Multi-directional excitation magneto-optical image registration method under hybrid drive

Abstract

The application discloses a multi-directional excitation magneto-optical image registration method under hybrid driving, and the method comprises the following steps: collecting the defect magnetic field distribution information of the surface of a ferromagnetic material under multi-directional excitation through a magneto-optical imaging device to obtain a reference image and a floating image; extracting the defect contour of the two images according to a gradient operator; performing spatial transformation on the defect contour of the floating image, and then performing or operation and morphological closing operation on the defect contour of the floating image and the defect contour of the reference image to obtain a defect contour map with closed shape; obtaining the magnetic leakage field distribution map of the defect contour map through defect magnetic leakage field model forward modeling; calculating the similarity measure of the reference image and the magnetic leakage field distribution map generated by the model; and finally continuously updating the spatial transformation parameters through an optimization search algorithm, and the registration is successful when the similarity measure reaches the maximum.

Description

Technical Field

[0001] This invention belongs to the field of image registration technology, and more specifically, relates to a multi-directional excitation magneto-optical image registration method based on image data and model hybrid driving. Background Technology

[0002] Magneto-optical imaging technology is a non-destructive testing technique for defects in ferromagnetic materials based on magnetic flux leakage detection that has been developed in the last 30 years. Compared with traditional non-destructive testing methods, this method has the advantages of fast detection speed, high reliability, result visualization and easy automation. It can be used to visualize the detection of defects on ferromagnetic specimens and is a research hotspot in the field of magnetic flux leakage detection.

[0003] Due to the inherent characteristics of magnetic flux leakage imaging, under unidirectional magnetic field excitation, only the defect contour perpendicular to the magnetic field direction is sensitive to the excitation. Complete and effective defect information cannot be obtained through unidirectional magnetic field excitation alone. Therefore, it is necessary to fuse data from multidirectional magnetic field excitation to obtain more accurate defect information. However, in practical engineering applications, due to insufficient workpiece flatness and structural limitations, magneto-optical images are mainly obtained by subjecting the defect to multiple DC excitations in different directions. Under these circumstances, the acquired images exhibit spatial positional offsets. To fuse defect information from multidirectional excitation, the first problem to solve is the geometric registration of the magnetic flux leakage images.

[0004] Traditional image registration methods are mainly divided into two types: feature-based image registration methods and gray-level-based image registration methods. Feature-based registration methods require a clear mapping relationship of feature points between the two images. However, during the imaging process of magneto-optical images, due to magnetic field distortion caused by magnetic leakage, the boundaries of defects in the generated images are relatively blurred, leading to problems such as inaccurate feature point extraction and inaccurate feature point matching, which affects the accuracy of image registration. Gray-level-based image registration methods use the gray-level statistical information of the images themselves to measure the similarity between them. However, during magneto-optical imaging, due to the influence of magnetic leakage coupling, magneto-optical images obtained under magnetic field excitation in different directions have certain shape distortions. The gray-level values ​​of corresponding regions in different magneto-optical images of the same defect differ significantly, which can cause the algorithm to fail. Therefore, this method is also not suitable for the registration of magnetic leakage images under multi-directional excitation. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of the prior art and provide a multi-directional excitation magneto-optical image registration method under hybrid driving, so as to achieve high-precision and high-robust registration of defect magneto-optical images under multi-directional magnetic field excitation.

[0006] To achieve the above-mentioned objectives, the present invention provides a multi-directional excitation magneto-optical image registration method under hybrid driving, characterized by comprising the following steps:

[0007] (1) Under DC excitation in different directions, the magneto-optical imaging device is used to collect the defect information on the test piece, and the magneto-optical image in one direction at the same defect location is used as reference image A, and the magneto-optical image in another direction is used as floating image B.

