A method for three-dimensional registration of neutron / X-ray heterogeneous source images of a cylindrical multilayer structure

Abstract

This invention provides a three-dimensional registration method for neutron / X-ray heterogeneous source images of cylindrical multilayer structures. The method includes edge extraction operators and PCA principal component analysis for vertical adjustment of image pairs; an edge contour extraction module to determine the height mapping of heterogeneous source tomographic images; a circle extraction algorithm and calculation of the circle center positioning translation; an image super-resolution reconstruction algorithm to achieve image scaling; and image rotation matching based on feature tomography and phase cross-correlation methods. This method effectively corrects geometric and modal differences between heterogeneous source images, achieving accurate alignment of metallic and non-metallic regions. It has significant advantages in registration accuracy, edge preservation, and modal feature alignment. It exhibits high adaptability and stability for complex structures and multimodal characteristic characterization, and is suitable for three-dimensional registration of neutron / X-ray heterogeneous source images of cylindrical multilayer structures such as cylindrical lithium batteries.

Description

Technical Field

[0001] This invention relates to the fields of image processing and non-destructive testing technology, specifically to a three-dimensional registration method for neutron / X-ray heterogeneous images of a cylindrical multilayer structure, which is particularly suitable for non-destructive testing and performance analysis of the internal structure of lithium batteries. Background Technology

[0002] Ensuring the effective storage and utilization of energy is a key issue for achieving sustainable development in today's society. Lithium-ion batteries, with their high energy density, good cycle stability, and lack of memory effect, have become the preferred energy solution for many industries, widely used in electric vehicles, aerospace equipment, and defense applications. With the expansion of their application scope, the performance requirements for lithium batteries are constantly increasing, especially in terms of safety. In actual operation, lithium batteries may encounter problems such as overcharging, overheating, and lithium dendrite growth, leading to corrosion of the current collector and gas generation. These problems not only shorten battery life but may also trigger serious accidents such as thermal runaway, fires, and even explosions. To reduce these potential risks, researchers have explored various improvement methods. First, by adopting a battery thermal management system, the battery temperature during operation can be effectively controlled to prevent overheating. In addition, flame-retardant electrolytes, redox shuttles, safer separator materials, and more stable cathode materials have been developed to reduce potential safety hazards. However, even with these measures, completely eliminating all risks remains challenging. Therefore, using non-destructive testing technology to monitor the quality of lithium batteries in real time to ensure their stability during use has become an important solution. Non-destructive testing technology can not only detect structural changes inside lithium batteries, but also provide valuable information for future material improvements and technology optimization.

[0003] Simultaneous use of neutron and X-ray imaging techniques to detect complex structural samples containing light elements (hydrogen, lithium, boron, etc.) and metals (iron, copper, etc.), such as multi-flow mixtures, lithium-ion batteries, and electronic packaging devices, can provide more comprehensive information on the internal structure and composition of the samples, facilitating accurate and multi-faceted interpretation of the sample's material, structure, and quality. Registering and fusing images from both modes helps in the analysis and evaluation of the internal quality state of lithium batteries, providing stronger support for lithium battery quality testing. However, since the two types of images are often acquired from different imaging devices, differences in imaging coordinate systems and parameters can lead to significant differences between X-ray tomography images and neutron tomography images in terms of image size, grayscale distribution, location, and scale. Therefore, a targeted image registration scheme needs to be designed to achieve high-precision alignment of heterogeneous images. Summary of the Invention

[0004] The purpose of this invention is to provide a three-dimensional registration method for neutron / X-ray heterogeneous source images of cylindrical multilayer structures, so as to solve the problem of insufficient registration accuracy in the prior art and realize high-precision non-destructive testing of the internal structure of lithium batteries.

[0005] This invention is achieved through the following technical solution. A method for three-dimensional registration of neutron / X-ray heterogeneous source images of a cylindrical multilayer structure is provided, comprising:

[0006] Edge extraction operators and PCA principal component analysis are used to adjust the verticality of image pairs;

[0007] The edge contour extraction module determines the height mapping of heterogeneous tomographic images;

[0008] A circle detection algorithm is used to detect the center of the circle and determine the amount of translation, so that the heterogeneous tomographic images are translated and aligned.

[0009] Based on the scale ratio, an image super-resolution reconstruction algorithm is used to achieve scale transformation super-resolution reconstruction of tomographic images;

[0010] Image rotation matching is achieved based on feature tomography and phase cross-correlation methods.

[0011] Furthermore, the verticality adjustment is based on the projected image; the image pair to be registered is a neutron projection image and an X-ray projection image; the verticality adjustment requires grayscale normalization of the input image pair, normalizing the grayscale values ​​to the range of [0,1]; for the normalized projected image, the OTU algorithm is used to perform threshold segmentation on the projected image to extract the edge contour of the projected image.

[0012] The PCA principal component analysis refers to: first, statistically analyzing the coordinates of the boundary pixels, and then using PCA to determine the first principal component, which is the direction of the axis of symmetry.

[0013] Furthermore, the key to threshold segmentation in the OTU algorithm is determining the segmentation threshold. By selecting a segmentation threshold T, the projected image is divided into foreground and background parts, such that the inter-class variance between the foreground and background is minimized. Maximum; for the acquired edge contour image, the vertical boundary of the foreground region is extracted again using the Sobel operator.

[0014] Furthermore, the purpose of the edge contour extraction module in determining the height mapping of heterogeneous tomographic images is to map the tomographic images of neutrons and X-rays to the same height; the height mapping determines the height corresponding to the first and last tomographic images of the cylindrical multilayer structure by determining the horizontal boundary of the projected image.

