Computed tomography reconstruction method and system for sparse angle of cylindrical lithium battery

Abstract

The invention relates to a computed tomography reconstruction method and system for a sparse angle of a cylindrical lithium battery, and the method comprises the steps: obtaining a cylindrical lithium battery projection image sequence as original projection data, and carrying out the diagonal interpolation processing of the original projection data, and generating complete sinogram data; performing Radon inverse transformation on the sinogram data to generate an initial iteration reconstruction image and taking the initial iteration reconstruction image as an iteration initial value, and performing iteration optimization by adopting a simultaneous algebraic reconstruction technology as a data fidelity term to obtain an optimized reconstruction image; designing a tangential smoothing filter, and filtering the optimized reconstructed image to suppress sparse artifacts; a radial weighting template is designed, and differential smoothing processing is carried out on the reconstructed image after sparse artifact suppression according to the distance between pixels and the center of the image; and performing reconstruction image optimization, sparse artifact suppression and differential smoothing processing circularly until a preset number of iterations is reached, thereby obtaining a final reconstruction image. According to the method disclosed by the invention, detail information of the image is effectively recovered while sparse artifacts are suppressed.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of industrial computed tomography, in particular to a cylindrical lithium battery sparse angle computed tomography reconstruction method and system thereof. BACKGROUND

[0002] With the growth of global energy demand and the enhancement of environmental protection awareness, lithium ion batteries are widely used in electric vehicles, aerospace, military equipment and consumer electronics due to their high energy density, long cycle life, no memory effect and other outstanding advantages. The safety performance of lithium ion batteries is particularly important, so accurate detection and evaluation of the internal structure and possible defects of the battery have become a key link in research and production.

[0003] Industrial computed tomography (CT) as a kind of efficient non-destructive testing technology can accurately obtain two-dimensional tomographic images and three-dimensional structural information of the internal structure of lithium batteries, which is convenient for researchers and engineers to analyze and evaluate the internal microstructure, electrolyte distribution, electrode performance and possible defects of the battery. However, traditional industrial CT usually adopts dense angle scanning, i.e. the interval between each angle is very small, which requires a long exposure time and increases the risk of battery radiation activation, which may have a negative impact on the performance and safety of the battery.

[0004] To solve the above problems, sparse angle CT technology has gradually attracted attention. This technology significantly reduces the dose and activation risk of battery irradiation by reducing the number of projection angles, but due to the sparsity of projection data, the use of traditional reconstruction algorithms often leads to a decrease in image quality and the generation of obvious sparse artifacts, which seriously affects the clarity and accuracy of the image and limits the widespread application of sparse angle CT technology. Therefore, it is urgent to develop a reconstruction method that can effectively improve the quality of sparse angle CT images and reduce sparse artifacts. SUMMARY

[0005] To solve the problem of severe image artifacts and low quality in traditional sparse angle CT reconstruction, the present disclosure proposes a cylindrical lithium battery sparse angle computed tomography reconstruction method to solve the above problems.

[0006] According to an aspect of the present disclosure, a cylindrical lithium battery sparse angle computed tomography reconstruction method is provided, comprising: S10, obtaining a sequence of projection images of a cylindrical lithium battery as original projection data, performing diagonal interpolation processing on the original projection data to generate complete sinogram data; S20, performing Radon inverse transform on the interpolated sinogram data to generate an initial iterative reconstruction image; S30, taking the initial iterative reconstruction image as an iterative initial value, using simultaneous algebraic reconstruction technique as a data fidelity term to iteratively optimize the initial iterative reconstruction image to obtain an optimized reconstruction image; S40, based on the rotational symmetry characteristics of the cylindrical lithium battery, designing a tangential smoothing filter to filter the optimized reconstruction image to suppress sparse artifacts to obtain a reconstruction image after suppressing sparse artifacts; S50, designing a radial weighting template with the rotation center as the reference, dynamically adjusting the smoothing weight according to the distance between the pixel and the image center, and performing differential smoothing processing on the reconstruction image after suppressing sparse artifacts; S60, cyclically performing steps S30-S50 until a preset iteration number is reached to obtain a final reconstruction image.

[0007] Preferably, the diagonal interpolation processing is performed on the original projection data to generate complete sinogram data, including: identifying the missing projection angles in the original projection data, and analyzing the gray similarity of the pixels along the diagonal line in the sinogram of the original projection data; interpolating along the diagonal direction of the sinogram, and preferentially selecting the pixel points similar to the diagonal line in the adjacent projection angles; generating the sinogram by using the linear interpolation method.

[0008] Preferably, the diagonal interpolation processing is performed on the original projection data to generate complete sinogram data, and further including: performing filter back projection reconstruction on the sinogram generated by the linear interpolation, and evaluating the interpolation effect by the peak signal-to-noise ratio, the structural similarity and the mean square error, which is expressed as: , , , In the formula, PSNR represents the peak signal-to-noise ratio, SSIM SSIM represents the structural similarity, and MSE represents the mean square error, n N represents the gray level of the image, f 1 and f 2 represent the standard image and the image to be evaluated, respectively, , , , represent the mean and variance of the images f 1 and f 2, respectively, Cov represents the covariance of the images f 1 and f 2, and are constants, M and Nrespectively represent the width and height of the image, represent the coordinates of a pixel in the image.

