Weak magnetic contour imaging method and system based on tensor feature value extraction

Abstract

The invention discloses a weak magnetic contour imaging method and system based on tensor eigenvalue extraction, and relates to the technical field of nondestructive testing imaging, and the method comprises the following steps: obtaining original magnetic field data, and constructing a magnetic field space vector grid; calculating a horizontal gradient component by using each node of the magnetic field space vector grid, calculating a vertical gradient component based on a magnetic field divergence-free physical constraint condition, and assembling a real symmetric magnetic gradient tensor matrix; eigenvalue decomposition is carried out on the magnetic gradient tensor matrix, three eigenvalues are solved, and the eigenvalue with the maximum absolute value is selected as a magnetic field main curvature characteristic quantity; and calculating local window statistical characteristics of the magnetic field main curvature characteristic quantity, and performing dynamic gain adjustment on the enhanced magnetic field main curvature characteristic quantity to generate a clear defect contour image. The method is completely based on spatial domain calculation, signal breakage is effectively eliminated, background noise is suppressed, and optical-level clear contour imaging of micro defects is achieved.

Description

Technical Field

[0001] This invention relates to the field of nondestructive testing imaging technology, and in particular to a weak magnetic contour imaging method and system based on tensor eigenvalue extraction and adaptive local enhancement. Background Technology

[0002] In industrial nondestructive testing, weak magnetic field testing technology has significant application value because it eliminates the need for bulky magnetization devices. Existing technologies typically utilize magnetic sensor arrays to acquire data and identify defects based on the modulus of the magnetic gradient tensor (i.e., the square root of the sum of squares of its components). However, existing technologies face the following significant shortcomings when pursuing high-quality imaging: Blurred outline: Traditional modulus algorithms are essentially energy integration. For micro-defects (such as cracks with a width of less than 0.5 mm), the high-frequency geometric distortion signals generated are easily smoothed out by the integration process, resulting in blurred image outlines.

[0003] Frequency domain artifacts: Current techniques often use two-dimensional Fourier transform (FFT) combined with frequency domain field extension operators to calculate gradients and higher-order derivatives. FFT requires the data to be periodic, which can introduce the "Gibbs effect" or ringing artifacts at the image edges, severely affecting image quality.

[0004] Discontinuous signal: Affected by the slight fluctuations in workpiece surface roughness or lift-off value, defect signals often appear as discontinuous spots in imaging, lacking an effective local connection mechanism and making it difficult to form a continuous, closed, and true topological profile. Summary of the Invention

[0005] The purpose of this invention is to provide a weak magnetic field contour imaging method and system based on tensor eigenvalue extraction. In existing weak magnetic field detection, traditional modulus algorithms are unable to clearly restore the contours of fine defects, and frequency domain processing is prone to edge artifacts and signal discontinuity. This invention solves the technical problems of contour blurring, edge artifacts and signal discontinuity by using a pure spatial domain algorithm path and principal curvature feature extraction, thereby achieving high-resolution imaging that is "what you see is what you get".

[0006] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a weak magnetic contour imaging method based on tensor eigenvalue extraction, comprising the following steps: S1: Obtain the original magnetic field data of the magnetic sensor array arranged on the surface of the test piece, perform hard magnetic and soft magnetic calibration on the original magnetic field data, eliminate the zero-point drift and triaxial sensitivity inconsistency interference of the sensor itself, and construct the magnetic field space vector grid using the cubic spline interpolation algorithm. S2: Calculate the horizontal gradient component using each node of the magnetic field space vector grid, and calculate the vertical gradient component based on the divergence-free physical constraint of the magnetic field, and assemble them into a real symmetric magnetic gradient tensor matrix. S3: Perform eigenvalue decomposition on the magnetic gradient tensor matrix, solve for three eigenvalues, and select the eigenvalue with the largest absolute value as the principal curvature characteristic of the magnetic field that characterizes the spatial geometric distortion of the defect. S4: Calculate the local window statistical characteristics of the principal curvature features of the magnetic field, and dynamically adjust the gain of the enhanced principal curvature features of the magnetic field to generate a clear defect contour image.

[0007] In some embodiments, the vertical gradient component is calculated based on the horizontal gradient component obtained by the finite difference method and the divergence-free physical constraint of the magnetic field. The calculation is performed entirely within the spatial domain. ; In the formula, They are respectively axis; , , The magnetic induction intensity vectors are respectively in Component of direction; These are the components of the gradient tensor matrix.