[0015] Furthermore, the height mapping method is as follows: an adaptive edge extraction operator is used to detect the horizontal boundary of the projected image, the pixels at the horizontal boundary are sorted, and the average of the row coordinates of the top 20% of the pixels is taken as the height of the horizontal boundary.

[0016] Furthermore, before edge detection, the input tomographic image is Gaussian smoothed using a Gaussian smoothing template, and then edge detection is performed. After edge detection, holes in the tomographic image are filled based on the region growing principle. After hole filling, the center of the circle is detected using a differentiable Lie group Hough circle detection algorithm. The coordinates of the center and the diameter of the circle can be calculated based on the region connectivity. The translation amount is determined according to the location of the center. After all the centers of the heterogeneous tomographic images are moved to the center, the translation and alignment of the heterogeneous tomographic images is completed.

[0017] Furthermore, the image super-resolution reconstruction algorithm includes using sine wave overlay for noise reduction, adaptive histogram equalization to improve image clarity, homomorphic filtering to sharpen details, bilinear interpolation algorithm for upsampling, and homomorphic filtering to enhance contrast, thereby achieving super-resolution reconstruction and scale registration of tomographic images.

[0018] Furthermore, using the tomographic image of the top shell region of the cylindrical multilayer structure as the registration object, the rotation angle was calculated and then all tomographic images were rotated and registered; the gradient of the tomographic image was first calculated and the central region of the tomographic image was selectively preserved, and the phase correlation method was used to match the heterogeneous gradient images.

[0019] The present invention also provides an electronic device, comprising: a processor, a communication interface, and a communication bus, wherein the processor, the communication interface, and the memory communicate with each other through the communication bus; the processor calls logical instructions in the memory to execute each step of the three-dimensional registration method for cylindrical multilayer neutron / X-ray heterogeneous source images.

[0020] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, when executed by a processor, the computer program implements the various steps of the described three-dimensional registration method for neutron / X-ray heterogeneous images of cylindrical multilayer structures.

[0021] Based on the complex internal structure and multi-material distribution of lithium batteries, this invention proposes a three-dimensional registration framework for heterogeneous imaging systems. The framework consists of four complementary stages, and three novel algorithms are introduced in key stages to improve overall robustness and accuracy: First, an adaptive edge extraction operator is used to accurately lock structural edges in a multi-scale fractional gradient domain, automatically estimating projection tilt and achieving pixel-level pose correction. Second, a bilinear interpolation algorithm performs super-resolution upsampling on low-resolution images. Third, this algorithm adaptively selects Padé poles in the conformal domain of the complex plane, combining Chebyshev nodes to minimize boundary ringing, effectively suppressing high-frequency artifacts and reconstructing details. Fourth, for the registration of tomographic images, a differentiable Lie group Hough circle detection algorithm is used to accurately estimate the center and radius of the workpiece fracture, embedding its gradient information into manifold optimization. This simultaneously solves for the Euclidean translation vector and rotation angle, achieving globally optimal registration of circular tomographic images. This four-stage process of "edge-geometry-scale-pose" couples detection, analysis, and optimization: the adaptive edge extraction operator ensures accurate pose, the bilinear interpolation algorithm provides high-fidelity details, and the differentiable Lie group Hough circle detection algorithm directly incorporates circle geometry into the pose solution in a differentiable manner. Ultimately, this not only eliminates size, grayscale, and spatial mismatches caused by hardware differences but also significantly improves the clarity and interpretable geometric consistency of the registered image, building a robust and precise alignment benchmark for subsequent cross-source data fusion.

[0022] This invention offers significant advantages in registration accuracy, edge preservation, and modal feature alignment, demonstrating high adaptability and stability for characterizing the complex structures and multimodal properties of cylindrical lithium-ion batteries. It can also be used for three-dimensional registration of neutron / X-ray heterogeneous source images of cylindrical multilayer structures, in addition to cylindrical lithium-ion batteries. Attached Figure Description

[0023] To more clearly illustrate the technical solutions in this invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this invention. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0024] Figure 1 This is a flowchart of the three-dimensional registration method for neutron / X-ray heterogeneous source images of a cylindrical multilayer structure provided by the present invention;

[0025] Figure 2 This is the overall flowchart of the present invention;

[0026] Figure 3 This is an example diagram showing the relative tilt of the projected image provided by the present invention;

[0027] Figure 4This is a grayscale distribution map of a neutron-X-ray projection image of a cylindrical lithium battery in an embodiment of the present invention;

[0028] Figure 5 This is an example diagram of vertical boundary extraction in an embodiment of the present invention;

[0029] Figure 6 This is a flowchart of the calculation and alignment of neutron-X-ray tomographic image translation in an embodiment of the present invention;

[0030] Figure 7 This is a flowchart of the super-resolution reconstruction process in an embodiment of the present invention;

[0031] Figure 8 This is the tomographic image registration result of lithium battery No. 1 in this embodiment of the invention;

[0032] Figure 9 This is a rendering effect of the superimposed tomographic image in an embodiment of the present invention;

[0033] Figure 10 This is a cross-sectional view of the three-dimensional registration entity of lithium battery No. 1 in this embodiment of the invention;

[0034] Figure 11 This is the tomographic image registration result of lithium battery No. 2 in this embodiment of the invention;

[0035] Figure 12 This is a cross-sectional view of lithium battery No. 2 in an embodiment of the present invention;

[0036] Figure 13 This is a tomographic image registration effect diagram of a miniature DC motor in an embodiment of the present invention;

[0037] Figure 14 This is a diagram showing the effect of three-dimensional image registration of a motor in an embodiment of the present invention;

[0038] Figure 15 This is a schematic diagram of the structure of the heterogeneous image three-dimensional registration device provided by the present invention;

[0039] Figure 16 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0040] To make the objectives, technical solutions, and advantages of this invention clearer, the technical solutions of this invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of this invention. All other embodiments obtained by those skilled in the art based on the embodiments of this invention without creative effort are within the scope of protection of this invention.