[0009] Preferably, the initial iterative reconstruction image is iteratively optimized by using simultaneous algebraic reconstruction technique as the data fidelity term, denoted as: , wherein, is the reconstruction image of the i-th iteration, k is the reconstruction image of the i+1-th iteration, is the reconstruction image of the i-th iteration, k is a diagonal matrix whose elements are determined by the column vector norm of the system matrix C is the weighted system matrix used to adjust the update step size, A is the relaxation factor to control the amplitude of each update, A is the projection data obtained from computed tomography. b Preferably, a tangential smoothing filter is designed, whose mathematical model is:

[0010] Preferably, a radial weighting template is designed with the rotation center as the reference, denoted as: , wherein, is the pixel with coordinates 1 and w 2 are the interpolation weights, w is the row coordinate of the tangential neighborhood pixel. Preferably, a radial weighting template is designed with the rotation center as the reference, denoted as:

[0011] , wherein, is the radial weighting sparse template, W and i respectively represent the horizontal and vertical coordinates of the pixel, j and respectively represent the horizontal and vertical coordinates of the image symmetry center, is the effective radius of the weighting template. R Preferably, the smoothing weight is dynamically adjusted according to the distance between the pixel and the image center, and the reconstruction image after the sparse artifact is suppressed is differentially smoothed, denoted as:

[0012] , wherein, represents the image after the simultaneous algebraic reconstruction technique calculation, represents the image after the tangential smoothing, represents the radial weighting image. f ​​

[0013] According to an aspect of the present disclosure, a computer tomography reconstruction system for sparse-angle cylindrical lithium battery is provided, comprising: A complete sinogram data generation module acquires a sequence of projection images of a cylindrical lithium battery as original projection data, performs diagonal interpolation processing on the original projection data, and generates complete sinogram data; An initial iterative reconstruction image generation module performs Radon inverse transformation on the interpolated sinogram data to generate an initial iterative reconstruction image; A reconstruction image optimization module takes the initial iterative reconstruction image as an iterative initial value, adopts simultaneous algebraic reconstruction technology as a data fidelity term, and iteratively optimizes the initial iterative reconstruction image to obtain an optimized reconstruction image; A reconstruction image sparse artifact suppression module, based on the rotational symmetry characteristics of the cylindrical lithium battery, designs a tangential smoothing filter, performs filter processing on the optimized reconstruction image to suppress sparse artifacts, and obtains a reconstruction image after suppressing sparse artifacts; A reconstruction image differential smoothing processing module designs a radial weighting template with the rotation center as the reference, dynamically adjusts the smoothing weight according to the distance between the pixel and the image center, and performs differential smoothing processing on the reconstruction image after suppressing sparse artifacts; A final reconstruction image acquisition module cyclically executes the reconstruction image optimization, sparse artifact suppression, and differential smoothing processing modules until a preset iteration number is reached, and obtains a final reconstruction image.

[0014] According to an aspect of the present disclosure, an electronic device is provided, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to execute the above-mentioned computer tomography reconstruction method for sparse-angle cylindrical lithium battery.

[0015] According to an aspect of the present disclosure, a computer-readable storage medium having computer program instructions stored thereon is provided, the computer program instructions being executed by a processor to implement the above-mentioned computer tomography reconstruction method for sparse-angle cylindrical lithium battery.

[0016] Compared with the prior art, the present disclosure has the following beneficial effects: 1) The present disclosure effectively solves the problems of stripe artifacts and edge blurring caused by sparse-angle projection data through diagonal interpolation processing and the design of tangential smoothing filter, significantly improving the image quality.

[0017] 2) The present disclosure combines the dynamic smoothing strategy of the radial weighting template, which not only suppresses artifacts but also perfectly preserves the microstructure details of the battery electrode, providing a more reliable image basis for lithium battery internal defect detection.

[0018] 3) The present disclosure significantly improves the convergence speed and greatly reduces the consumption of computing resources by optimizing the iteration and periodic tangential filtering in parallel through SART.

[0019] It should be understood that the above general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present disclosure.

[0020] Other features and aspects of the present disclosure will become apparent from the following detailed description of example embodiments, taken in conjunction with the accompanying drawings. BRIEF DESCRIPTION OF DRAWINGS

[0021] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments consistent with the present disclosure and serve to explain the technical solutions of the present disclosure together with the specification.

[0022] Figure 1 A flow chart of a computer tomography reconstruction method for sparse angles of a cylindrical lithium battery is shown; Figure 2 A flow chart of a SART_TI iterative reconstruction algorithm in an embodiment of the present disclosure is shown; Figure 3 A schematic diagram of a sinogram diagonal interpolation in an embodiment of the present disclosure is shown; Figure 4 A schematic diagram of the influence of diagonal interpolation on a reconstructed image in an embodiment of the present disclosure is shown; Figure 5 A schematic diagram of the variation trend of sparsity in an embodiment of the present disclosure is shown; Figure 6 A schematic diagram of the geometric relationship of a cylindrical lithium battery slice image in an embodiment of the present disclosure is shown; Figure 7 A diagram of the geometric relationship of a pixel neighborhood in an embodiment of the present disclosure is shown; Figure 8 A simulation phantom iterative convergence curve diagram in an embodiment of the present disclosure is shown; Figure 9 An X-ray lithium battery image iterative convergence curve diagram in an embodiment of the present disclosure is shown; Figure 10 A neutron lithium battery image iterative convergence curve diagram in an embodiment of the present disclosure is shown; Figure 11 A diagram showing the influence of different modules on the quality coefficient of a reconstructed image in an embodiment of the present disclosure is shown; Figure 12 A block diagram of a computer tomography reconstruction system structure for sparse angles of a cylindrical lithium battery in an embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0023] Various exemplary embodiments, features, and aspects of the present disclosure will be described below with reference to the accompanying drawings. The same reference numbers in different drawings represent the same or similar elements. Although various aspects of the embodiments are illustrated in the drawings, the drawings are not necessarily drawn to scale unless specifically indicated.