[0008] In some embodiments, the horizontal gradient component is: ; In the formula, , All of these are the step sizes of the interpolation grid.

[0009] In some embodiments, the divergence-free physical constraint condition of the magnetic field is: the divergence of the magnetic induction intensity of the static magnetic field is always zero in the source-free region. ; In the formula, The gradient operator represents the partial derivative operation with respect to spatial coordinates; is the magnetic induction intensity vector.

[0010] In some embodiments, the principal curvature characteristic of the magnetic field is: ; This eigenvalue characterizes the maximum principal curvature of the magnetic field equipotential surface and is extremely sensitive to spatial geometric distortions caused by minute defects.

[0011] In the formula, , , These are the three eigenvalues ​​of the magnetic gradient tensor matrix at this node, used to extract the direction component with the largest curvature in the spatial distribution of the magnetic field.

[0012] In some embodiments, the principal curvature features of the magnetic field are enhanced using an adaptive local contrast normalization algorithm, and the calculation expression is as follows: ; In the formula, This refers to the original principal curvature characteristic quantity; For The mean value within a local window centered on the target; For The standard deviation within the local window centered on the target; The adaptive regularization parameter is proportional to the mean standard deviation of all image features and is used to suppress noise amplification in the background region.

[0013] In some embodiments, the principal curvature features of the magnetic field enhanced by the adaptive local contrast normalization algorithm are... A non-linear mapping is performed, using the Sigmoid activation function for "soft binarization" to softly binarize the defect edges, thereby improving contour clarity. ; In the formula, This is the sharpening gain coefficient; The enhanced principal curvature characteristic of the magnetic field; The center of the intensity threshold.

[0014] Secondly, the present invention provides a weak magnetic contour imaging system based on tensor eigenvalue extraction, utilizing the method described in the first aspect, comprising the following modules: Data acquisition module: acquires data from the surface of the workpiece by means of a magnetic sensor array arranged on the surface of the workpiece, and constructs a spatial vector grid of the magnetic field; Tensor Construction Module: Calculates the horizontal gradient component using each node of the magnetic field space vector grid in the data acquisition module, and calculates the vertical gradient component based on the divergence-free physical constraint of the magnetic field, assembling them into a real symmetric magnetic gradient tensor matrix. Feature extraction module: Perform eigenvalue decomposition on the magnetic gradient tensor matrix in the tensor construction module, solve for three eigenvalues, and select the eigenvalue with the largest absolute value as the principal curvature feature of the magnetic field that characterizes the geometric distortion of the defect space; Image enhancement module: Calculates the local window statistical characteristics of the principal curvature features of the magnetic field in the feature extraction module, and dynamically adjusts the gain of the enhanced principal curvature features of the magnetic field to generate a clear defect contour image before outputting it.

[0015] Thirdly, the present invention provides an electronic device including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the method described in the first aspect.

[0016] Fourthly, the present invention provides a computer-readable medium having a processor-executable program code that causes the processor to perform the method described in the first aspect.

[0017] Compared with the prior art, the present invention has the following beneficial effects: Providing high-resolution contour imaging: The analysis method based on the principal curvature characteristics of the magnetic field in this invention can capture signals with weak energy but extremely large spatial curvature, achieving "optical-grade" clear imaging of minute defects.

[0018] Edge-free artifacts: This invention uses spatial domain difference and algebraic operations throughout the process, avoiding the Gibbs effect caused by FFT transformation and eliminating blind spots in image edge detection.

[0019] Good visual connectivity: The present invention uses an adaptive LCN algorithm to effectively repair signal breaks caused by environmental interference and suppress background noise, so that the final imaging contour is continuous and closed, and the defect topology is realistically restored. Attached Figure Description

[0020] Figure 1 This is a schematic diagram of the overall process of Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the spatial distribution of the sensor array sampling points and the high-density interpolation grid in Embodiment 1 of the present invention; Figure 3 This is a schematic diagram showing the signal enhancement effect of the adaptive local contrast normalization (LCN) algorithm in Embodiment 1 of the present invention. Figure 4 The images show a comparison of the final imaging effects of the present invention, where (a) is traditional gradient modulus imaging and (b) is eigenvalue sharpening imaging of the present invention. Figure 5 This is a schematic diagram of the prefabricated crack test block in Embodiment 1 of the present invention and an imaging effect diagram of cracks of different sizes; Figure 6 This is a system structure diagram of Embodiment 2 of the present invention. Figure 7 This is a structural diagram of the electronic device according to Embodiment 2 of the present invention. Detailed Implementation

[0021] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention without creative effort are within the scope of protection of the present invention.