[0041] Figure 2This invention provides a method for three-dimensional registration of neutron / X-ray heterogeneous images of a cylindrical multilayer structure. The cylindrical multilayer structure in this embodiment uses a cylindrical lithium battery as the research object. While CT imaging of a cylindrical lithium battery using neutrons and X-rays can reveal different information, neutron CT and X-ray CT are two independent imaging systems. When imaging an object, differences in the object's pose, the geometry of the imaging system, and imaging parameters (such as X-ray energy, sensor size, and magnification ratio) can lead to differences in verticality, rotation angle, foreground position, tomographic height, and image resolution between the acquired projection image and the reconstructed tomographic image. To address these problems, this invention designs... Figure 2 The illustrated 3D registration process for heterogeneous images of cylindrical lithium batteries decouples image matching into five sub-problems: data loading, verticality adjustment, interlayer matching, translation and scale transformation, and rotation angle matching, which are solved step by step.

[0042] Reference Figure 1 and Figure 2 The embodiments of this invention provide a method for three-dimensional registration of neutron / X-ray heterogeneous source images of a cylindrical multilayer structure, comprising:

[0043] S1, edge extraction operator and PCA principal component analysis are used to adjust the verticality of image pairs.

[0044] This invention designs a method for adjusting the verticality of projected images. Image fusion requires that the tomographic images to be fused reflect information from the same location in the actual object, that is, that the tomographic images correspond to slices of the same plane as the cylindrical lithium battery. However, in actual imaging and detection, differences in system structure or external experimental factors may cause the projected images of the cylindrical lithium battery on the two imaging systems to be relatively tilted, such as... Figure 3 As shown, the slices of the two sets of tomographic images are not on the same plane. Under this condition, the reconstructed two sets of tomographic images cannot be mapped to the same plane, and the verticality of the acquired images needs to be adjusted. Since the information reflected in neutron and X-ray projection images differs significantly, similarity matching based on grayscale information or feature matching cannot achieve satisfactory registration results. Considering that the edges of the projection image of a cylindrical lithium battery are relatively clear, this invention uses an edge extraction algorithm to extract the vertical boundary of the cylindrical lithium battery, and then calculates the tilt angle between the vertical boundaries to achieve the verticality adjustment of the projection image.

[0045] like Figure 4As shown, since the grayscale distributions of X-ray projection images and neutron projection images are different, it is necessary to normalize the grayscale of the input image pair (neutron projection image and X-ray projection image) to normalize the grayscale values ​​to the range of [0,1]. For the normalized projection image, the OSTU algorithm (Otsu algorithm) is used to perform threshold segmentation on the projection image to extract the edge contour of the projection image. The key to threshold segmentation in the OSTU algorithm is to determine the segmentation threshold. The mathematical principle of the segmentation threshold calculation is shown in formula (1). By selecting a segmentation threshold T, the projection image is divided into two parts: foreground and background, so that the inter-class variance between the foreground and background is reduced. maximum.

[0046] (1);

[0047] In formula (1), when the segmentation threshold is T, and These represent the proportions of pixels in the foreground and background regions to the total number of pixels, respectively. and This represents the average gray level of the foreground and background regions.

[0048] (2);

[0049] Once the segmentation threshold is determined using formula (1), the projected image can be segmented according to formula (2) to obtain the edge contour image. .in This represents the grayscale value of the pixel at row i and column j. For the acquired edge contour image, the Sobel operator is used to extract the vertical boundary of the foreground region.

[0050] Because edge extraction algorithms have limited accuracy, directly calculating the tilt angle of vertical boundaries yields low-precision results. Figure 5 It can be seen that the vertical boundary image of the lithium battery is a binary image, and the image is symmetrical from left to right. If the axis of symmetry on the left and right sides is calculated, and then the tilt angle of the axis of symmetry is calculated, the verticality adjustment of the cylindrical lithium battery can be achieved. Based on this consideration, the present invention first counts the coordinates of the boundary pixels (i.e., pixels with a gray value of 1), and then uses PCA to determine the first principal component, which is the direction of the axis of symmetry. The principle of PCA is shown in formula (3). PCA is a statistical method that aims to project high-dimensional data into a new coordinate system through linear transformation, and in this new coordinate system, the variance of the data is arranged in descending order. The goal of this method is to find the direction with the largest variance in the dataset, which is the direction of the principal component. The first principal component is the direction with the largest variance in the dataset, so it can explain the largest change in the data.

[0051] (3);

[0052] In formula (3), It is image data The average gray level, It is the image data after removing the mean. yes transpose; It is the covariance matrix. It is the eigenvector of the covariance matrix, representing the direction of the i-th direction (principal component) in the dataset; It is the eigenvalue in the i-th direction, representing the variance of the data in the i-th direction; It is the eigenvector corresponding to the largest eigenvalue; Let be the position coordinate vector of the i-th pixel. for exist The projection value in the direction. After determining the axis of symmetry of the boundary image, the tilt angle of the axis of symmetry can be calculated. By rotating the projected image according to the tilt angle, the verticality of the projected image can be adjusted.

[0053] S2, the edge contour extraction module determines the height mapping of heterogeneous tomographic images.

[0054] This invention addresses the problem of tomographic location differences caused by varying resolutions in neutron / X-ray CT devices by designing a tomographic image height mapping method.