[0024] The term "exemplary" is used herein to mean "serving as an example, instance, or illustration." Any implementation described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other implementations.

[0025] The term "and / or", used herein only to describe the connection relationship of associated objects, means that there can be three relationships, for example, A and / or B, which can represent the existence of A alone, the existence of A and B at the same time, and the existence of B alone. In addition, the term "at least one" herein means any one of the plurality or any combination of at least two of the plurality, for example, including at least one of A, B, and C, which can mean including any one or more elements selected from the set consisting of A, B, and C.

[0026] In addition, in order to better illustrate the present disclosure, numerous specific details are given in the specific embodiments below. Those skilled in the art should understand that the present disclosure can also be implemented without certain specific details. In some examples, methods, means, elements and circuits well known to those skilled in the art are not described in detail, in order to highlight the main idea of the present disclosure.

[0027] In order to make the purpose, technical solutions and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0028] Embodiment 1 Based on the above idea, the present disclosure embodiment proposes a computer tomography reconstruction method for sparse angles of cylindrical lithium batteries. Figure 1 A flowchart of a computer tomography reconstruction method for sparse angles of cylindrical lithium batteries is shown. The method comprises: S10, acquiring a projection image sequence of a cylindrical lithium battery as original projection data, performing diagonal interpolation processing on the original projection data to generate complete sinogram data; S20, performing Radon inverse transform on the interpolated sinogram data to generate an initial iterative reconstruction image; S30, taking the initial iterative reconstruction image as an iterative initial value, performing iterative optimization on the initial iterative reconstruction image by using a simultaneous algebraic reconstruction technique as a data fidelity term to obtain an optimized reconstruction image; S40, based on the rotational symmetry characteristic of the cylindrical lithium battery, designing a tangential smoothing filter to perform filter processing on the optimized reconstruction image to suppress sparse artifacts, to obtain a reconstruction image after sparse artifact suppression; S50, designing a radial weighting template with the rotation center as a reference, and dynamically adjusting the smoothing weight according to the distance of a pixel from the image center, to perform differential smoothing processing on the reconstruction image after sparse artifact suppression; S60, performing steps S30-S50 in a loop until a preset iteration number is reached, to obtain a final reconstruction image.

[0029] The flowchart of the SART_TI iterative reconstruction algorithm in the embodiment of the present disclosure is as shown in Figure 2 The specific steps are as follows: parameter initialization; generating sinogram data through diagonal interpolation processing; further obtaining an iterative initial value; obtaining an optimized reconstruction image through simultaneous algebraic reconstruction technique (SART) iterative updating; designing a tangential smoothing filter to suppress sparse artifacts of the reconstruction image, and performing differential smoothing processing when Mod( k , regu Freq) = 0 (the current iteration number); k Radial weighting smoothing, calculating the radial weight of a pixel, the closer to the center, the lower the weight (weaker smoothing), and the weight in the edge region is 1 (strong smoothing); multiplying the tangential filtered image and the weight template point by point to output a differential smoothing image. The computer tomography reconstruction method for sparse angles of cylindrical lithium batteries proposed in the embodiment of the present disclosure specifically includes the following steps: S10, obtaining a sequence of projection images of a cylindrical lithium battery as original projection data, and performing diagonal interpolation processing on the original projection data to generate complete sinogram data.

[0030] In the embodiment, a real lithium battery image is collected by an X-ray and neutron CT system, and data collection and processing are performed by an X-ray digital imaging system developed by the applicant. The system includes a 225kV micro-focus X-ray source, a high-resolution amorphous silicon flat panel detector, a 5-axis control system, and auxiliary components. The X-ray uses an XWT type X-ray source of WorX GmbH, and the main parameters are shown in Table 1. The detector is Perkin Elmer XRD0822, and the key technical parameters are shown in Table 2.

[0031] Table 1 Main technical indicators of XWT type X-ray source

[0032] Table 2 Main technical parameters of Perkin Elmer XRD0822 detector

[0033] The neutron CT imaging experiment of cylindrical lithium battery was carried out in China Advanced Research Reactor of China Institute of Atomic Energy. The main parameters of the system are as follows: the full power of the reactor is 60 MW, the collimation ratio (L / D) is 70-400, and the neutron flux is 4x107ncm-2s-1(20 MW). The main parameters during imaging are shown in Table 3.

[0034] Table 3 Neutron CT imaging parameters of cylindrical lithium battery

[0035] The simulation data of lithium battery was designed by using matlab, and the image size was 512x512.

[0036] In the process of CT iterative reconstruction, the definition of initial image matrix is an important consideration factor. The common method is to initialize the initial image matrix to zero, or to use the image reconstructed by filtered back projection (FBP) as the initial image. Studies have shown that using the image reconstructed by FBP as the initial image can accelerate the convergence of the algorithm and improve the signal-to-noise ratio of the reconstructed image. In this embodiment, the idea of existing research is used for reference, and the original sinogram is interpolated to reduce the sparsity of the projection, and then the interpolated sinogram is reconstructed by FBP.