[0008] (2) The gradient operator was used to process the reference image A and the floating image B to extract the defect contour reference image A. s and defect contour floating image B s ;

[0009] (3) Floating image B of the defect contour s Perform spatial transformation;

[0010] (3.1) Set the rigid transformation matrix Tran;

[0011]

[0012] Where Δx is the pixel translation on the x-axis, Δy is the pixel translation on the y-axis, and Δθ is the rotation angle;

[0013] (3.2) Based on the rigid transformation matrix, the floating image B of the defect contour is transformed. s Perform translation and rotation transformations with the transformation parameters ΔR = [Δx, Δy, Δθ] to obtain the transformed floating image B. tran :

[0014] B tran =B s ·Tran

[0015] (4) Use the defect contour reference image A s With the transformed floating image B tran First, perform a pixel OR operation, then perform a morphological closing operation to obtain a closed defect contour image C;

[0016] (5) The leakage magnetic field distribution of the defect contour image C is simulated and calculated using the leakage magnetic field forward model, so as to obtain the defect leakage magnetic field distribution map D;

[0017] (6) The similarity measure function is used to calculate the similarity between the defect leakage magnetic flux distribution map D and the reference image A to obtain the similarity measure;

[0018] (7) Use the particle swarm optimization algorithm to find the transformation parameters that make the similarity measure optimal. If the iteration terminates, proceed to step (8); otherwise, change the transformation parameters and return to step (3.2) to continue searching for the transformation parameters that make the similarity measure optimal.

[0019] (8) Output the optimal transformation parameters and apply the rigid transformation matrix to the floating image B of the defect contour.s Perform translation and rotation transformations to finally output the registered floating image B. after .

[0020] The objective of this invention is achieved as follows:

[0021] This invention discloses a multi-directional excitation magneto-optical image registration method under hybrid driving. It acquires defect magnetic field distribution information on the surface of a ferromagnetic material under multi-directional excitation using a magneto-optical imaging device, obtaining a reference image and a floating image. Defect contours are extracted from both images using a gradient operator. The floating image's defect contour undergoes a spatial transformation, and then an OR operation and morphological closure operation are performed with the reference image's defect contour to obtain a closed-shape defect contour map. The leakage magnetic field distribution map of the defect contour map is obtained through forward modeling using a defect leakage magnetic field model. The similarity measure between the reference image and the model-generated leakage magnetic field distribution map is then calculated. Finally, the spatial transformation parameters are continuously updated through an optimized search algorithm. Registration is successful when the similarity measure reaches its maximum.

[0022] Meanwhile, the multi-directional excitation magneto-optical image registration method under hybrid driving of the present invention also has the following beneficial effects:

[0023] During the registration process, combining magneto-optical image data with the leakage magnetic field distribution of the defect contour in the magneto-optical image can overcome the registration difficulties caused by the distortion of the defect magneto-optical image under multi-directional magnetic field excitation, and achieve high-precision and robust registration of the defect magneto-optical image. This helps to fuse effective information about the defect from multiple directions in the subsequent process, and achieve accurate identification of the defect. Attached Figure Description

[0024] Figure 1 This is a flowchart of a multi-directional excitation magneto-optical image registration method under hybrid driving according to the present invention;

[0025] Figure 2 These are magneto-optical images of the same defect location;

[0026] Figure 3 This is an example diagram of the interaction between a defective surface solenoid;

[0027] Figure 4 This is a schematic diagram showing the positions of the defect surface and the inspection surface. Detailed Implementation

[0028] The specific embodiments of the present invention will now be described with reference to the accompanying drawings to enable those skilled in the art to better understand the invention. It should be particularly noted that in the following description, detailed descriptions of known functions and designs that might obscure the main content of the invention will be omitted here.

[0029] Example

[0030] Figure 1This is a flowchart of a multi-directional excitation magneto-optical image registration method under hybrid driving according to the present invention.

[0031] In this embodiment, as Figure 1 As shown, the present invention provides a multi-directional excitation magneto-optical image registration method under hybrid driving, comprising the following steps:

[0032] S1. Under DC excitation in different directions, a magneto-optical imaging device is used to collect defect information on the test piece. A magneto-optical image in one direction at the same defect location is obtained as reference image A, and a magneto-optical image in another direction is obtained as floating image B. For example... Figure 2 As shown, (a) is the reference image A and (b) is the floating image B.

[0033] S2. The gradient operator is used to process the reference image A and the floating image B to extract the defect contour reference image A. s and defect contour floating image B s ;

[0034] In this embodiment, the gradient operator selected is the Sobel operator, which is a first-order discrete difference operator that specifically includes convolution templates in both vertical and horizontal directions.

[0035] Below, we will use the defect contour reference image A. s Taking the Sobel operator as an example, we will use it to extract the defect contour reference image A. s The process is as follows:

[0036]

[0037]

[0038]

[0039] Among them, G x G is the convolution template in the vertical direction. y This is the convolution template in the horizontal direction;

[0040] Similarly, we can also extract the floating image B of the defect contour using the method described above. s .