[0055] To align heterogeneous tomographic images, the height corresponding to the first and last tomographic images of the cylindrical lithium battery can be determined by identifying the horizontal boundary of the projected image. To achieve this, an adaptive edge extraction operator is used to detect the horizontal boundary of the projected image. Due to the limited accuracy of edge detection and the fact that the projected images of the upper and lower boundaries of the cylindrical lithium battery are not strictly straight lines, the row coordinates of the pixels on the edge are not the same. To improve the calculation accuracy, the pixels at the horizontal boundary are sorted, and the average of the row coordinates of the top 20% of the pixels is taken as the height of the horizontal boundary. The specific principle is shown in formula (4).

[0056] (4);

[0057] In formula (4), This represents the binarized image after edge extraction, where i is the row index of the pixel and j is the column index of the pixel. Represents the matrix after index association; flatten() flattens a two-dimensional image into a one-dimensional array; The sorted array, This represents sorting a one-dimensional array in descending order. This represents the mode extraction function. Indicates the height of the first fault. Indicates the height of the tail fault. This indicates filtering the top N data items. This indicates filtering the bottom N data items, where N equals 20% of the image width. and These represent the row indices of the first 20% and the last 20% of a sorted one-dimensional array, respectively.

[0058] After determining the positions of the first and last tomographic images of the neutron and X-ray sources according to formula (4), since the height of the cylindrical lithium battery is fixed and the verticality of the projected image has been adjusted, the height positions of all tomographic images can be accurately aligned and mapped using the positions of the first and last tomographic images as the standard.

[0059] S3. Use a circle detection algorithm to detect the center of the circle and determine the translation amount to align the heterogeneous tomographic images.

[0060] During CT imaging, changes in the distance of the object relative to the center of rotation of the turntable will cause changes in the object's position in the projected image. Both neutron and X-ray systems require 360° projection sampling around the object during CT imaging. Different initial irradiation angles during sampling will cause rotation between the reconstructed images. The principle behind this can be explained by formulas (5)-(7):

[0061] (5);

[0062] in, This is the tomographic image to be reconstructed, where x is the horizontal coordinate of the pixel and y is the vertical coordinate of the pixel. Indicates the angle Projected data along the direction of line t; This is the Dirac function, used to select the projection direction. As can be seen from formula (5), at the angle... Projection data at the location It can be viewed as the cumulative projection of the object at that angle.

[0063] Assuming the initial projection angle becomes Then we have:

[0064] (6);

[0065] in, For rotation angle, This is the rotated projection data;

[0066] Tomographic images are typically obtained using a filtered back-projection algorithm. Essentially, this involves accumulating projection values ​​from different angles onto the pixels traversed by the ray path. The calculation principle is as follows:

[0067] (7);

[0068] in, Indicates the angle Along Projection data of the direction;

[0069] When the initial projection angle changes, the angles corresponding to all projected images change relative to the original projection. With the change of , the tomographic image reconstruction formula becomes:

[0070] (8);

[0071] in, The image is a reconstructed tomographic image after rotation;

[0072] Let the rotated projection angle :

[0073] (9);

[0074] As can be seen from formula (9), when the projection data changes... After the change, the reconstructed fault image will also have the same rotation angle relative to the original fault image.

[0075] Based on the above analysis, after adjusting the verticality of the projected image and aligning the height of the tomographic image, it is also necessary to consider the scale differences, translation, and rotation present in the tomographic image pair. Considering that the foreground region of the tomographic image of a cylindrical lithium battery is circular, this invention designs a differentiable Lie group Hough circle detection algorithm to extract the circular boundary of the tomographic image.

[0076] like Figure 6 As shown, this invention uses an adaptive edge extraction operator to extract the horizontal boundaries of the tomographic image. Since the operator is sensitive to noise, a 5×5 Gaussian smoothing template is used to Gaussian smooth the input tomographic image pair before edge detection to achieve noise reduction. Then, edge detection is performed. After edge detection, many unwanted structures are detected inside the outermost closed circle. To improve the accuracy of circle detection, holes are filled in the tomographic image based on the region growing principle. After hole filling, a differentiable Lie group Hough circle detection algorithm is used to detect the circle center. Based on region connectivity, the coordinates of the circle center and the diameter of the circle can be calculated. The translation amount is determined according to the location of the circle center (by default, the circle center is moved to the exact center of the image). After moving the centers of all the heterogeneous tomographic images to the center, the translation and alignment of the heterogeneous tomographic images is completed.

[0077] S4. Based on the scale ratio, an image super-resolution reconstruction algorithm is used to realize the scale transformation of the tomographic image.

[0078] Neutron and X-ray CT imaging systems differ in their geometry and imaging methods. Neutron beams are parallel, thus the image formed is the same size as the actual object. X-ray beams are cone-shaped, and during imaging, the size of the projected image increases synchronously with the decrease in distance between the object and the radiation source; the projected image is always magnified relative to the actual object. Furthermore, the camera size and number of pixels in neutron imaging systems and detectors in X-ray imaging systems generally differ. These factors directly lead to differences in the resolution of the reconstructed tomographic images and the size of the foreground region.

[0079] Step S3 moves the foreground region in the tomographic image to the center of the tomographic image (i.e., the center of the circle coincides with the image center), and the diameter of the tomographic image can also be determined. Assume the diameter of the lithium battery in the X-ray tomographic image is... The diameter of the lithium battery in the neutron tomography image is Calculate the size ratio: To achieve image resolution matching, super-resolution reconstruction of the low-resolution image is required.