[0037] Figure 3 The sinograms of different sparsity corresponding to the X-ray tomographic images of cylindrical lithium battery are shown. By analyzing the sinogram of lithium battery, it is found that the pixels along the diagonal line have higher similarity, so the diagonal interpolation method is used to obtain the interpolated sinogram.

[0038] Further, the original projection data is processed by diagonal interpolation to generate complete sinogram data, including: identifying the missing projection angles in the original projection data, analyzing the gray similarity of pixels along the diagonal line in the sinogram of the original projection data; interpolating along the diagonal direction of the sinogram, preferentially selecting the pixel points similar to the diagonal line in the adjacent projection angles; and generating the sinogram by using linear interpolation method.

[0039] It can be seen that the interpolated sinusoidal graph has a high degree of smoothness, but the enlarged image of the local area shows that the edge structure has a certain degree of jaggedness after interpolation. Compared with the original sinusoidal graph, the sparse artifacts in the image reconstructed after diagonal interpolation are significantly suppressed, and some details and contours are effectively restored. In order to objectively quantify the impact of diagonal interpolation, the tomographic images reconstructed with 360 projections are used as the standard, and the structural similarity (SSIM), mean square error (MSE) and peak signal-to-noise ratio (PSNR) are used as evaluation indicators to evaluate the impact of diagonal interpolation on the reconstructed image. The interpolation effect is evaluated by peak signal-to-noise ratio, structural similarity and mean square error, which are expressed as: , , , Where PSNR represents the peak signal-to-noise ratio, SSIM represents structural similarity, MSE represents mean square error, n Indicates the grayscale level of the image. This embodiment uses the normalized image, so set n =1, f 1 and f 2 represent the standard image and the image to be evaluated, 、 、 、 Represents images respectively f 1 and f 2 mean and variance, Representing an image f 1 and f The covariance of 2, and is a constant, M and N Represents the width and height of the image respectively, Represents the coordinates of pixels in the image.

[0040] Table 4 Quantitative analysis of algorithm performance

[0041] Table 4 shows that applying FBP to the interpolated sinograms improves the image's PSNR and SSIM, while also reducing the MSE. However, as the number of projections increases, the magnitude of the changes in the image's PSNR, SSIM, and MSE tends to decrease, indicating that the sinogram diagonal interpolation method is more effective for high-sparse reconstruction. Figure 4 Schematic diagram showing the effect of diagonal interpolation on the reconstructed image in the embodiment of the present disclosure. Figure 4It can be seen that the uninterpolated sparse projection data (e.g., 20 projection angles) exhibits significant discontinuity and missing data; diagonal interpolation fills the gaps, forming a continuous, complete sinusoidal graph. When FBP reconstruction is performed directly on the original sparse sinusoidal graph, the image exhibits severe streak artifacts and blurred structures. However, when the same reconstruction algorithm is applied to the diagonally interpolated sinusoidal graph, artifacts are significantly reduced, and the electrode edges and internal pore structure are more clearly defined.

[0042] Figure 5 A schematic diagram of the sparsity variation trend in the disclosed embodiment is shown. Experimental data comparison visually demonstrates the performance difference between the disclosed embodiment's method (SART_TI) and the traditional SART algorithm under different sparsity angle conditions (varying number of projections), verifying its robustness and rapid convergence at low projection numbers.

[0043] S20 , performing an inverse Radon transform on the interpolated sinusoidal data to generate an initial iteratively reconstructed image.

[0044] In traditional sparse angular CT reconstruction, directly performing an inverse Radon transform on the original sparse sinusoidal image can lead to severe star-shaped artifacts and noise amplification. Existing techniques typically use zero-value initialization or low-pass filtering preprocessing, but these methods struggle to balance computational efficiency and initial image quality. This embodiment significantly improves the fidelity of the initial image by combining diagonal interpolation preprocessing with an adaptive inverse Radon transform.

[0045] S30 , using the initial iterative reconstructed image as an iteration initial value, adopting a simultaneous algebraic reconstruction technique as a data fidelity item, and iteratively optimizing the initial iterative reconstructed image to obtain an optimized reconstructed image.

[0046] In this embodiment, the simultaneous algebraic reconstruction technique (SART) is an algorithm that combines the efficiency of ART with the robustness of SIRT. Its update strategy aims to accelerate convergence and improve image quality by considering all ray paths and updating the reconstructed image simultaneously. Using the simultaneous algebraic reconstruction technique as the data fidelity term, the initial iteratively reconstructed image is iteratively optimized, expressed as: , Where, For the k +1 reconstructed image of the iteration, For the k The reconstructed image of the iteration, C is a diagonal matrix whose elements are determined by the system matrix A The column vector norm of A is the weighted system matrix, used to adjust the update step size, To control the relaxation factor of each update amplitude, bFor the projection data obtained from computed tomography. The denominator in represents the weighted system matrix A Adjustment of the update step to optimize the reconstruction results and the convergence speed of the algorithm.

[0047] SART can effectively balance the reconstruction speed and image quality. By considering the contribution of all ray paths simultaneously in each iteration, SART not only speeds up the convergence, but also improves the image quality by fine-tuning the reconstructed image, especially in cases of incomplete data or high noise. In addition, the parallel design of SART supports modern computing platforms, making large-scale calculations feasible, which is particularly important for handling large amounts of data and achieving fast reconstruction. Based on the above considerations, this embodiment adopts SART as the data fidelity term of the sparse-angle reconstruction process of cylindrical lithium batteries.

[0048] S40, based on the rotational symmetry characteristics of cylindrical lithium batteries, a tangential smoothing filter is designed to filter the optimized reconstructed image to suppress sparse artifacts, obtaining a reconstructed image after suppressing sparse artifacts.