[0041] S3, For the floating image B of the defect contour s Perform spatial transformation;

[0042] S3.1, Set the rigid transformation matrix Tran;

[0043]

[0044] Where Δx is the pixel translation on the x-axis, Δy is the pixel translation on the y-axis, and Δθ is the rotation angle;

[0045] S3.2, Apply the rigid transformation matrix to the floating image B of the defect contour. s Perform translation and rotation transformations with the transformation parameters ΔR = [Δx, Δy, Δθ] to obtain the transformed floating image B. tran :

[0046] B tran =B s ·Tran

[0047] S4. Transfer the defect contour reference image A s With the transformed floating image B tran First, perform a pixel OR operation, then perform a morphological closing operation to obtain a closed defect contour image C;

[0048] S5. Use the forward modeling of the leakage magnetic field to simulate and calculate the leakage magnetic field distribution of the defect contour image C, thereby obtaining the defect leakage magnetic field distribution map D;

[0049] In this embodiment, the forward modeling of the leakage magnetic field includes, but is not limited to, the magnetic dipole model, the finite element model, and the solenoid model. We will use the solenoid model as an example below:

[0050] S5.1. A semi-infinite solenoid is constructed inside the medium to simulate the leakage magnetic field source of a defect surface. The magnetic field strength dH generated by any infinitesimal element of the defect surface at the leakage magnetic field point P is... ls for:

[0051]

[0052] Among them, M de M represents the magnetization intensity at the defect surface. d The effective component in the normal direction of the defect surface, i.e. Here, ds is the normal vector of the defect surface, and r is the infinitesimal area element of the defect surface. s It is the distance from the center point of the solenoid end face to the field point P;

[0053] In this embodiment, as Figure 4 As shown, the coordinates of field point P are (x, y, z), and the coordinates of the center point of the solenoid on the defect surface are (x0, y0, z0).

[0054] S5.2 Considering the non-uniform magnetization of complex defects in reality, to make the model closer to reality, the magnetization of the solenoid is assumed to be quasi-saturated, and the interaction force of the solenoid is introduced; to eliminate the singularity of the solenoid's neighborhood, the arctangent function is introduced into the formula to obtain the magnetic field strength H of the solenoid's interaction field. inner The calculation formula is as follows:

[0055]

[0056] Where, r l It is the distance from the center point of one solenoid end face to the center point of the other solenoid end face;

[0057] To simplify the model, 3×3 solenoids are uniformly distributed on the defect surface to simulate the source of the force. It is assumed that the forces between the solenoids do not affect the amplitude of the magnetization, but only change the direction of the solenoids, thus altering the effective magnetic moment of the solenoids on the defect surface. In this embodiment, the interaction between solenoid 1 and other solenoids is as follows: Figure 3 As shown, the solenoid No. 1 in the upper left corner is affected by the interaction forces of all other solenoids. The formula for calculating its magnetic field strength is as follows:

[0058] H1 = H 21 +H 31 +H 41 +…+H 91

[0059] S5.3 Calculate the direction of magnetization at the defect surface under the influence of the excitation magnetic field and the aforementioned 3×3 source, and use this as the new... The direction; combined with the normal vector of the defect surface. Calculate the corrected effective magnetic moment M de Substituting the formula from step S5.1 into the formula, we can obtain the magnetic field strength at the leakage magnetic field point considering the solenoid interaction force, and obtain the defect leakage magnetic field distribution map D of the model forward modeling.

[0060] S6. The similarity measure function is used to calculate the similarity between the defect leakage magnetic field distribution map D and the reference image A to obtain the similarity measure;

[0061] Mutual information, a common measure of image similarity, is based on entropy. To reduce the sensitivity of mutual information to overlapping parts of images, this embodiment uses normalized mutual information as the similarity measure. The specific calculation formula is as follows:

[0062]

[0063]

[0064]

[0065]

[0066] Where H(A) represents the entropy of the reference image A, a represents the different gray values ​​of each pixel in the reference image A, and P A (a) represents the probability that a pixel with gray value a appears in the reference image A; H(D) represents the entropy of the distribution map D, b represents the different gray values ​​of each pixel in the distribution map D, and represents the probability that a pixel with gray value b appears in the distribution map D; H(A,D) represents the joint entropy of the reference image A and the distribution map D, P A (a,b) represents the probability that a pixel at the same location has a gray value of a in the reference image A but a gray value of b in the distribution map B. NMI(A,D) represents the normalized mutual information between the reference image A and the distribution map D.