[0080] Super-resolution reconstruction essentially involves using different interpolation methods to interpolate grayscale values ​​based on the original image information, such as nearest neighbor interpolation, linear interpolation, B-spline interpolation, multi-scale interpolation, and deep learning-based interpolation methods. When using interpolation algorithms for image super-resolution reconstruction, the presence of noise significantly affects the reconstruction quality. Noise can cause artifacts and blurring during the interpolation process, leading to loss of detail or unclear images. Since interpolation algorithms typically assume the image is smooth, noise reduction should be performed first when noise levels are high to improve the quality of the final reconstructed image. In CT imaging, the turntable rotates 360 degrees while the detector samples, resulting in symmetrical projected images that are 180 degrees apart. Considering that noise is random (both in intensity and location), the noise distribution differs between the two projected images that are 180 degrees apart. Based on this characteristic, the two images can be superimposed and averaged to reduce the noise content.

[0081] To improve the quality of the super-resolution reconstructed image, highlighting details and enhancing contrast and sharpness, image enhancement was performed both before and after upsampling the low-resolution image. Furthermore, to avoid over-enhancement, an adaptive histogram equalization method was used to adjust the grayscale of the tomographic image.

[0082] The algorithm flow for super-resolution image reconstruction is as follows: Figure 7 As shown, sine wave overlay is used for noise reduction, adaptive histogram equalization to improve image clarity, homomorphic filtering to sharpen details, bilinear interpolation algorithm for upsampling, and homomorphic filtering to enhance contrast, to achieve super-resolution reconstruction and scale registration of tomographic images.

[0083] The image sharpness is improved after adaptive histogram equalization, and the image details are sharpened after homomorphic filtering.

[0084] Bilinear interpolation algorithms better preserve image details, achieve smooth edge transitions, effectively reduce jagged edges, and make images look more natural and realistic. Furthermore, compared to nearest-neighbor interpolation, the designed method avoids interpolation distortion in high-frequency regions (such as image edges), reducing visual distortion and improving the overall clarity and visual quality of the image. This invention uses this method for upsampling images during interpolation, where... Figure 8 As can be seen, the edges and details of the interpolated image are more prominent, and the image resolution is significantly improved. Bilinear interpolation inevitably introduces some blurring, especially in areas with abundant detail, where it may produce slight artifacts or jagged edges. Filtering can smooth these imperfect areas and improve edge performance. Based on this consideration, homomorphic filtering is applied again to the upsampled image for enhancement, thereby improving the contrast of details.

[0085] S5, Image rotation angle matching.

[0086] Analysis of the cylindrical lithium battery structure revealed that the features of the battery casing are more pronounced, exhibiting a certain degree of similarity in heterogeneous tomographic images. Therefore, when performing rotational matching, the tomographic image of the top casing region of the cylindrical lithium battery was selected as the registration object. After calculating the rotation angle, rotational registration was performed on all tomographic images. To further improve similarity, the image gradient was calculated first, and the central region of the image was selectively preserved.

[0087] Phase cross-correlation can utilize the phase information of an image, exhibiting strong robustness to image rotation. Phase cross-correlation transforms the image from the spatial domain to the frequency domain via Fourier transform, globally considering the image's frequency components without requiring local pixel-by-pixel optimization. Furthermore, since phase cross-correlation primarily relies on the image's phase information, it demonstrates good resistance to noise and local artifacts. Considering the low signal-to-noise ratio and similarity in neutron and X-ray tomographic images, this invention employs this method for matching heterogeneous gradient images, the principle of which is as follows:

[0088] First, the tomographic images to be matched are subjected to Fourier transform according to formula (10) to obtain the spectrum:

[0089] (10);

[0090] in, For neutron tomography image functions in the spatial domain, For spatial domain X-ray tomographic image functions, This is the frequency domain representation of the neutron tomography image after Fourier transform. The frequency domain representation of the X-ray tomographic image after Fourier transform is given by: x is the spatial abscissa, y is the spatial ordinate, u is the frequency domain abscissa, v is the frequency domain ordinate, e is the natural constant, and j is the imaginary unit.

[0091] Next, the cross-power spectrum is calculated based on the spectral data:

[0092] (11);

[0093] in, For cross-power spectrum, for The complex conjugate;

[0094] By performing an inverse Fourier transform on the power spectrum, relevant information about the image to be matched can be obtained. The angle corresponding to the largest correlation coefficient is the desired rotation angle.

[0095] (12);

[0096] in, The correlation coefficient, This is the inverse Fourier transform. The angle is the rotation angle.

[0097] The registration of heterologous tomographic images can be completed by rotating the image at the rotation angle determined by the above process. The registration result is as follows: Figure 8 As shown. To more intuitively demonstrate the effect of image registration, Figure 8 The registered images are superimposed together. It can be seen from the superimposed images that the registration method designed in this invention has high registration accuracy.

[0098] This invention utilizes neutron CT and X-ray CT equipment to image two cylindrical lithium batteries and obtain true projection and tomographic images. The main imaging parameters are shown in Table 1.

[0099] Table 1. X-ray CT imaging parameters of cylindrical lithium batteries

[0100]

[0101] The neutron CT imaging experiment of cylindrical lithium batteries was conducted at the China Advanced Research Reactor (CAR) of the China Institute of Atomic Energy. The main parameters of the system were: full-load power of 60 MW, collimation ratio (L / D) of 70-400, and neutron flux of 4 × 10⁷ ncm⁻² s⁻¹ (20 MW). The main parameters during imaging are shown in Table 2.

[0102] Table 2. Neutron CT Imaging Parameters for Cylindrical Lithium-ion Batteries

[0103]

[0104] Registration effect as Figure 8 As shown in the images, the superimposed images of the registered neutron tomography and X-ray tomography exhibit a high degree of consistency. In tomography 1, the structure of the multi-layer helical coil is perfectly aligned, and the boundary of each layer of the coil appears as a continuous and clear circular shape in the fused image, demonstrating the high accuracy of the registration method in correcting minute curves and interlayer distances. In tomography 2, due to the presence of local material variations and heterogeneous regions, the superimposed image successfully shows the transition details between different materials. Especially with the combined effect of the neutron tomography's penetration information for thick materials and the high-resolution information from X-rays, the heterogeneous regions appear more realistic and complete. The registration results not only clearly preserve the accurate boundary depiction of metallic regions by X-rays but also enhance the characterization ability of neutron rays for non-metallic regions (such as electrolytes or membrane layers).