[0049] In this embodiment, if only SART is used for iterative calculation, there will be sparse artifacts in the reconstructed image. Therefore, it is necessary to design special rules for subsequent processing of the image. How to formulate effective constraint principles according to the characteristics of the image to suppress artifacts while maximizing the recovery of image information is an important research direction in the field of iterative reconstruction.

[0050] As Figure 6 shown, for tomographic images such as cylindrical lithium batteries with rotational symmetry, the pixels in the image foreground region have the highest similarity with their tangential neighbors with the COR (Center of Rotation) as the origin. Figure 7 In the formula, pixels A and B constitute the tangential neighborhood of pixel x). Based on the above analysis, a tangential smoothing filter is designed based on the tangential self-similarity of the image. After completing the SART iterative process, the tangential smoothing filter is used to perform secondary recovery of the structure in the image. The mathematical model of the tangential smoothing filter is: , In the formula, is the pixel with coordinates , w 1 and w 2 are interpolation weights, is the row coordinate of the tangential neighborhood pixel.

[0051] As Figure 7 shown, the pixel xThe tangential neighborhood of the pixel point is not strictly located at the pixel point, and the gray value of the tangential neighborhood needs to be obtained by interpolation using the surrounding pixels. The linear interpolation method is used to calculate the formula w 1 and w 2. Assuming that the distance between adjacent pixels is 1, the distance between A and is 1, and the distance between B and w is 2. w

[0052] S50, a radial weighting template is designed with the rotation center as the reference, the smoothing weight is dynamically adjusted according to the distance of the pixel from the image center, and the reconstructed image after the sparse artifact is suppressed is differentially smoothed.

[0053] Since the closer to the center, the better the details and structure of the image are recovered, and on the contrary, the more serious the sparse artifact is. According to this rule, in the tangential smoothing of the image, a radial weighting template is introduced in this embodiment, so that the pixels in different sparse regions are smoothed to different degrees. A radial weighting template is designed with the rotation center as the reference, which is represented as: , In the formula, W is a radial weighting sparse template, i and j respectively represent the horizontal and vertical coordinates of the pixel, and respectively represent the horizontal and vertical coordinates of the image symmetry center, R is the effective radius of the weighting template.

[0054] Further, the smoothing weight is dynamically adjusted according to the distance of the pixel from the image center, and the reconstructed image after the sparse artifact is suppressed is differentially smoothed, which is represented as: , In the formula, represents the image after the simultaneous algebraic reconstruction technique is calculated, represents the image after the tangential smoothing, f represents the image after the radial weighting.

[0055] S60, steps S30-S50 are repeatedly executed until a preset iteration number is reached, and a final reconstructed image is obtained.

[0056] ​​In this embodiment, when performing iterative reconstruction, the setting of parameters directly affects the quality of the reconstructed image and the convergence speed of the algorithm. In order to analyze the influence of different parameters on the performance of the algorithm and determine the optimal reconstruction parameters, the real lithium battery X-ray tomographic image is taken as the reconstruction target, and the effective radius of the radial weighting template, the symmetric center and the frequency of the tangential smoothing are analyzed according to the idea of the control variable method.

[0057] The tangential smoothing filter uses the similarity of the diagonal neighborhood of the pixels to restore the image information. The filter will blur the details and edges in the image while restoring the image information. The higher the frequency used, the more blurred the details and edges in the image will be. Therefore, it is necessary to determine a suitable filtering frequency to ensure that the image has sufficient contrast while restoring the image information. In this embodiment, the effective radius of the radial weighting template is set to 125, the symmetric center is set to (131, 135), and the number of projections is 36. The experimental results are shown in Table 5.

[0058] Table 5 Influence of tangential smoothing filter frequency on reconstruction effect

[0059] In Table 5, Mod represents the modulo operation, k k represents the current iteration number. When Mod(k, 1)=0, the tangential smoothing filter is applied to the image. From the information in the table, it can be seen that applying the tangential smoothing filter to the image once every 2 iterations can best restore the image information.

[0060] In CT imaging, the object rotates continuously with the turntable at a pre-set speed, and the detector synchronously performs image acquisition to form a sparse projection sequence. The rotation center of the turntable is equivalent to the symmetric center of the reconstructed tomographic image. In this embodiment, the symmetric center is designed for the image when performing radial weighting and tangential filtering. Therefore, it is necessary to study the influence of the calculation error of the symmetric center on the quality of the reconstructed image. In this embodiment, the effective radius of the radial weighting template is set to 125, the tangential smoothing filter frequency is set to Mod(k, 2), and the number of projections is 36. The experimental results are shown in Table 6.

[0061] Table 6 Influence of symmetric center on reconstruction effect

[0062] The actual symmetric center of the lithium battery X-ray tomographic image in the experiment is (131, 135). From Table 6, it can be seen that when the symmetric center is the actual value, the image quality is the highest, and the quality of the reconstructed image gradually decreases as the error of the symmetric center increases. In addition, when the error of the horizontal and vertical coordinates of the symmetric center is within 5 pixels, the image quality decreases at a lower rate, indicating that SART_TI has good robustness to the symmetric center.

[0063] The setting of the weighted radius directly affects the smoothing degree of the pixels in different regions in the image. Selecting a suitable weighted radius is important for balancing information recovery and detail blurring. In this embodiment, the symmetric center is set to (131, 135), the tangential smoothing filter frequency is set to Mod(k, 2), and the number of projections is 36. The experimental results are shown in Table 7.