[0022] Example 1: Please see Figures 1-5 A weak magnetic contour imaging method based on tensor eigenvalue extraction and adaptive local enhancement, in a specific embodiment, includes the following steps: Step 1: Data Acquisition and Spatial Domain Interpolation Reconstruction; The raw magnetic field data of the test piece is acquired using a magnetic sensor array arranged on its surface. However, due to manufacturing errors and interference from surrounding ferromagnetic materials, the magnetic sensor array can introduce hard magnetic errors (zero-point offset) and soft magnetic errors (inconsistent sensitivity and non-orthogonality errors) during actual measurements. Therefore, the raw magnetic field data needs to be calibrated before data processing: the coefficient matrix is ​​calibrated using methods such as ellipsoidal fitting or least squares calculation, and the raw data is linearly transformed to correct for distortion, restoring the true magnetic field vector signal. Subsequently, a high-density magnetic field spatial vector grid with second-order derivative continuity is constructed using a cubic spline interpolation algorithm.

[0023] like Figure 2 As shown, to achieve high-precision gradient calculation in subsequent steps, the limited sampling density of the physical sensor must be addressed. In the figure, solid dots represent the physical sampling positions of the original sensor, and their spatial resolution is limited by hardware size. This embodiment utilizes the cubic spline interpolation algorithm to construct an extremely fine virtual mesh between the physical sampling points (shown as dashed meshes or small crosshairs in the figure, with a step size of up to 1 mm). A magnified view shows that the interpolation points achieve uniform and smooth filling between the physical points. The key to choosing cubic spline interpolation is that it ensures the second derivative continuity of the reconstructed magnetic field surface in space (…). (continuous), thus avoiding numerical oscillations in subsequent data processing.

[0024] Step 2: Constructing the Hessian matrix in the pure spatial domain; At each node of the magnetic field space vector grid, the horizontal gradient component is calculated using spatial finite difference, and the vertical gradient component is calculated based on the physical constraint of no divergence in the magnetic field, thereby assembling a real symmetric magnetic gradient tensor matrix.

[0025] This embodiment completely abandons the complex frequency domain transformations or exponential decay operators found in existing technologies, ensuring the purity of the physical logic. The magnetic gradient tensor, i.e., the Hessian matrix of the magnetic scalar potential, is directly constructed on the reconstructed high-density space vector grid. .

[0026] The construction process is as follows: First, the four gradient components in the horizontal plane are calculated using the Spatial Finite Difference Method. For any point on the grid... The expression for calculating its horizontal gradient component is: ; In the formula, , All of these are the step size of the interpolation grid; in this embodiment, it is set to 1 mm.

[0027] Then, for the vertical gradient components that cannot be directly measured by a single-layer array in physical terms... This embodiment is based on the divergence-free physical constraint of the static magnetic field in the passive region. According to Maxwell's equations, the divergence of the magnetic induction intensity is always zero. ; In the formula, The gradient operator represents the partial derivative operation with respect to spatial coordinates; is the magnetic induction intensity vector.

[0028] This directly leads to the vertical gradient component: In the formula, They are respectively axis; , , The magnetic induction intensity vectors are respectively in Component of direction; These are the components of the gradient tensor matrix.

[0029] Finally, based on the properties of magnetic scalar potential, using and Alternatively, using divergence-free physical constraints Perform verification and assemble into a complete system. Real symmetric Hessian matrix : ; This step is completed entirely in the spatial domain through algebraic operations, effectively avoiding the Gibbs effect and edge ringing artifacts introduced by Fourier transform in the frequency domain method.

[0030] Step 3: Principal curvature feature extraction; The magnetic gradient tensor matrix is ​​decomposed into eigenvalues ​​to obtain three eigenvalues. The eigenvalue with the largest absolute value is selected as the principal curvature characteristic of the magnetic field that characterizes the spatial geometric distortion of the defect.

[0031] For each node on the magnetic field space vector grid, after the construction is complete After obtaining the magnetic gradient tensor matrix, this embodiment does not use the traditional modulus calculation (i.e., the sum of squares of all components), but directly solves for the eigenvalues ​​of the matrix mathematically. Since the magnetic gradient tensor matrix is ​​a real symmetric matrix, according to the spectral theorem, it must have three real eigenvalues. , , These real eigenvalues ​​physically correspond precisely to the principal curvatures of the magnetic field equipotential surfaces in the three orthogonal principal directions. For microcracks, although the leakage magnetic field has weak energy, it causes the magnetic field lines to bend sharply in the direction perpendicular to the crack direction.