[0105] To more clearly observe and analyze the registration effect, Figure 8 The superimposed image was rendered in pseudo-color, such as Figure 9 As shown, pseudo-color processing further highlights the differences in properties of different materials, such as the distribution and interface clarity of key components like the current collector, positive electrode, negative electrode, and electrolyte, significantly improving the image's visualization effect. From Figure 9 As can be seen, the light blue area corresponds to the electrolyte distribution, which is particularly prominent in neutron-ray imaging. After registration, it can be precisely superimposed with the metal current collector (highlighted area) and electrode structure (positive and negative electrodes) in X-ray imaging. The central circular area is displayed as black or dark gray, corresponding to the central core of the battery, and its material boundary with the surrounding area can be clearly distinguished. The positive and negative electrodes are represented by pseudo-colors of different brightness, with obvious layering, and the alignment boundaries of each layer remain highly consistent. In addition, the bright area of ​​the current collector has clear boundaries, showing its accurate geometric relationship with the spiral coil, while preserving the complementary information of neutron and X-ray imaging: neutron rays highlight non-metallic materials (such as electrolyte and negative electrode), while X-rays accurately depict the details of the metal current collector and electrodes.

[0106] To more comprehensively evaluate the 3D registration effect, the tomographic image of lithium battery No. 1 was displayed in 3D, and the 3D model was sectioned to further analyze the longitudinal registration accuracy of the image. Figure 10This presentation showcases neutron tomography (CT) images, X-ray tomography (XT) images, and a fused 3D model of lithium-ion battery No. 1. In the neutron tomography images, the electrolyte 1 is clearly distributed, and the penetrating power of neutron imaging on non-metallic materials makes the electrolyte's morphology complete and its boundaries distinct. In the X-ray tomography images, metallic structures such as the positive terminal 2 and the current collector 3 are rendered with high-resolution detail. Through the registered fused image, key components such as electrolyte 1, positive terminal 2, current collector 3, and negative terminal 5 achieve precise spatial alignment. In particular, the morphology and layering of the casing 4 and the negative electrode 7 are highly reproduced in the fused image. The longitudinal cross-sectional images show that the uniformity of the current collector and the positive and negative electrodes throughout the lithium-ion battery is fully presented through the fused image, while preserving the high penetration of the neutron tomography images and the edge sharpness of the X-ray tomography images. The fused 3D model effectively complements the material and structural characteristics of the heterogeneous images, simultaneously demonstrating the spatial relationships of different materials and components within the lithium-ion battery.

[0107] from Figure 11 The registration results show that the X-ray tomographic images clearly reveal the boundary structure of the metal component at the top of the lithium battery, while the neutron rays highlight the distribution of the non-metallic regions. The registered neutron tomographic images and the stacked tomographic images show that the metal component and the non-metallic regions at the top are precisely spatially aligned, with clear layers in the stacked images and highly consistent relative positions of the components. The registration results demonstrate a high degree of consistency between the non-metallic distribution and the metallic structure in the stacked tomographic images.

[0108] Figure 12 This paper presents longitudinal and 3D cross-sectional images of a No. 2 lithium battery. The cross-sectional results demonstrate the superior alignment effect of the proposed method in both longitudinal and 3D spaces. In the longitudinal cross-sectional image, the X-ray tomographic image clearly shows the boundary between the battery casing and the internal current collector, particularly highlighting the high-resolution details of the metallic materials. The neutron beam image emphasizes the distribution characteristics of non-metallic materials such as the electrolyte and separator inside the battery. After registration, the neutron beam and X-ray tomographic images are successfully spatially aligned, clearly displaying both metallic and non-metallic properties within the battery and revealing its complete internal structure. In the 3D cross-section, the fused image exhibits significant hierarchy and detail consistency. The registered 3D model demonstrates excellent spatial consistency; for example, complex geometric structures (such as the winding of the positive and negative electrodes and the distribution of the electrolyte) in the top and bottom regions of the battery are accurately represented in the fused image.

[0109] Table 3. Quantitative Analysis of Registration of Cylindrical Lithium-ion Batteries

[0110]

[0111] As shown in Table 3, the registration performance of the proposed method was quantitatively analyzed using DSC (Diesel similarity coefficient), NMI (normalized mutual information), and GS (gradient similarity) as evaluation metrics. For battery 1, the DSC metrics were 96.78% and 97.72% at the two tomographic locations, demonstrating high spatial alignment accuracy and indicating that the registration algorithm has high matching ability in complex battery structures. The NMI metrics were 41.31% and 43.17%, showing significant effectiveness in information alignment of heterogeneous images. The GS metrics were 79.91% and 81.25%, further validating the consistency of image edge features and accurate alignment of geometric details. According to DSC, the four fault locations of battery 2 all reached high values ​​(42.47%, 98.43%, 98.37%, and 98.67%, respectively). Among them, fault 1, due to its large initial heterogeneity, showed a particularly significant improvement after registration. The NMI ranged from 27.32% to 43.44%, and the consistency of information after registration gradually increased as the fault moved from the top to the bottom. The GS ranged from 63.39% to 68.11%, indicating the stability of the proposed algorithm in maintaining geometric alignment and feature fusion restoration.