[0064] Table 7 Influence of symmetric center on reconstruction effect

[0065] In the experiment, the size of the image is 256x256, and the actual symmetric center of the image is (131, 135). Therefore, when the effective radius is equal to 125, the radial weighting target and completely cover the image region. The data in Table 7 show that when the weighted template just covers the image region, the quality of the reconstructed image is the highest.

[0066] Further, an experiment is designed to verify the performance of the method in this embodiment.

[0067] (1) A sparsity experiment is designed to verify the performance of the method in this embodiment.

[0068] Sparse angle imaging can significantly reduce the irradiation time of the object to be measured, thereby reducing the irradiation dose of the rays. For lithium-ion batteries, reducing the number of projection images as much as possible can effectively reduce the risk of battery activation. However, with the reduction in the number of projection images, the sparse artifacts in the reconstructed tomographic image are more serious, which seriously affects the judgment of the internal information of the battery. In order to test the performance of SART_TI under different sparsity conditions, this embodiment designs a sparsity experiment for simulation phantoms, X-ray lithium batteries and neutron lithium batteries.

[0069] In the experiment, 20, 25, 30, 36 and 45 projection images are used for iteration 50 times. The size of the tomographic image in the experiment is 256x256, and the reconstruction results are compared with SART.

[0070] With the reduction in the number of projection images, the quality of the images reconstructed based on SART and SART_TI methods both decreases obviously, but the SART_TI method can more effectively suppress sparse artifacts, restore the structure and details of the image. As can be seen from the enlarged view of the local region, even if the number of projections is only 20, SART_TI can still better restore the image information, and only the sawtooth artifacts exist in the outer region of the battery. At the same time, it can be seen that the image reconstructed by SART_TI using 20 projection images has a quality comparable to that of SART using 45 projection images in terms of subjective quality.

[0071] To analyze the convergence of the algorithm and quantitatively analyze the quality of the reconstructed image, the iterative convergence curves are plotted based on the MSE, SSIM and PSNR evaluation indexes as shown in Figure 8 Figure 8 It can be seen that (simulated lithium battery piece-no noise), the SART_TI method has converged after 30 iterations, indicating that the SART_TI method has a faster convergence speed. In addition, from the three image quality evaluation indexes, it can be seen that SART_TI can more effectively restore the details of the image and improve the signal-to-noise ratio of the image.

[0072] The SART method cannot effectively restore the details of the image under different sparseness. With the increase of the number of projections, the information near the center of the battery area is gradually restored, but there are still obvious noise and artifacts. For SART_TI, the image reconstructed by 45 projections has a high subjective similarity with the standard image, and the details of the image are effectively restored. With the decrease of the number of projections, the edges and background areas of the reconstructed image begin to appear sparse artifacts, but the structure of the battery still has a high similarity with the standard image.

[0073] From the convergence curves in Figure 9 It can be seen that the MSE and SSIM curves oscillate to a certain extent in the initial stage of iteration, and the three curves begin to converge when the iteration is about 25 rounds. Compared with the SART method, SART_TI can more effectively restore the image information, especially the signal-to-noise ratio of the image.

[0074] From Figure 10 It can be seen that the convergence curve of the neutron image oscillates more obviously during the iteration process compared with the simulation data and the X-ray data, and continues throughout the iteration process. But the SART_TI method gradually converges when the iteration is about 30 rounds, and the convergence curves of the three evaluation indexes show that the method can obtain better image quality.

[0075] (2) Design ablation experiments to verify the influence of different modules on the method in the embodiment.

[0076] The SART_TI method designed in the embodiment mainly involves three improvements: tangential smoothing, radial weighting and sine map diagonal interpolation. In order to evaluate the influence of the three modifications on the performance of the algorithm, ablation experiments are designed in the embodiment. The tangential smoothing module, the radial weighting module and the sine map interpolation module are sequentially added to the SART, and the corresponding images are reconstructed. The influence of different modules on the performance of the algorithm is evaluated by analyzing the quality of the images under different conditions. The experimental objects are simulation phantom, X-ray lithium battery and neutron lithium battery images, and the number of projection images participating in the iterative reconstruction in the experiment is 36.

[0077] ​The image based on SART reconstruction restores the basic structure, but the details are not clear enough, and the noise is relatively high. After introducing the tangential smoothing module, the clarity of the three images has been significantly improved. The ring structure of the simulation phantom image is more distinct, the edges of the X-ray lithium battery are more sharp, and the texture contrast of the neutron lithium battery image is also enhanced. This shows that the tangential smoothing module plays an important role in improving the edge and texture contrast of the image. The radial weighting module further optimizes the image, especially in the representation of details. The contrast and clarity of the simulation phantom are enhanced, and the texture and structural details in the X-ray lithium battery and neutron lithium battery images are more fine, indicating that the radial weighting module has a significant improvement effect on the reconstruction quality of the central region. Finally, after adding the sinogram interpolation module, the three images have the most significant improvement in artifact suppression and detail clarity. The ring structure of the simulation phantom becomes clearer, the internal stripes and edges of the X-ray lithium battery are clearly visible, and the structural contrast and clarity of the neutron lithium battery image are most prominent.

