center of the jth grid cell to the probability maximum position ij The calculation expression for (i=1, 2, ..., n) is:

[0075]

[0076] Where (x i ,y i , z i ) is the coordinate of the maximum probability position of the i-th magnetic body, (x j ,y j , z j ) is the coordinate of the center of the jth grid cell, a x , a y , a z They are the prior proportion weights in the three directions respectively. If there is no prior information, the default value is 1.

[0077] S6: Using the conjugate gradient method to solve the inversion equation to estimate the magnetic susceptibility m of each grid cell; specifically including:

[0078] The probability extreme value weighting function is introduced to constrain the model and construct the inversion objective function, which is expressed as:

[0079]

[0080] The first term is the data fitting function, which represents the degree of fit between the predicted data and the measured data. d represents the observed data, m is the model physical property parameter to be solved, and G is the mapping matrix from the physical property space to the data space, also known as the kernel matrix;

[0081] The second term is the model objective function, which is used to reduce the degrees of freedom in the inversion to obtain a unique solution. Among them, m0 is the reference model, which can be set to a zero-value vector if there is no corresponding prior information; Wm is the model weighting matrix, which is composed of different types of constraint matrices, such as the probability extreme value weighting matrix and the roughness matrix; β is the regularization parameter, which is used to adjust the weight between the data fitting objective function and the model objective function.

[0082] Theoretically, the minimization of φ(m) is The solution m under these conditions is the optimal solution of the objective function.

[0083]

[0084] Simplifying the above formula, we can get the inversion target equation:

[0085] (G T G+βW m T W m )m=(G T d+βW m T W m m0)

[0086] The measured magnetic gradient tensor components are substituted into the above inversion target equation and the conjugate gradient method is used to solve the inversion target equation to estimate the magnetic susceptibility of each grid cell.

[0087] The three-dimensional contour of the magnetic target is depicted using visualization technology to achieve high-precision three-dimensional reconstruction of multiple magnetic targets; the specific steps are as follows:

[0088] The optimal solution m of the physical property parameters is rearranged into a three-dimensional matrix according to the spatial position of the grid cells;

[0089] 40% of the local extreme value of the optimal solution m is used as the segmentation threshold, and the grid cell values ​​with magnetic susceptibility less than this threshold are set to 0;

[0090] The three-dimensional matrix is ​​processed using visualization technology to achieve high-precision three-dimensional reconstruction of multiple magnetic targets.

[0091] Example:

[0092] Taking the theoretical model as an example, the magnetic body distribution and its gradient tensor component Bzz are as follows: Figure 2 As shown, the spatial distribution of magnetic susceptibility of the abnormal body model is reconstructed in three dimensions by a three-dimensional reconstruction method based on a probability extreme value weighting function. For other application scenarios, the actual method for obtaining magnetic gradient tensor data can be implemented as in step 1. Of course, the magnetic gradient tensor components of the target detection area can also be obtained by other methods, and the examples of the present invention do not specifically limit this.

[0093] The specific steps are as follows:

[0094] Step 1: Divide the inversion space into cubic grid units, simulate magnetic targets of various complex shapes through the combination of grid units, complete forward modeling, and calculate the corresponding mapping matrix.

[0095] Step 2: Through forward calculation, the magnetic gradient tensor components of the theoretical model are obtained, such as Figure 2 As shown. The theoretical model consists of two relatively independent cuboids. The magnetic susceptibility of the top cuboid model is 2.5, and the center coordinates are (15m, 0m, 10m). The magnetic susceptibility of the bottom cuboid model is 1, and the center coordinates are (-15m, 0m, 20m).

[0096] Step 3: Use the forward operator to calculate the theoretical magnetic gradient tensor generated by the magnetic dipole on each grid cell, and use the correlation imaging method to calculate the normalized mutual correlation coefficient C between it and the observed magnetic gradient tensor ab,q , the calculation formula is:

[0097]

[0098] Among them, B ab (x i ,y i ,z i ) is a component of the measured magnetic gradient tensor at any point in the measurement area, B ab,q (x i ,y i ,z i ) is a component of the magnetic gradient tensor generated by the underground magnetic dipole q at any point in the survey area, a, b = x, y, z, N is taken as 441, which is the total number of observation points in the survey area.

[0099] Step 4: Use the correlation coefficient C ab,q Indicates the possibility of the existence of a dipole at the q point. The mutual correlation coefficients of all the cube center positions in the inversion space are calculated in turn, which are used as prior information to describe the distribution of underground magnetic targets, and the maximum value of the correlation coefficient is used to represent the location of the maximum value of the probability of the existence of the magnetic body. The result of the correlation coefficient at the y=0 section is as follows Figure 3 As shown, the grid coordinates where the maximum value of the correlation coefficient is calculated are Czz1 = (15, 0, 10) and Czz2 = (-15, 0, 18), indicating that these two grids are the locations where the probability of the existence of magnetic bodies is the maximum.