[0032] In this embodiment, the core imaging feature quantity is defined as: ; In the formula, , , Let be the three eigenvalues ​​of the magnetic gradient tensor matrix at this node.

[0033] The core imaging feature essentially constructs a "maximum curvature filter". Regardless of the orientation of the crack in the plane, this operator can always extract the component with the greatest curvature, thereby keenly extracting the geometric skeleton of the crack from the gentle background magnetic field.

[0034] Step 4: Image enhancement and non-linear sharpening; The local window statistical characteristics of the principal curvature features of the magnetic field are calculated, and the principal curvature features of the magnetic field are enhanced using an adaptive local contrast normalization algorithm. The enhanced features are then dynamically adjusted to generate a clear defect contour image.

[0035] like Figure 3 As shown, to address the signal discontinuity problem, the principal curvature features of the magnetic field are first enhanced using an adaptive local contrast normalization (LCN) algorithm. ; In the formula, This refers to the original principal curvature characteristic quantity; For The mean value within a local window centered on the target; For The standard deviation within the local window centered on the target; For adaptive regularization parameters.

[0036] Subsequently, to further enhance the edge sharpness of the contour, this embodiment enhances the magnetic field principal curvature feature quantity after the LCN algorithm. Perform nonlinear mapping. Unlike simple linear stretching, this embodiment uses the Sigmoid activation function for "soft binarization": ; In the formula, To sharpen the gain coefficient (e.g., take...) ); The enhanced principal curvature characteristic of the magnetic field; Center of intensity threshold (e.g., take) ).

[0037] The description refers to: the enhanced principal curvature characteristic of the magnetic field. medium to low The background noise region is quickly compressed to 0 (black), which will be higher than The crack signal area is quickly pulled up to 1 (white), thereby eliminating the grayscale transition band and outputting a high-resolution contour image with clear black and white contrast and sharp edges.

[0038] To visually verify the beneficial effects of this embodiment, such as Figure 4 As shown, (a) represents the modulus of the conventional magnetic gradient tensor ( (a) Imaging. It can be seen that due to the energy integration effect, the crack appears as a blurry spot with a width of about 5 mm, and the signals at both ends are weak, making it impossible to identify the true width and length of the crack. (b) Imaging using principal curvature feature extraction and LCN sharpening in this embodiment. The results show that the blurry spot is focused into a clear line with a width of only about 1 mm. The crack outline is continuous and closed, and the background noise is completely suppressed, achieving "optical-level" restoration of the topological morphology of the microcrack.

[0039] To further verify the detection limit and imaging resolution of the present invention for microcracks, a sample was fabricated as follows. Figure 5 The upper part shows a pre-fabricated cracked specimen of ferromagnetic material. The specimen size is 500 mm. 100mm 15mm, with artificial cracks of varying widths (0.15mm to 0.5mm) and depths (0.15mm to 10mm) on the surface.

[0040] Figure 5 The lower half illustrates this embodiment for four typical crack sizes (0.5 mm). 10mm, 0.3mm 5mm, 0.2mm 2mm and 0.15mm The final imaging effect (2mm). As shown in the figure, even for extremely fine cracks with a width of only 0.15mm, the principal curvature features extracted by this invention, after enhancement, can still generate a defect contour image with a very high signal-to-noise ratio and sharp edges. This result strongly demonstrates that this invention can accurately restore the spatial geometry of micron-level defects while suppressing background noise, verifying the effectiveness and high-resolution characteristics of the method.

[0041] Example 2 like Figure 6 As shown, a weak magnetic field contour imaging system based on tensor eigenvalue extraction and adaptive local enhancement, utilizing the aforementioned weak magnetic field contour imaging method, includes the following modules: Data acquisition module: The magnetic sensor array arranged on the surface of the test piece acquires the data of the test piece surface, and uses a cubic spline interpolation algorithm to construct a high-density magnetic field space vector grid with second derivative continuity. Tensor Construction Module: Calculates the horizontal gradient component using each node of the magnetic field space vector grid in the data acquisition module, and calculates the vertical gradient component based on the divergence-free physical constraint of the magnetic field, assembling them into a real symmetric magnetic gradient tensor matrix. Feature extraction module: Perform eigenvalue decomposition on the magnetic gradient tensor matrix in the tensor construction module, solve for three eigenvalues, and select the eigenvalue with the largest absolute value as the principal curvature feature of the magnetic field that characterizes the geometric distortion of the defect space; Image enhancement module: Calculates the local window statistical characteristics of the principal curvature features of the magnetic field in the feature extraction module, and dynamically adjusts the gain of the enhanced principal curvature features of the magnetic field to generate a clear defect contour image before outputting it.