[0078] In order to objectively and quantitatively analyze the influence of different modules on image quality, the quality coefficient histogram is drawn with SSIM, MSE and PSNR as evaluation indexes. As shown in Figure 11 MSE of all types of images decreases with the superposition of different modules, indicating that the reconstruction error is decreasing, reflecting that the difference between the reconstructed image and the original image at the pixel level is becoming smaller. This shows that the addition of modules effectively improves the reconstruction quality in terms of accuracy. Secondly, the value of SSIM is high in different image types. By further adding modules, the SSIM value of the three images is improved, indicating that the structural fidelity of the reconstructed image has been significantly improved. The improvement of PSNR after the superposition of modules is particularly obvious, especially in the simulation and X-ray lithium battery images, which shows that the visual quality has been significantly improved.

[0079] Considering these three indicators, it can be concluded that with the increase of modules, the image reconstruction quality has been improved as a whole. In the simulation image, these improvements may be particularly prominent, because the simulation environment allows accurate evaluation of algorithm performance. In actual applications, such as X-ray and neutron lithium battery image reconstruction, these enhanced techniques have also proven their effectiveness, demonstrating lower reconstruction error and better structural fidelity. The improvement of evaluation indicators shows the effectiveness of the proposed method in the task of sparse angle reconstruction, and the addition of radial weighting and diagonal interpolation modules plays a key role in improving the accuracy, structural fidelity and visual quality of the image.

[0080] The above results show that the method provided by the embodiment can effectively restore the detail information of the image in the sparse iterative reconstruction of the cylindrical lithium battery. The method of the embodiment effectively solves the key problems of the existing sparse angle CT reconstruction technology, and has a broad practical application prospect.

[0081] Embodiment 2 As another aspect of the embodiments of the present disclosure, a computer tomography reconstruction system 100 for sparse-angle projection data of a cylindrical lithium battery is also provided, as shown in the accompanying drawings, comprising: Figure 12 a complete sinogram data generation module 1, which acquires a sequence of projection images of a cylindrical lithium battery as original projection data, performs diagonal interpolation processing on the original projection data, and generates complete sinogram data; an initial iterative reconstruction image generation module 2, which performs Radon inverse transformation on the interpolated sinogram data to generate an initial iterative reconstruction image; a reconstruction image optimization module 3, which takes the initial iterative reconstruction image as an iterative initial value, adopts simultaneous algebraic reconstruction technology as a data fidelity term, and iteratively optimizes the initial iterative reconstruction image to obtain an optimized reconstruction image; a reconstruction image sparse artifact suppression module 4, which designs a tangential smoothing filter based on the rotational symmetry characteristics of the cylindrical lithium battery, performs filter processing on the optimized reconstruction image to suppress sparse artifacts, and obtains a reconstruction image after suppression of sparse artifacts; a reconstruction image differential smoothing processing module 5, which designs a radial weighting template with the rotation center as the reference, dynamically adjusts the smoothing weight according to the distance of the pixel from the image center, and performs differential smoothing processing on the reconstruction image after suppression of sparse artifacts; a final reconstruction image acquisition module 6, which cyclically executes the reconstruction image optimization, sparse artifact suppression, and differential smoothing processing modules until a preset iteration number is reached, and obtains a final reconstruction image.

[0082] Without causing contradictions, the above modules in the system of the embodiments of the present disclosure can realize any of the embodiments of the above methods.

[0083] Based on the description of the above embodiments, the embodiments of the present disclosure can achieve the following technical effects: 1) The present disclosure effectively solves the problems of stripe artifacts and edge blurring caused by sparse-angle projection data through diagonal interpolation processing and the design of a tangential smoothing filter, significantly improving the image quality.

[0084] 2) The present disclosure combines the dynamic smoothing strategy of the radial weighting template, perfectly preserves the microstructure details of the battery electrode while suppressing artifacts, and provides a more reliable image basis for lithium battery internal defect detection.

[0085] 3) The present disclosure significantly improves the convergence speed compared to traditional methods through the synergistic optimization of SART parallel iteration and periodic tangential filtering, greatly reducing the consumption of computing resources.

[0086] ​The embodiments of the present disclosure also provide an electronic device, comprising: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to implement the above-mentioned computer tomography reconstruction method for sparse-angle cylindrical lithium battery. The electronic device can be provided as a terminal, a server or other forms of devices.

[0087] The embodiments of the present disclosure also provide a computer-readable storage medium having stored thereon computer program instructions, which, when executed by a processor, implement the above-mentioned computer tomography reconstruction method for sparse-angle cylindrical lithium battery. The computer-readable storage medium can be a non-volatile computer-readable storage medium.

[0088] Those skilled in the art can understand that, in the above-mentioned three-dimensional reconstruction method and system of strong ambient light filtering and depolarization error correction, the writing order of each step does not mean a strict execution order and does not constitute any limitation on the implementation process, and the specific execution order of each step should be determined according to its function and possible internal logic.

[0089] The flow diagrams and the block diagrams in the drawings are presented to illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flow diagrams and the block diagrams can represent a module, a segment, or a portion of code, which comprises one or more executable instructions for implementing the specified logical functions. In some alternative implementations, the functions noted in the blocks can occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and / or flow diagrams, and combinations of blocks in the block diagrams and / or flow diagrams, can be implemented by special purpose hardware-based systems that perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.

[0090] The embodiments of the present disclosure have been described above, and the above description is exemplary and is not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes are obvious to those skilled in the art without departing from the scope and spirit of the described embodiments. The selection of the terms used herein is intended to best explain the principles, practical applications or technical improvements of the technology in the market, or to enable other ordinary skilled persons in the art to understand the embodiments disclosed herein.