[0112] This invention designs Figure 7 The super-resolution reconstruction module shown can improve image quality while performing image scale transformation and achieving image resolution calibration. To quantitatively and objectively evaluate the performance of this module in improving image quality, the peak signal-to-noise ratio (PSNR) and average gradient (AG) of the images before and after super-resolution were compared, as shown in Table 4.

[0113] Table 4. Quantitative Analysis of Super-Resolution Reconstruction

[0114]

[0115] As shown in Table 4, the designed super-resolution reconstruction module has a significant effect on improving image quality. The improvements in both PSNR and AG indicators demonstrate the high stability and adaptability of this method in resolving cylindrical lithium battery tomographic images. The significant improvement in PSNR indicates the reconstruction method's advantages in reducing image noise and improving signal quality, especially for images with low initial quality, effectively improving signal fidelity and ensuring significant optimization of visual and quantization quality in the reconstructed image. Simultaneously, the increase in average gradient verifies the method's superior performance in detail enhancement and edge feature strengthening. Analysis of multi-tomographic images reveals that the super-resolution reconstruction method effectively recovers high-frequency information in the image, making the details of complex structures clearer and more discernible. This effect not only improves image sharpness and contrast but also enhances the ability to represent multi-layered complex geometric structures.

[0116] The image registration method designed in this invention is for cylindrical lithium batteries, and should be equally effective for samples with circular outlines. To verify this hypothesis, a three-dimensional image registration experiment was designed using a micro DC motor as the research object. The main parameters of the experiment are shown in Tables 5 and 6.

[0117] Table 5. X-ray CT scan parameters

[0118]

[0119] Table 6. Neutron X-ray CT Scan Parameters

[0120]

[0121] The X-ray tomographic image size is 370×370, and the neutron beam tomographic image size is 1058×1058. The registration results using the registration method designed in this invention are as follows: Figure 13 As shown, the detailed features of the registered image demonstrate a high degree of consistent alignment of image boundaries, and structural features exhibit good consistency in the overlay image, avoiding misalignment or blurring caused by differences in heterogeneous imaging. The Region of Interest (ROI) further validates the high precision of the registration algorithm; the geometric features and positions in this region are completely coincident in the neutron and X-ray overlay images, with clear boundaries, indicating the algorithm's excellent ability to correct subtle differences in heterogeneous materials in space. Furthermore, region analysis reveals the accurate location of non-metallic materials within the structure after fusion, while the metallic parts exhibit clear contrast and integrity. Further analysis of material properties shows that neutron imaging highlights the thicker non-metallic regions, while X-ray imaging captures the high-resolution characteristics of metallic components. The fused image fully preserves the advantages of both imaging modalities, achieving a comprehensive presentation of material properties and structural features.

[0122] To quantitatively evaluate the registration accuracy of the images, we calculated the DSC, NMI, and GS of the registered images. The results show that the similarity of the images is significantly improved after using the proposed method for heterogeneous image registration. The numbers in parentheses in Table 7 represent the increase in the coefficients after registration. As can be seen from Table 7, all three indicators of the registered images are significantly improved, verifying the high-precision registration capability of the proposed method. Among them, the DSC reached 88.79%, 81.18%, and 57.35% in the three images, respectively, showing a high degree of overlap between the structures of the images. In particular, image 1 has the highest registration accuracy due to its relatively regular geometric features. NMI reflects the degree of information sharing between images. The NMI of image 1 and image 2 are 0.5674 and 0.4722, respectively, reflecting the effectiveness of the algorithm in aligning information between heterogeneous images. GS further shows the consistency of edge features after registration. The GS of the three images is significantly improved, reaching a maximum of 83.85%.

[0123] Table 7. Quantitative Analysis of Registration of Micro DC Motors

[0124]

[0125] To more intuitively demonstrate the effect of 3D registration of heterogeneous images, we visualized the tomographic images in 3D, weighted and fused the 3D model, and rendered it in pseudo-color.

[0126] Figure 14 This paper presents a heterogeneous fusion model generated based on the proposed method. The integrity of the internal structure of the micro DC motor and the accurate reconstruction of the material distribution can be clearly observed from the weighted fusion image. This reflects the powerful capability of the proposed method in characterizing heterogeneous materials and demonstrates its superior performance in capturing details.

[0127] Metallic materials exhibit high contrast and uniform distribution in the fused image, and complex geometric shapes are accurately reproduced using the proposed method, further validating the algorithm's efficiency in registering metal-nonmetal boundaries. Compared to single-modal imaging, the fusion model retains both the high resolution of X-ray imaging and the penetrating power of neutron imaging, enhancing the representation of multiple materials in the same view and improving the hierarchical representation of minute components.

[0128] The above processing results show that the three-dimensional registration method for neutron / X-ray heterogeneous images of cylindrical multilayer structures provided by the present invention can solve the problem of inconsistent image size, grayscale distribution and spatial position caused by differences in geometric characteristics and parameters in heterogeneous imaging systems, and realize high-precision registration of heterogeneous cross-modal three-dimensional images of cylindrical workpieces.

[0129] The following describes the three-dimensional registration device for neutron / X-ray heterogeneous images of a cylindrical multilayer structure provided by the present invention. The three-dimensional registration device for neutron / X-ray heterogeneous images of a cylindrical multilayer structure described below can be referred to in correspondence with the three-dimensional registration method for heterogeneous images described above.

[0130] like Figure 15 As shown, the three-dimensional registration device for cylindrical multilayer neutron / X-ray heterogeneous source images provided by the present invention includes:

[0131] Projected image verticality adjustment module 231 is used to adjust the tilt of the projected image;

[0132] The tomographic image height mapping module 232 is used to map the height of neutron and X-ray tomographic images;

[0133] Translation matching module 233 is used for two-dimensional translation registration of tomographic images;

[0134] The super-resolution reconstruction module 234 is used to upsample low-resolution images to achieve image scale registration;

[0135] The image rotation matching module 235 is used to match the rotation angle of heterogeneous tomographic images.