[0100] Step 5: Use the position coordinates of the maximum probability value of the magnetic body to construct a probability extreme value weighting function, and its calculation expression is:

[0101]

[0102] Where z0 is the height of the observation plane relative to the horizontal reference plane; z j is the central depth of the j-th cubic unit; n is the number of maximum values ​​of the probability of the existence of magnetic bodies, which is 2 in this case; r 1j To r nj Represents the distance from the center of the jth grid unit to the 1st to nth probability maxima; r0 is a small stable constant. In order to avoid the model being too concentrated at the probability maximum, r0 is taken as the side length of a grid unit, which is 2m in this case;

[0103] Step 6: Calculate the reweighted distance based on the prior information, the reweighted distance r from the center of the jth grid cell to the probability maximum position ij The calculation expression for (i=1, 2, ..., n) is:

[0104]

[0105] Where (x i ,y i , z i ) is the coordinate of the position of the ith probability maximum, (x j ,y j , z j ) is the coordinate of the center of the jth grid cell, a x , a y , a z They are the prior proportion weights in the three directions respectively. If there is no prior information, the default value is 1.

[0106] The probability extreme value weighting function can compensate for the spatial attenuation of the kernel matrix G to a certain extent by assigning different weights to the grid cells, and constrain the imaging results to shrink to the probability maximum position.

[0107] Step 7: Introduce the probability extreme value weighting function to constrain the model and construct the inversion objective function, which is expressed as:

[0108]

[0109] The first term is the data fitting function, which represents the degree of fit between the predicted data and the measured data. d represents the observed data, m is the model physical property parameter to be solved, and G is the mapping matrix from the physical property space to the data space, also known as the kernel matrix;

[0110] The second term is the model objective function, which is used to reduce the degrees of freedom in the inversion to obtain a unique solution. Among them, m0 is the reference model, which can be set to a zero-value vector if there is no corresponding prior information; Wm is the model weighting matrix, which is composed of different types of constraint matrices, such as the probability extreme value weighting matrix and the roughness matrix; β is the regularization parameter, which is used to adjust the weight between the data fitting objective function and the model objective function, and is determined by the L curve method.

[0111] Theoretically, the minimization of φ(m) is The solution m under these conditions is the optimal solution of the objective function.

[0112]

[0113] Simplifying the above formula, we can get the inversion target equation:

[0114] (G T G+βW m T W m )m=(G T d+βW m T W m m0)

[0115] The magnetic gradient tensor components of the theoretical model are substituted into the above inversion objective equation, and the inversion objective equation is solved using the conjugate gradient method to estimate the magnetic susceptibility of each grid cell.

[0116] Step 8: Rearrange the optimal solution m of the physical parameters into a three-dimensional matrix according to the spatial position of the grid cells. 40% of the local extreme value of the optimal solution m is used as the segmentation threshold, and the grid cell values ​​with magnetic susceptibility less than the threshold are set to 0. Finally, the three-dimensional matrix is ​​processed using visualization technology to achieve high-precision three-dimensional reconstruction of the theoretical model, such as Figure 4 and Figure 5 shown. Figure 4 This is the horizontal 3D reconstruction result of the theoretical model; Figure 4 (a) weighting for using extreme probability values; Figure 4 (b) weighting the extreme value of the probability of non-use; Figure 5 This is the result of the vertical 3D reconstruction of the theoretical model; Figure 5 (a) weighting for using extreme probability values; Figure 5 (b) Weighting for the extreme value of unused probability.

[0117] It can be found that after being processed by the three-dimensional reconstruction method based on the probability extreme value weighted function, the imaging results are more accurate, the magnetic parameters and three-dimensional contours of the reconstructed anomaly model are closer to the real target, and the inversion effect has been significantly improved.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solution of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solution of the present invention can be modified or replaced by equivalents without departing from the purpose and scope of the technical solution, which should be included in the scope of the claims of the present invention.

Claims

1. A method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function, characterized in that: The method comprises the following steps: S1: Obtain the magnetic gradient tensor data in the observation plane, divide the inversion space into cubic grid cells, and calculate the kernel matrix of the magnetic gradient tensor forward modeling; S2: Calculate the theoretical magnetic gradient tensor generated by the magnetic dipole on each grid cell, and use the correlation imaging method to calculate the normalized cross-correlation coefficient C between it and the measured magnetic gradient tensor data ab,q ; S3: Calculate the grid position of the maximum value of the correlation coefficient using the 26-neighborhood method. The correlation coefficient represents the probability of the existence of the magnetic body, and obtain the maximum value position of the probability of the existence of the magnetic body. S4: Taking the probability maximum position as prior information, construct a probability maximum weighting function; weighting is performed according to the distance from each grid to the probability maximum position, and grids close to the probability maximum position are assigned lower weights, so that the inversion results are constrained to the space near the probability maximum position; S5: If the magnetic body is unevenly distributed in the three directions of x, y, and z, when calculating the distance from each grid to the probability maximum position, different weights should be given to the three directions based on prior information, that is, the reweighted distance should be calculated; S6: Add the probability extreme value weighting function to the inversion objective function for solution, and then use the visualization method to obtain the three-dimensional reconstruction result.

2. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 1, characterized in that: In S1, assuming that the sides of the cube are a, b, and c, and the center coordinates are (x0, y0, z0), the three-component expression of the magnetic induction intensity of the magnetic cube at any observation point is: Where (x, y, z) is the coordinate of the observation point in the measurement plane, μ0 is the magnetic permeability in vacuum, J is the total magnetization intensity of the cube, L0, M0, N0 are the direction cosines of the total magnetization intensity, (ξ, η, ζ) is the coordinate of the source point of the cube field, r = [(ξ-x) 2 +(η-y) 2 +(ζ-z) 2 ] 1 / 2 , represents the distance between the observation point and any point in the cube; The corresponding expressions of magnetic gradient tensor components are obtained by differentiating the three-component expressions of magnetic induction intensity in different directions.

3. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 2, characterized in that: In S1, according to the superposition property of the potential field, the magnetic gradient tensor component of the entire model for any observation point is equal to the linear superposition of the magnetic gradient tensor components generated at that point by the grid units constituting the model.

4. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 3, characterized in that: In S2, the normalized correlation coefficient C ab,q The calculation formula is: Among them, B ab (x i ,y i ,z i ) is a component of the measured magnetic gradient tensor at any point in the measurement area, B ab,q (x i ,y i ,z i ) is a component of the magnetic gradient tensor generated by the underground magnetic dipole q at any point in the survey area, a, b = x, y, N is the total number of observation points in the survey area.

5. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 4, characterized in that: In S3, the mutual correlation coefficient C ab,q Indicates the possibility of the existence of a magnetic body at the q point; C ab,q It is between -1 and 1. The closer it is to 1, the greater the probability of the existence of a magnetic body. The maximum position of the correlation coefficient reflects the location of the maximum probability of the existence of a magnetic body. If there are n magnetic targets, then n such maximum probability positions are obtained.

6. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 5, characterized in that: In S4, the probability extreme value weighting function is constructed using the maximum value position coordinates of the probability of existence of the magnetic body, and its calculation expression is: Where z0 is the height of the observation plane relative to the horizontal reference plane; z j is the central depth of the jth cubic unit; n is the number of maximum probability values ​​of magnetic bodies; r 1j To r nj Represents the distance from the center of the jth grid unit to the 1st to nth probability maximum positions; r0 is a small stable constant. To avoid the model being too concentrated at the probability maximum position, r0 is taken as the side length of a grid unit.

7. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 6, characterized in that: In S5, the reweighted distance r from the center of the jth grid unit to the probability maximum position ij The calculation expression is: Where (x i ,y i , z i ) is the coordinate of the probability maximum of the i-th magnetic body, (x j ,y j , z j ) is the coordinate of the center of the jth grid cell, a x , a y , a z are the prior proportion weights in three directions respectively. If there is no prior information, the default values ​​are all 1, i=1, 2,…, n.

8. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 7, characterized in that: In S6, the expression of the inversion objective function is: The first term is the data fitting function, which represents the degree of fit between the predicted data and the measured data; d represents the observed data, m is the physical property parameter of the model to be solved, and G is the mapping matrix from the physical property space to the data space, which is called the kernel matrix; The second term is the model objective function, which is used to reduce the degrees of freedom in the inversion to obtain a unique solution; where m0 is the reference model, and is set to a zero-value vector if there is no corresponding prior information; W m is the model weighting matrix, which is composed of different types of constraint matrices, including the probability extreme value weighting matrix and the roughness matrix; β is the regularization parameter, which is used to adjust the weight between the data fitting objective function and the model objective function; Theoretically, the minimization of φ(m) is The solution m under the condition is the optimal solution of the objective function; Simplifying the above formula, we can get the inversion target equation: (G T G+βW m T W m )m=(G T d+βW m T W m m0)。 9. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 1, characterized in that: In S6, the inversion target equation is solved by using the conjugate gradient method according to the measured magnetic gradient tensor components to estimate the magnetic susceptibility of each grid cell.

10. The method for 3D reconstruction of multiple magnetic targets based on probability extreme value weighted function according to claim 1, characterized in that: In S6, the specific steps of achieving high-precision three-dimensional reconstruction for multiple magnetic targets include: The optimal solution m of the physical property parameters is rearranged into a three-dimensional matrix according to the spatial position of the grid cells; 40% of the local extreme value of the optimal solution m is used as the segmentation threshold, and the grid cell values ​​with magnetic susceptibility less than this threshold are set to 0; The three-dimensional matrix is ​​processed using visualization technology to achieve high-precision three-dimensional reconstruction of multiple magnetic targets.

Images