Claims

1. A method for computer tomography reconstruction of rarefaction angles of cylindrical lithium batteries, characterized in that: The steps include: S10, acquiring a cylindrical lithium battery projection image sequence as original projection data, performing diagonal interpolation processing on the original projection data, and generating complete sinusoidal graph data; S20, performing an inverse Radon transform on the interpolated sinusoidal data to generate an initial iteratively reconstructed image; S30, using the initial iterative reconstructed image as an iteration initial value, using a simultaneous algebraic reconstruction technique as a data fidelity item, and iteratively optimizing the initial iterative reconstructed image to obtain an optimized reconstructed image; S40, designing a tangential smoothing filter based on the rotational symmetry characteristics of the cylindrical lithium battery, and performing filtering processing on the optimized reconstructed image to suppress sparse artifacts, thereby obtaining a reconstructed image with suppressed sparse artifacts; S50, designing a radial weighted template based on the rotation center, dynamically adjusting the smoothing weight according to the distance between the pixel and the image center, and performing differential smoothing on the reconstructed image after suppressing the sparse artifacts; S60 , looping through steps S30 - S50 until a preset number of iterations is reached to obtain a final reconstructed image.

2. The method according to claim 1, characterized in that Perform diagonal interpolation on the original projection data to generate complete sinogram data, including: Identifying missing projection angles in the original projection data, and analyzing grayscale similarity of pixels along a diagonal line in a sinusoidal graph of the original projection data; Interpolate along the diagonal direction of the sinusoidal graph, giving priority to pixels with similar diagonals in adjacent projection angles; The sinogram is generated using linear interpolation.

3. The method according to claim 2, characterized in that The original projection data is diagonally interpolated to generate complete sinusoidal data. This also includes: filtering back-projection reconstruction of the sinusoidal data generated by linear interpolation. The interpolation effect is evaluated by peak signal-to-noise ratio, structural similarity, and mean square error, which is expressed as: ， ， ， Where PSNR represents the peak signal-to-noise ratio, SSIM represents structural similarity, MSE represents mean square error, n Indicates the grayscale level of the image, f 1 and f 2 represent the standard image and the image to be evaluated, 、 、 、 Represents images respectively f 1 and f 2 mean and variance, Representing an image f 1 and f The covariance of 2, and is a constant, M and N Represents the width and height of the image respectively, Represents the coordinates of pixels in the image.

4. The method according to claim 1, wherein The simultaneous algebraic reconstruction technique is used as the data fidelity item to iteratively optimize the initial iteratively reconstructed image, which can be expressed as: ， Where, For the k +1 reconstructed image of the iteration, For the k The reconstructed image of the iteration, C is a diagonal matrix whose elements are determined by the system matrix A The column vector norm of A is the weighted system matrix, used to adjust the update step size, To control the relaxation factor of each update amplitude, b The projection data are obtained from computed tomography.

5. The method according to claim 1, wherein A tangential smoothing filter is designed, and the mathematical model of the tangential smoothing filter is: ， Where, For coordinates pixels, w 1 and w 2 is the interpolation weight. The tangential neighborhood of pixel x is often not strictly located at the pixel point. is the row coordinate of the tangential neighboring pixel.

6. The method according to claim 1, characterized in that Design a radial weighted template based on the rotation center, which is expressed as: ， Where, W is a radially weighted sparse template, i and j Represent the horizontal and vertical coordinates of the pixel respectively, and Respectively represent the horizontal and vertical coordinates of the image symmetry center, R is the effective radius of the weighted template.

7. The method according to claim 1, characterized in that The smoothing weight is dynamically adjusted according to the distance between the pixel and the image center, and differential smoothing is performed on the reconstructed image after suppressing the sparse artifacts, which is expressed as: ， Where, represents the image calculated by the simultaneous algebraic reconstruction technique, represents the image after tangential smoothing, f Represents the radially weighted image.

8. A computed tomography reconstruction system for rarefaction angles of cylindrical lithium batteries, characterized in that: include: A complete sinusoidal graph data generation module obtains a cylindrical lithium battery projection image sequence as raw projection data, performs diagonal interpolation processing on the raw projection data, and generates complete sinusoidal graph data; An initial iterative reconstruction image generation module performs an inverse Radon transform on the interpolated sinusoidal data to generate an initial iterative reconstruction image; a reconstructed image optimization module, which uses the initial iterative reconstructed image as an iteration initial value, adopts a simultaneous algebraic reconstruction technique as a data fidelity item, and iteratively optimizes the initial iterative reconstructed image to obtain an optimized reconstructed image; A reconstructed image sparse artifact suppression module is configured to design a tangential smoothing filter based on the rotational symmetry of a cylindrical lithium battery, and to filter the optimized reconstructed image to suppress sparse artifacts, thereby obtaining a reconstructed image after suppressing sparse artifacts; A differential smoothing processing module for reconstructed images is designed, which designs a radial weighted template based on the rotation center, dynamically adjusts the smoothing weight according to the distance between the pixel and the image center, and performs differential smoothing processing on the reconstructed image after suppressing the sparse artifacts; The final reconstructed image acquisition module cyclically executes the reconstructed image optimization, sparse artifact suppression and differential smoothing processing modules until a preset number of iterations is reached to obtain the final reconstructed image.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the computer program, it implements the computed tomography reconstruction method of the sparse angle of the cylindrical lithium battery according to any one of claims 1 to 7.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the computer tomography reconstruction method of the sparse angle of the cylindrical lithium battery according to any one of claims 1 to 7 is implemented.

Images