[0136] The heterogeneous image three-dimensional registration device provided by the present invention can be used to perform the above-mentioned three-dimensional registration method for neutron / X-ray heterogeneous images of cylindrical multilayer structures, so as to solve the problem of inconsistent image size, gray scale distribution and spatial position caused by differences in geometric characteristics and parameters in related technologies. It can effectively correct the geometric and modal differences between heterogeneous images, achieve accurate alignment of metal and non-metal regions, and significantly improve the robustness and accuracy of registration.

[0137] Figure 16 A schematic diagram of the physical structure of an electronic device is provided. This electronic device may include a processor 2410, a communications interface 2420, a memory 2430, and a communication bus 2440. The processor 2410, communications interface 2420, and memory 2430 communicate with each other via the communication bus 2440. The processor 2410 can call logical instructions in the memory 2430 to execute various steps of the three-dimensional registration method for cylindrical multilayer neutron / X-ray heterogeneous source images. Furthermore, the logical instructions in the memory 2430 can be implemented as software functional units and, when sold or used as independent products, can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this invention. The aforementioned storage media include: USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks or optical disks, and other media that can store program code.

[0138] On the other hand, the present invention also provides a computer program product, which includes a computer program that can be stored on a non-transitory computer-readable storage medium. When the computer program is executed by a processor, the computer is able to perform each step of the above-described three-dimensional registration method for neutron / X-ray heterogeneous images of cylindrical multilayer structures.

[0139] In another aspect, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the various steps of the above-described method for three-dimensional registration of cylindrical multilayer neutron / X-ray heterogeneous source images.

[0140] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0141] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0142] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for three-dimensional registration of neutron / X-ray heterogeneous source images of a cylindrical multilayer structure, characterized in that, include: Edge extraction operators and PCA principal component analysis are used to adjust the verticality of image pairs; The verticality adjustment is achieved based on the projected image; The image pair to be registered consists of a neutron projection image and an X-ray projection image; the verticality adjustment requires gray-level normalization of the input image pair, normalizing the gray-level values ​​to the range [0,1]; for the normalized projection image, the OTU algorithm is used to perform threshold segmentation to extract the edge contours of the projection image; the PCA principal component analysis refers to: firstly, statistically analyzing the coordinates of the boundary pixels, and then using PCA to determine the first principal component, which is the direction of the axis of symmetry; the key to threshold segmentation in the OTU algorithm is determining the segmentation threshold, by selecting a segmentation threshold T to divide the projection image into foreground and background parts, such that the inter-class variance between the foreground and background is minimized. Maximum; for the acquired edge contour image, the Sobel operator is used to extract the vertical boundary of the foreground region; The edge contour extraction module determines the height mapping of heterogeneous tomographic images, mapping neutron and X-ray tomographic images to the same height; the height mapping determines the height corresponding to the first and last tomographic images of the cylindrical multilayer structure by determining the horizontal boundary of the projected image; A circle detection algorithm is used to detect the center of the circle and determine the amount of translation, so that the heterogeneous tomographic images are translated and aligned. Based on the scale ratio, an image super-resolution reconstruction algorithm is used to achieve scale transformation super-resolution reconstruction of tomographic images; Image rotation matching is achieved based on feature tomography and phase cross-correlation: the tomographic image of the top shell region of a cylindrical multilayer structure is used as the registration object, the rotation angle is calculated, and then all tomographic images are rotated and registered; the gradient of the tomographic image is calculated first and the central region of the tomographic image is selectively preserved, and the phase correlation method is used to match the heterogeneous gradient images.

2. The heterogeneous image three-dimensional registration method according to claim 1, characterized in that, The height mapping method is as follows: an adaptive edge extraction operator is used to detect the horizontal boundary of the projected image, the pixels at the horizontal boundary are sorted, and the average of the row coordinates of the top 20% of the pixels is taken as the height of the horizontal boundary.

3. The heterogeneous image three-dimensional registration method according to claim 1, characterized in that, Before edge detection, the input tomographic image is Gaussian smoothed using a Gaussian smoothing template. Then, edge detection is performed. After edge detection, holes in the tomographic image are filled based on the region growing principle. After hole filling, the center of the circle is detected using a differentiable Lie group Hough circle detection algorithm. The coordinates of the center and the diameter of the circle can be calculated based on the region connectivity. The translation amount is determined according to the location of the center. After moving the centers of all the heterogeneous tomographic images to the center, the translation and alignment of the heterogeneous tomographic images is completed.

4. The heterogeneous image three-dimensional registration method according to claim 1, characterized in that, The image super-resolution reconstruction algorithm includes using sinusoidal graph overlay for noise reduction, adaptive histogram equalization to improve image clarity, homomorphic filtering to sharpen details, bilinear interpolation algorithm for upsampling, and homomorphic filtering to enhance contrast, thereby achieving super-resolution reconstruction and scale registration of tomographic images.

5. An electronic device, comprising: The system includes a processor, a communication interface, and a communication bus, wherein the processor, communication interface, and memory communicate with each other via the communication bus; characterized in that the processor calls logical instructions in the memory to execute each step of the three-dimensional registration method for cylindrical multilayer neutron / X-ray heterogeneous source images as described in any one of claims 1-4.

6. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When executed by a processor, the computer program performs the steps of the three-dimensional registration method for neutron / X-ray heterogeneous images of a cylindrical multilayer structure as described in any one of claims 1-4.

Images