[0042] The weak magnetic contour imaging system based on tensor eigenvalue extraction and adaptive local enhancement of the present invention can be installed in computer equipment.

[0043] The module described in this invention refers to a series of computer program segments that can be executed by the processor of a computer device and can perform a fixed function, and which are stored in the memory of the computer device.

[0044] The system provided in this embodiment of the invention has the same implementation principle and technical effects as the aforementioned method embodiment, and the corresponding content in the aforementioned method embodiment can be referred to.

[0045] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions marked in the blocks may occur in a different order than those shown in the drawings. For example, two consecutive blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or action, or using a combination of dedicated hardware and computer instructions.

[0046] Figure 7 A block diagram is shown that is suitable for implementing embodiments of the present application. Figure 7 The electronic device shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of this application.

[0047] like Figure 7 As shown, the electronic device is represented in the form of a general-purpose computing device. The components of the electronic device may include, but are not limited to: one or more processors 410, memory 430, and communication bus 440 connecting different system components (including memory 430 and processing unit 410).

[0048] Communication bus 440 represents one or more of several bus architectures, including a memory bus or memory controller, a peripheral bus, a graphics acceleration port, a processor, or a local bus using any of the various bus architectures. For example, these architectures include, but are not limited to, Industry Standard Architecture (ISA) buses, Micro Channel Architecture (MAC) buses, Enhanced ISA buses, Video Electronics Standards Association (VESA) local buses, and Peripheral Component Interconnect (PCI) buses.

[0049] Electronic devices typically include a variety of computer-readable media. These media can be any available media that can be accessed by the electronic device, including volatile and non-volatile media, and removable and non-removable media.

[0050] Memory 430 may include computer system readable media in the form of volatile memory, such as random access memory (RAM) and / or cache memory. The electronic device may further include other removable / non-removable, volatile / non-volatile computer system storage media. Memory 430 may include at least one program product having a set (e.g., at least one) of program modules configured to perform the functions of the embodiments of this application.

[0051] A program / utility having a set (at least one) of program modules can be stored in memory 430. Such program modules include—but are not limited to—an operating system, one or more application programs, other program modules, and program data. Each or some combination of these examples may include an implementation of a network environment. The program modules typically perform the functions and / or methods described in the embodiments of this application.

[0052] Processor 410 executes various functional applications and data processing by running programs stored in memory 430, such as implementing embodiments of this application. Figure 1 The RIES reliability assessment method provided in the illustrated embodiment.

[0053] This application provides a non-transitory computer-readable storage medium that stores computer instructions, which cause the computer to execute embodiments of this application. Figure 1 The weak magnetic contour imaging method provided in the illustrated embodiment.

[0054] The aforementioned computer-readable storage medium may be any combination of one or more computer-readable media. A computer-readable medium may be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium may be, for example—but not limited to—an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), or flash memory, optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium may be any tangible medium containing or storing a program that may be used by or in connection with an instruction execution system, apparatus, or device.

[0055] Computer-readable signal media may include data signals propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals may take various forms, including—but not limited to—electromagnetic signals, optical signals, or any suitable combination thereof. Computer-readable signal media may also be any computer-readable medium other than computer-readable storage media, capable of transmitting, propagating, or transmitting programs for use by or in connection with an instruction execution system, apparatus, or device.

[0056] The program code contained on a computer-readable medium may be transmitted using any suitable medium, including—but not limited to—wireless, wire, optical fiber, RF, etc., or any suitable combination thereof.

[0057] Computer program code for performing the operations of the embodiments of this application can be written in one or more programming languages ​​or a combination thereof. These programming languages ​​include object-oriented programming languages—such as Java, Smalltalk, and C++—and conventional procedural programming languages—such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN), or it can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0058] The foregoing has described specific embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps described in the claims may be performed in a different order than that shown in the embodiments and may still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require the specific or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0059] The above examples illustrate the present invention only to aid in understanding it and are not intended to limit the scope of the invention. Those skilled in the art can make various simple deductions, modifications, or substitutions based on the principles of this invention.