[0067] S7. Use the particle swarm optimization algorithm to find the transformation parameters that make the similarity measure optimal. If the iteration terminates, proceed to step S8; otherwise, change the transformation parameters and return to step S3.2 to continue searching for the transformation parameters that make the similarity measure optimal.

[0068] In this embodiment, the specific process of using the particle swarm optimization algorithm to find the transformation parameters that optimize the similarity measure is as follows:

[0069] S7.1 Initialize the particle swarm: Set the total number of particles in the particle swarm to 80, the maximum number of iterations to 100, the inertia weight factor w, the self-learning factor c1, the swarm learning factor c2, and randomly initialize the initial position and initial velocity of each particle.

[0070] S7.2 Define arrays pbest and gbest;

[0071] Calculate the normalized mutual information of each particle at its initial position as its fitness value. Store the current position and fitness value of each particle in each particle's pbest. Then, store the fitness value and position of the individual with the largest fitness value in pbest as the historical best fitness value and corresponding position of the group in gbest.

[0072] S7.3. Select the position corresponding to the historical best fitness value of each particle, and denote it as... And the position corresponding to the group's historical best fitness value is denoted as

[0073] S7.4 Determine whether the current iteration number t has reached the maximum iteration number. If it has, proceed to step S7.7; if not, proceed to step S7.5.

[0074] S7.5 Update the velocity and position of each particle;

[0075]

[0076] Among them, v i (t) represents the velocity of particle i at the t-th iteration, x i (t) represents the position of particle i in the t-th iteration, and r1 and r2 are random numbers between [0,1].

[0077] S7.6 Increment the current iteration number t by 1, then return to step S7.2;

[0078] S7.7 Output the historical best fitness value of the swarm particles and the corresponding particle position to obtain the optimal transformation parameters.

[0079] S8. Output the optimal transformation parameters and apply the rigid transformation matrix to the floating image B of the defect contour. s Perform translation and rotation transformations to finally output the registered floating image B. after .

[0080] Although the illustrative specific embodiments of the present invention have been described above to enable those skilled in the art to understand the invention, it should be understood that the invention is not limited to the scope of the specific embodiments. For those skilled in the art, various changes are obvious as long as they are within the spirit and scope of the invention as defined and determined by the appended claims, and all inventions utilizing the concept of the present invention are protected.

Claims

1. A multi-directional excitation magneto-optical image registration method under hybrid driving, characterized in that, Includes the following steps: (1) Under DC excitation in different directions, the magneto-optical imaging device is used to collect the defect information on the test piece, and the magneto-optical image in one direction at the same defect location is used as reference image A, and the magneto-optical image in another direction is used as floating image B. (2) The gradient operator was used to process the reference image A and the floating image B to extract the defect contour reference image A. s and defect contour floating image B s ; (3) For the floating image B of the defect contour s Perform spatial transformation; (3.1) Set the rigid transformation matrix ; ； in, For pixels translated on the x-axis, For pixels translated on the y-axis, The angle of rotation; (3.2) Based on the rigid transformation matrix, the floating image B of the defect contour is transformed. s For translation and rotation transformations, the transformation parameters are: Thus, the transformed floating image is obtained. : ； (4) Refer to the defect contour reference image A s With the transformed floating image First, perform a pixel OR operation, then perform a morphological closing operation to obtain a closed defect contour image C; (5) The leakage magnetic field distribution of the defect contour image C is simulated and calculated using the leakage magnetic field forward model, so as to obtain the defect leakage magnetic field distribution map D; (6) The similarity measure function is used to calculate the similarity between the defect leakage magnetic flux distribution map D and the reference image A to obtain the similarity measure; (7) Use the particle swarm optimization algorithm to find the transformation parameters that make the similarity measure optimal. If the iteration terminates, proceed to step (8); otherwise, change the transformation parameters and return to step (3.2) to continue searching for the transformation parameters that make the similarity measure optimal. (8) Output the optimal transformation parameters and apply them to the floating image B of the defect contour using a rigid transformation matrix. s Perform translation and rotation transformations to finally output the registered floating image. .

2. The multi-directional excitation magneto-optical image registration method under hybrid driving as described in claim 1, characterized in that, The forward modeling model of the leakage magnetic field includes the magnetic dipole model, the finite element model, and the solenoid model.

Images