Claims

1. A weak magnetic field contour imaging method based on tensor eigenvalue extraction, characterized in that, Includes the following steps: S1: Acquire the raw magnetic field data of the magnetic sensor array arranged on the surface of the test piece, and construct a magnetic field space vector grid; S2: Calculate the horizontal gradient component using each node of the magnetic field space vector grid, and calculate the vertical gradient component based on the divergence-free physical constraint of the magnetic field, and assemble them into a real symmetric magnetic gradient tensor matrix. S3: Perform eigenvalue decomposition on the magnetic gradient tensor matrix, solve for three eigenvalues, and select the eigenvalue with the largest absolute value as the principal curvature characteristic of the magnetic field. S4: Calculate the local window statistical characteristics of the principal curvature features of the magnetic field, and dynamically adjust the gain of the enhanced principal curvature features of the magnetic field to generate a clear defect contour image.

2. The weak magnetic field contour imaging method based on tensor eigenvalue extraction according to claim 1, characterized in that, In the spatial domain, the vertical gradient component is calculated based on the horizontal gradient component obtained by the finite difference method and the divergence-free physical constraint of the magnetic field: ； In the formula, They are respectively axis; , , The magnetic induction intensity vectors are respectively in Component of direction; These are the components of the gradient tensor matrix.

3. The weak magnetic field contour imaging method based on tensor eigenvalue extraction according to claim 2, characterized in that, The horizontal gradient component is: ； In the formula, , All of these are the step sizes of the interpolation grid.

4. The weak magnetic field contour imaging method based on tensor eigenvalue extraction according to claim 2, characterized in that, The physical constraint condition for the divergence-free magnetic field is: within a source-free region, the divergence of the magnetic induction intensity of the static magnetic field is always zero. ； In the formula, The gradient operator represents the partial derivative operation with respect to spatial coordinates; is the magnetic induction intensity vector.

5. The weak magnetic field contour imaging method based on tensor eigenvalue extraction according to claim 1, characterized in that, The principal curvature characteristic of the magnetic field is: ； In the formula, , , Let be the three eigenvalues ​​of the magnetic gradient tensor matrix at this node.

6. The weak magnetic field contour imaging method based on tensor eigenvalue extraction according to claim 1, characterized in that, The principal curvature features of the magnetic field are enhanced using an adaptive local contrast normalization algorithm, and the calculation expression is as follows: ； In the formula, This refers to the original principal curvature characteristic quantity; For The mean value within a local window centered on the target; For The standard deviation within the local window centered on the target; For adaptive regularization parameters.

7. The weak magnetic field contour imaging method based on tensor eigenvalue extraction according to claim 6, characterized in that, The principal curvature features of the magnetic field enhanced by the adaptive local contrast normalization algorithm Perform nonlinear mapping, using the Sigmoid activation function for "soft binarization": ； In the formula, This is the sharpening gain coefficient; The enhanced principal curvature characteristic of the magnetic field; The center of the intensity threshold.

8. A weak magnetic field contour imaging system based on tensor eigenvalue extraction, comprising the method described in any one of claims 1-7, characterized in that, Includes the following modules: Data acquisition module: acquires data from the surface of the workpiece by means of a magnetic sensor array arranged on the surface of the workpiece, and constructs a spatial vector grid of the magnetic field; Tensor Construction Module: Calculates the horizontal gradient component using each node of the magnetic field space vector grid in the data acquisition module, and calculates the vertical gradient component based on the divergence-free physical constraint of the magnetic field, assembling them into a real symmetric magnetic gradient tensor matrix. Feature extraction module: Perform eigenvalue decomposition on the magnetic gradient tensor matrix in the tensor construction module, solve for three eigenvalues, and select the eigenvalue with the largest absolute value as the principal curvature feature of the magnetic field that characterizes the geometric distortion of the defect space; Image enhancement module: Calculates the local window statistical characteristics of the principal curvature features of the magnetic field in the feature extraction module, and dynamically adjusts the gain of the enhanced principal curvature features of the magnetic field to generate a clear defect contour image before outputting it.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that, When the processor executes the computer program, it implements the method according to any one of claims 1 to 7.

10. A computer-readable medium having processor-executable capability, characterized in that, The program code causes the processor to execute the method according to any one of claims 1 to 7.

Images